Module 1

HYDROLOGIC CYCLE

1.1 Introduction
1.2 Components of Hydrologic Cycle
1.3 Scales for study of hydrologic cycle
1.4 Mathematical Representation of the Hydrologic Cycle
1.5 Global water balance
1.6 Influence of Human Activities and Land use Changes on Hydrologic Cycle
1.7 Impact of climate changes on the hydrologic cycle
1.8 Closure

Keywords: Hydrologic cycle, precipitation, water balance, components, water balance, human
impact, climate change,

1.1 Introduction
Water can occur in three physical phases: solid, liquid, and gas and is found in nature in all these
phases in large quantities. Depending upon the environment of the place of occurrence, water
can quickly change its phase.

A number of cycles are operating in nature, such as the carbon cycle, the nitrogen cycle,
and several biogeochemical cycles. The Hydrologic Cycle, also known as the water cycle, is one
such cycle which forms the fundamental concept in hydrology. Hydrologic cycle was defined by
the National Research Council (NRC, 1982) the as “the pathway of water as it moves in its
various phases to the atmosphere, to the earth, over and through the land, to the ocean and back
to the atmosphere”. This cycle has no beginning or end and water is present in all the three states
(solid, liquid, and gas). A pictorial view of the hydrological cycle is given in Fig. 1.1. The
science of hydrology primarily deals with the land portion of the hydrologic cycle; interactions
with the oceans and atmosphere are also studied. NRC (1991) called the hydrologic cycle as the
integrating process for the fluxes of water, energy, and the chemical elements.

The hydrologic cycle can be visualized as a series of storages and a set of activities that
move water among these storages. Among these, oceans are the largest reservoirs, holding about
97% of the earth’s water. Of the remaining 3% freshwater, about 78% is stored in ice in
Antarctica and Greenland. About 21% of freshwater on the earth is groundwater, stored in
sediments and rocks below the surface of the earth. Rivers, streams, and lakes together contain
less than 1% of the freshwater on the earth and less than 0.1% of all the water on the earth.
 Hydrologic
H cycle
c consid
ders the mo otion, loss, and rechargge of the eearth's wateers. It
connects the atmosphere and tw wo storages ofo the earth system: the oceans, andd the land spphere
(lithosph
here and ped dosphere). The
T water ev vaporated frrom the eartth and the ooceans enterrs the
atmospheere. Water leeaves the atmosphere th hrough preciipitation. Thhe oceans receive water from
the atmoosphere by means
m of preecipitation and
a from thhe land throuugh rivers aand ground w water
flow. Water
W goes out
o of ocean ns only thro ough evapooration. The water leavves land thrrough
evapotrannspiration, streamflow,
s and ground water flow. Evaporationn and precippitation proccesses
take placce in the veertical plane while streaamflow and ground watter flow occcur mostly iin the
horizontaal plane.

Fig. 1.1 Pictorial

P vieew of the hydrologiccal cycle. ((Source: W
Wikipedia, w
www.
htttp://en.wikipedia.org/wiki/Water_cy
ycle).

The
T exchangee of water amonga the oceans, land,, and the atm mosphere waas termed ass ‘the
turnover’’ by Shikhlo omanov (199 99). This turnnover affectts the global patterns of the movemeent of
ocean waaters and gases in the attmosphere, thereby
t greaatly influenccing climate. Since wateer is a
very goood solvent, chemistry is an integral part
p of the hhydrologic ccycle. Usuallly, rain and snow
are consiidered as thee purest form
m of water allthough thesse may also bbe mixed wiith pollutants that
are preseent in the atm
mosphere. Duuring the jouurney on eartth, many cheemical comppounds are m mixed
with watter and conssequently thee water quallity undergooes a changee. One can aalso visualizze the
hydrologgic cycle as a perpetual distillation
d and
a pumpingg system in which the gglaciers and snow
packs aree replenishedd and rivers get water off good qualitty.
 We need to study the hydrologic cycle since water is essential for survival of life and is
an important input in many economic activities. From the use point of view, the land phase of the
hydrologic cycle is the most important.

In view of the complexities and extensive coverage, the study of the complete hydrologic
cycle is truly interdisciplinary. For instance, the atmospheric part is studied by meteorologists,
the pedospheric part by soil scientists, the lithosphere part by geologists, and the part pertaining
to oceans falls in the domain of oceanographers. A host of other professionals study hydrologic
cycle: agricultural engineers, energy managers, ecologists and environmentalists, public health
officers, industrialists, chemists, and inland navigation managers.

1.2 Components of Hydrologic Cycle

The hydrologic cycle can be subdivided into three major systems: The oceans being the major
reservoir and source of water, the atmosphere functioning as the carrier and deliverer of water
and the land as the user of water. The amount of water available at a particular place changes
with time because of changes in the supply and delivery. On a global basis, the water movement
is a closed system but on a local basis it is an open system.

The major components of the hydrologic cycle are precipitation (rainfall, snowfall, hale,
sleet, fog, dew, drizzle, etc.), interception, depression storage, evaporation, transpiration,
infiltration, percolation, moisture storage in the unsaturated zone, and runoff (surface runoff,
interflow, and baseflow).

Evaporation of water takes place from the oceans and the land surface mainly due to solar
energy. The moisture moves in the atmosphere in the form of water vapour which precipitates
on land surface or oceans in the form of rain, snow, hail, sleet, etc. A part of this precipitation is
intercepted by vegetation or buildings. Of the amount reaching the land surface, a part infiltrates
into the soil and the remaining water runs off the land surface to join streams. These streams
finally discharge into the ocean. Some of the infiltrated water percolates deep to join
groundwater and some comes back to the streams or appears on the surface as springs.

This immense movement of water is mainly driven by solar energy: the excess of
incoming radiation over the outgoing radiation. Therefore, sun is the prime mover of the
hydrologic cycle. The energy for evaporation of water from streams, lakes, ponds and oceans and
other open water bodies comes from sun. A substantial quantity of moisture is added to the
atmosphere by transpiration of water from vegetation. Living beings also supply water vapor to
the atmosphere through perspiration. Gravity has an important role in the movement of water on
the earth’s surface and anthroprogenic activities also have an increasingly important influence on
the water movement.

An interesting feature of the hydrologic cycle is that at some point in each phase, there
usually occur: (a) transportation of water, (b) temporary storage, and (c) change of state. For
example, in the atmospheric phase, there occurs vapor flow, vapor storage in the atmosphere and
condensation or formation of precipitation created by a change from vapor to either the liquid or
solid state. Moreover, in the atmosphere, water is present in the vapor form while it is mostly
(saline) liquid in the oceans.

1.3 Scales for study of hydrologic cycle

From the point of view of hydrologic studies, two scales are readily distinct. These are the global
scale and the catchment scale.

Global scale
From a global perspective, the hydrologic cycle can be considered to be comprised of three
major systems; the oceans, the atmosphere, and the landsphere. Precipitation, runoff and
evaporation are the principal processes that transmit water from one system to the other. This
illustration depicts a global geophysical view of the hydrologic cycle and shows the interactions
between the earth (lithosphere), the oceans (hydrosphere), and the atmosphere. The study at the
global scale is necessary to understand the global fluxes and global circulation patterns. The
results of these studies form important inputs to water resources planning for a national, regional
water resources assessment, weather forecasting, and study of climate changes. These results
may also form the boundary conditions of small-scale models/applications.

Catchment Scale
While studying the hydrologic cycle on a catchment scale, the spatial coverage can range from a
few square km to thousands of square km. The time scale could be a storm lasting for a few
hours to a study spanning many years. When the water movement of the earth system is
considered, three systems can be recognized: the land (surface) system, the subsurface system,
and the aquifer (or geologic) system. When the attention is focused on the hydrologic cycle of
the land system, the dominant processes are precipitation, evapotranspiration, infiltration, and
surface runoff. The land system itself comprises of three subsystems: vegetation subsystem,
structural subsystem and soil subsystem. These subsystems subtract water from precipitation
through interception, depression and detention storage. This water is either lost to the
atmospheric system or enters subsurface system. The exchange of water among these subsystems
takes place through the processes of infiltration, exfiltration, percolation, and capillary rise.

Fig. 1.2 shows the schematic of the hydrologic cycle at global scale, in the earth system, and
micro-scale view of the cycle in the land system. Fig. 1.3 gives a schematic presentation of the
hydrologic cycle of the earth system. Detailed schematic of the hydrologic cycle in the land
system is shown in Fig. 1.4.

Precipitation

Atmospheric Precip. Earth Riverflow Ocean

system system GW flow system
ET

Evaporation

Fig. 1.2 A global schematic of the hydrologic cycle.

Precipitation
ET
Surface
Land system runoff

Exfiltration Infiltration

Subsurface system Subsurface

runoff Stream
Percolation Upward flow
(recharge) moisture movement
Ground
Aquifer system water runoff

Fig. 1.3 A schematic of the hydrologic cycle of the earth system.

Time scales in hydrologic cycle

The time required for the movement of water through various components of the hydrologic
cycle varies considerably. The velocity of streamflow is much higher compared to the velocity
of ground water. The time-step size for an analysis depends upon the purpose of study, the
availability of data, and how detailed the study is. The estimated periods of renewal of water
resources in water bodies on the earth is given in Table 1.1. The time step should be sufficiently
small so that the variations in the processes can be captured in sufficient detail but at the same
time, it should not put undue burden on data collection and computational efforts.

Precip. ET Precip. Evaporation Precip.

Structural Soil
Vegetation system
system Depression
system Inter- Surface
ception storage
runoff
Throughflow
Infiltration
Moisture supply

Fig. 1.4 A detailed schematic of the hydrologic cycle in the land system.

Table 1.1 Periods of water resources renewal on the Earth

Water of hydrosphere Period of renewal

World Ocean 2500 years
Ground water 1400 years
Polar ice 9700 years
Mountain glaciers 1600 years
Ground ice of the permafrost zone 10000 years
Lakes 17 years
Bogs 5 years
Soil moisture 1 year
Channel network 16 days
Atmospheric moisture 8 days
Biological water Several hours
Source: Shiklomanov (1999).

The range of spatial and temporal dimensions of many processes related to the hydrologic cycle
is shown in Fig. 1.5.
1.4 Mathematical Representation of the Hydrologic Cycle
The quantities of water going through the various components of the hydrologic cycle can be
evaluated by the so-called hydrologic equation, which is a simple spatially-lumped continuity or
water budget equation:

I - Q = ΔS (1.1)

where I = inflow of water to a given area during any given time period, Q = outflow of water
from the area during the selected time period, and ΔS = change in storage of water in the given
area during the time period. If I and Q vary continuously with time, then equation (1.1) can be
written as

d S(t)/dt = I(t) – Q(t) (1.2)

GLOBAL
GLOBAL GLOBAL
       
WEATHER   C02        
SYSTEMS VARIATIONS
10000 km
DEVELOPMENT
             
OF MAJOR    
RIVER BASINS
SOIL
1000 km FORMATION
        RUNOFF          
CYCLE
SPACE

100 km MESOSCALE DRAINAGE

WEATHER NETWORK
MESOSCALE
    SYSTEMS 
(FLOODS)
 SOIL MOISTURE
  SOIL   
FORMATION    
VARIATION EROSION

SHALLOW
10 km GROUND
WATER
        NUTRIENT      
CIRCULATION
   
CYCLES
THUNDER
1 km
STORM
        TIME
         

LOCAL Log Sec

2 4 6 8 10 12 14 16

SEC MIN DAY YEAR CENTAURY ONE MILLION ONE BILLION

YEARS YEARS

Fig. 1.5 Illustrative range of process scales [source: NRC, 1991].

By integrating, this equation can also be written as

 dS(t) =  [I(t) – Q(t)] dt

t t
or S (t )  S (0)   I (t )dt   Q(t )dt
0 0

= VI(t) – V0(t) (3)

where S(0) is the initial storage at time t=0, S(t) is the storage at time t, V0(t) and VI(t) are
volumes of outflow and inflow at time t. Each of the terms of this lumped equation is the result
of a number of other terms. These can be sub-divided and even eliminated from the equation
depending upon the temporal and spatial scale of the study. The continuity equation is one of the
governing equations of almost all hydrologic problems. For a watershed, equation (1.1) may be
written as

P + QSI + QGI - E - QSO - QGO - S -  = 0 (4)

where P is precipitation, QSI is surface inflow, QGI is ground water inflow, E is evaporation from
the watershed, QSO is surface water outflow, QGO is ground water outflow, S is change in the
storage of water in the watershed, and  is a discrepancy term. For large watersheds, QGI and QGO
are usually negligible. The discrepancy term is included in equation (1.4) because the sum of all
other terms may not be zero due to measurement errors and/or simplifying assumptions.
However, a small value of discrepancy term does not necessarily means that all other terms have
been correctly measured/estimated.

Depending on the specific problem, the terms of equation (1.1) may be further
subdivided. For example, when applying the hydrologic equation for short time intervals, the
change in the total water storage (S) may be subdivided into several parts: changes of moisture
storage in the soil (M), in aquifers (G), in lakes and reservoirs (L), in river channels (SC),
in glaciers (SG), and in snow cover (SS). Thus, S can be expressed as:

S = M + G + L + SC + SG + SS (1.5)

The hydrologic equation may be applied for any time interval; the computation of the mean
annual water balance for a basin being the simplest, since it is possible to disregard changes in
water storages in the basin (S), which are difficult to measure and compute. In general, the
shorter the time interval, the more stringent are the requirements for measurement or
computation of the components and the more subdivided are the terms of equation (1.1). This
results in a complex equation which is difficult to close with acceptable errors.

The hydrologic equation may be applied for areas of any size, but the complexity of
computation greatly depends on the extent of the area under study. The smaller is the area, the
more complicated is its water balance because it is difficult to estimate components of the
equation. Finally, the components of the hydrologic equation may be expressed in terms of the
mean depth of water (mm), or as a volume of water (m3), or in the form of flow rates (m3/s or
mm/s).

1.5 Global Water Balance

According to estimates (Seckler et al., 1998), the annual average depth of precipitation on the
land surface is about 108*103 km3. Out of this, about 61*103 km3 is returned to the atmosphere
as evapotranspiration and the runoff from land to oceans is 47*103 km3. As far as the water
balance of oceans is concerned, the depth of precipitation over them is about 410*103 km3, 47
*103 km3 of water is received as runoff from the land, and 457*103 km3 is lost as evaporation. If
we consider the water balance of atmosphere, 457*103 km3 of water is received as evaporation
from oceans and 61*103 km3 from land. The precipitation over oceans is 410*103 km3 and it is
108*103 km3 over land. Table 1.2 gives quantitative values of water present in various forms on
the earth.
1.6 Influence of Human Activities and Land use Changes on Hydrologic Cycle
Watersheds are subjected to many types of changes, major or minor, for various reasons. Some
of these are natural changes and some are due to human activities. Watershed changes affect
virtually all elements of the hydrologic cycle. The quality of water is significantly deteriorating
at many places due to industrial and agricultural activities. There has been a growing need to
quantify the impact of major human-induced changes on the hydrologic cycle in order to
anticipate and minimize the potential environmental detriment and to satisfy water resources
requirements of the society. Even if the water of adequate quantity were present at a place, its
use may be limited because of poor quality. The modern viewpoints of role of the humans in the
hydrologic cycle are shown in Fig. 1.6.

Watershed changes can be categorised as point changes and non-point changes. Structural
changes, such as dam construction, channel improvement, and detention storage are examples of
point changes and affect watershed response in terms of evaporation, seepage, residence time,
etc. Forestry, agriculture, mining, and urbanization are non-point land use changes. A qualitative
discussion of the hydrologic consequences due to watershed changes is given in the following.

(a) Effects of Agricultural Changes

These changes imply that a land area that was earlier forested or a barren land is now being
cultivated. As a result, the vegetal cover changes, soil crusting and infiltration characteristics
change, and artificial bunds may be placed. The effect of these changes on the hydrologic regime
is pronounced and may be multiplicative. The water may be withdrawn from the ground water
zone or canal irrigation may be applied on the land leading to noticeable changes in the water
table behavior. The impacts are also noticed in evapotranspiration, overland flow, channel flow,
and infiltration. Fertilizers, pesticides, and insecticides that are applied to crops affect the water
quality of runoff from agriculture areas.

Table 1.2 Water reserves in various phases of the hydrologic cycle

Distributio Volume (103 Layer (m) Percentage of global
n area (103 km3) reserves
km2) Of total Of fresh-
water water
World ocean 361,300 1,338,000 3,700 96.5 -
Groundwater 134,800 23,400 174 1.7 -
 Freshwater 10,530 78 0.76 30.1
Soil moisture 16.5 0.2 0.001 0.05
Glaciers and permanent 16,227 24,064 1,463 1.74 68.7
snow cover
Antarctic 13,980 21,600 1,546 1.56 61.7
Greenland 1,802 2,340 1,298 0.17 6.68
Arctic islands 226 83.5 369 0.006 0.24
Mountainous regions 224 40.6 181 0.003 0.12
Ground ice/permafrost 21,000 300 14 0.022 0.86
Water reserves in lakes 2,058.7 176.4 85.7 0.013 -
Fresh 1,236.4 91 73.6 0.007 0.26
Saline 822.3 85.4 103.8 0.006 -
Swamp water 2,682.6 1.47 4.28 0.0008 0.03
River flows 148,800 2.12 0.014 0.0002 0.006
Biological water 510,000 1.12 0.002 0.0001 0.003
Atmospheric water 510,000 12.9 0.025 0.001 0.04
Total water reserves 510,000 1,385,984 2.718 100 -
Total freshwater reserves 148,800 35,029 235 2.53 100
Source: Unesco (1978).

(b) Effects of Urbanization

A land area that was being used for purposes, such as forestry, agriculture, might be transformed
into an urban area where houses, roads, parks, parking lots, sewers, etc. are constructed. A large
increase in the impervious surface takes place which considerably reduces infiltration and the
removal of storm water is accelerated. Urban development usually increases the volume and
peak of direct runoff for a given rainfall event. The time of travel of water is reduced, resulting
in a lower lag time and a lower time of concentration.

Natural system
Oceans

Atmosphere Land Surface

Humans Anthropogenic Processes

 Fig. 1.6 A viewpoint showing role of humans in hydrologic cycle.

(c) Effects of Forest Activities

These activities may be directed towards planting trees as well as cutting them. When a forested
area is deforested and forest litter removed, the interception of precipitation is virtually
eliminated. Litter removal changes infiltration capacity of soil and has a pronounced effect on
raindrop impact and the resulting soil erosion. With the loss of vegetation, evapotranspiration is
generally decreased. These changes amount to increased production of direct runoff, reduced
surface roughness, and decreased recharge to ground water. The hydrograph of direct runoff
rises more quickly because of the reduced time to peak. However, when additional trees are
planted in an area, the effect is reverse though the impact takes place gradually as the trees grow.

(d) Effects of Structural Changes

Typical structural changes include a dam, a weir, channel improvement works, etc. A dam-
reservoir is constructed for many purposes. Regardless of its intended function, it does affect the
hydrology of the stream on which it is built. In general, the peak of outflow from a reservoir is
less and the flow may be more even than the pre-project condition. The volume of flow
downstream may be considerably less in the after-project scenario if the reservoir water is
diverted elsewhere.

Decreasing channel roughness increases flow velocity and peak discharge for the same
channel size. The removal of vegetation, lining of the channel, and proper maintenance can
greatly reduce roughness. The other alterations, such as straightening the channel, maintenance
of bands, or increasing slope, significantly affect travel time and flow velocity. Depending upon
the bed material, infiltration through the bed and banks also modifies flow characteristics.

1.7 Impact of climate changes on the hydrologic cycle

An analysis of measured data series as well as paleoclimatological records suggests that the
climate of the earth may be undergoing a significant change. The climatic change may have
important impacts on the hydrologic cycle (the converse is also true). The change may be taking
place due to natural and human causes. The major changes that could affect climate are changes
in vegetation (changes in evapotranspiration, soil moisture, albedo, and radiation balance),
increased use of water for day-to-day needs, burning of fossil fuels, and industrial activities.
Large-scale water transfers from one basin to another may also cause climate change in the long
run. The change manifests itself in changed patterns of spatial and temporal variability in the
components of the hydrologic cycle.

The increased emission of green-house gases is believed to be the cause of gradual

increase in earth’s temperature. The increase in the temperature of the atmosphere would lead to
higher evapotranspiration, changes in precipitation pattern, timing, and distribution, melting of
polar ice caps and recession of glaciers. Higher melting of polar ice and glaciers will cause rise
of sea water level and inundation of islands of low elevations as well as cities adjacent to seas.
Another possibility is that an increase in temperature may mean more precipitation, some of
which will be in the form of snow at the poles, leading to an additional accumulation of ice. The
coupled atmosphere-ocean general circulation models are widely used to study the response of
climate to various changes. Notwithstanding a large number of studies, it is not known with
sufficient degree of certainty and accuracy as to what is going to happen and where?

1.8 Closure
Water is central to the environment and is considered to be the driver of nature. All studies
dealing with water are concerned with some part or the other of the hydrologic cycle. The
occurrence and availability of water is closely linked with human development. As a result of
many human activities, the features/characteristics of many components of the hydrologic cycle
are undergoing changes which are likely to result in long-term changes in the climate. It is
necessary to predict the magnitude and extent of these changes so that the required ameliorative
measures can be initiated well in time.

References
NRC (1982). Scientific Basis of Water Resource Management. National Research Council,
National Academy Press, Washington D.C.
NRC (1991). Opportunities in the Hydrologic Sciences. Committee on ‘Opportunities in the
Hydrologic Sciences’ of National Research Council, National Academy Press, Washington
D.C.
Seckler, David, Upali Amarasinghe, Molden David, Radhika de Silva, and Randolph Barker.
(1998). World water demand and supply, 1990 to 2025: Scenarios and issues. Research
Report 19. International Water Management Institute. Colombo, Sri Lanka.
Shiklomanov, I. A. (1999). World water resources: modern assessment and outlook for the 21st
century. (Prepared in the framework of IHP, Unesco). State Hydrology Institute, St.
Petersburg.
Singh,V.P. (1992). Elementary Hydrology. Prentice Hall, Engelwood Cliffs, New Jersey.
Unesco (1978). World water balance and water resources of the Earth. United Nations
Educational, Scientific, and Cultural Organization, Paris.
WMO (2008). Guide to Hydrological Practices, Volume I: Hydrology – From Measurement to
Hydrological Information. WMO-No. 168, World Meteorological Organization, Geneva.
 Module 2
Measurement and Processing of Hydrologic Data
2.1 Introduction
2.1.1 Methods of Collection of Hydrologic Data
2.2 Classification of Hydrologic Data
2.2.1 Time-Oriented Data
2.2.2 Space-Oriented Data
2.2.3 Relation-Oriented Data
2.3 Design of Hydrometeorological Data Networks
2.3.1 Classification of Observation Networks
2.4 Precipitation Networks
2.5 Stream Gauging Networks
2.5.1 Network Design Process
2.5.2 Criteria for Location of Stations
2.5.3 Evaluation and Adequacy of Networks
2.5.4 Site Selection Surveys
2.5.5 General site selection guidelines
2.5.6 Criteria for Water Level Gauging Sites
2.5.7 Criteria for Streamflow Measurement Sites
2.5.8 Criteria for Natural Control Sites
2.5.9 Criteria for Artificial Control Sites
2.5.10 Bureau of Indian Standards (BIS) criteria for selection of river gauging sites
2.5.11 World Meteorological Organisation (WMO) criteria for selection of site
2.5.12 International Standards for Hydrometry
2.6 Errors in Hydrological Observations
2.6.1 Sources of Errors
2.6.2 Secondary Errors of Measurement
2.7 Validation of Hydrologic Data
2.7.1 Levels of Validation
2.7.2 Primary Validation
2.7.3 Secondary Validation
2.7.4 Hydrological Validation
2.7.5 Validation of Climatic Data
2.7.6 Single series tests of homogeneity
2.7.7 Multiple stations validation
2.A Definitions of terms related to measurement errors
2.11 References

Keywords: Hydrologic Data, Time-Oriented, Space-Oriented, Relation-Oriented, Networks,

Precipitation, Stream Gauging, Design, Site Selection, Measurement Errors, Validation

2.1 Introduction
Data are the foundations on which any analysis rests. The practice of hydrological measurements
is very old. Kautilya initiated systematic precipitation measurements in India in the fourth
century BC. Streamflow was probably first monitored by Hero of Alexandria in the first century
AD. Equipment and techniques of hydrologic data collection have evolved with growth in
technology and water sciences.

For a water resources study, one needs data of a number of variables in the vertical as
well as horizontal planes. The data needed for water resources development come from a vast
swath of disciplines: hydro-meteorologic, geomorphologic, agricultural, pedologic, geologic,
hydrologic, social, economic, ecological and environmental sciences, etc. Hydrometeorologic
data include rainfall, snowfall, temperature, radiation, humidity, vapor pressure, sunshine hours,
wind velocity, and pan evaporation. Agricultural data include crop cover, irrigation application,
and fertilizer application. Pedologic data include soil type and texture; soil particle size; porosity;
moisture content; steady-state infiltration, and saturated hydraulic conductivity. Geologic data
include stratigraphy, lithology, and structural controls. Frequently, data on the type, depth and
areal extent of aquifers are needed. Ecological and environmental data includes water quality
variables, aquatic plants and animals and their habitats. Each data set is examined with respect to
homogeneity, completeness, and accuracy.

Geomorphologic data include topographic maps showing elevation contours, river

networks, drainage areas, slopes and slope lengths, and watershed area. Hydrologic data include
flow depth, discharge, base flow, stream-aquifer interaction, depth to water table, and
drawdowns. Fig. 2.1 shows the activities of a hydrological service. The term hydrological data
processing is a widely used but loosely defined and includes a range of activities varying from
simple analysis to complete modeling. Before this, of course, the data are observed and this step
is known as origination and collection. Hydrological data processing is a multi-step process that
begins with a preliminary checking of raw data in the field and successively higher levels of
validation before the data are accepted as fully validated. Passage of data from field to storage is
not a one-way process and contains feedbacks. Further, processing and validation of hydrological
data is not a purely statistical exercise –an understanding of field practices, the principles of
observation, and the physics of the variable being measured are required. The activities in data
processing life cycle are shown in Fig. 2.1.

Data processing also includes aggregation of data observed at a given time interval (e.g.
hourly) to a different interval (e.g., daily and daily to monthly) or disaggregation, i.e., conversion
from a long to short (say daily to hourly) time step is also carried out.
Typical stages in hydrological data processing are:
 Scrutiny of raw data;
 data entry to computer, validation, and correction; and
 data archival and dissemination.

2.1.1 Methods of Collection of Hydrologic Data

Hydrological observations are the scientific ways for collection of water related data at a specific
location. There are many ways in which the hydrologic data can be collected. The major
techniques are described below.

Direct Measurement
This is the most common way to measure hydrometeorological variables, such as precipitation
and streamflow. A gauging site is established and is equipped with the devices that can measure
the variable(s) of interest. In case of manual observations, an observer visits the site, measures
the values of the concerned variables, and records or transmits them to the controlling office for
processing and storage. On the other hand, at an automated hydrologic or weather station the
seasons can measure a number of hydrometeorological variables and store/transmit the data to
the controlling office without any human intervention. The equipment may be programmed to
transmit the data at selected time interval or it can be interrogated as per the needs to get the data.
With improvement in communication technology, it is possible to get the desired data from the
stations widely spread over an area at a central place in real-time.

Remote Sensing
In this technique, the data about an object are obtained without coming in physical contact with
the object. This technique is now very commonly used to provide spatial data of terrain features.
Similarly, weather radars are being increasingly used for measurement of precipitation.

OBSERVATION INPUT
STORE

FEEDBACK RETRIEVE

DATA
PROCESSING
OPERATIONS
UTILISE

Observation and input, processing and storage, Retrieval and use, Feedback,

Figure 2.1: An illustration of data processing life cycle activities

Chemical Tracers
In this approach, some chemicals, known as tracers, are added to the process whose data are to
be obtained. Tracers can also be used to determine the flow path of water or a pollutant. The
nuclear or isotope techniques are being employed to trace the movement of water molecules in
different parts of the hydrological cycle. Nuclear techniques are helpful to assess the rate of
sediment deposition in a water body, identify the rainfall recharge and recharge areas of aquifers,
study of seawater intrusion in coastal regions, measure seepage and leakage from surface water
bodies, analyse surface water and ground water interaction, etc.

2.2 Classification of Hydrologic Data

Hydrologic data can be classified in several ways. Most commonly, data are classified in three
categories: time-oriented data, space-oriented data, and relation-oriented data.

Hydrologic data can also be classified as time varying or time non-varying data. The time
non-varying or static data includes most space-oriented data which do not change (or very-very
slowly change) with time, for example catchment topographic map, soil map, etc. Some features,
such as river network and land use in a catchment, might gradually change with time and can be
considered as semi-static. A brief description of each type of data is presented next.

Design of data collection network

Data acquisition
Data collection and transmission

Validation, processing and storage

Data processing and

Analysis of data
analysis

Data for design and operation

Decision making
Public information

Fig. 2.2 Activities of a hydrological service [adapted from WMO (1994)].

2.2.1 Time-Oriented Data
Values of most hydrometeorological variables change with time and such variables are known as
time-oriented data. The time-series data include all the measurements which have an observation
time associated with them and most water resources data have this property. The variable could
be an instantaneous value (e.g. river water level); an accumulated value (e.g., daily rainfall); or
an averaged value (e.g., mean daily discharge). The distinction between instantaneous and
accumulative values is important when the data are processed. These data can be further
classified as meteorological data, hydrological data, and water quality data.

Depending on the frequency of observations, the time-series data can also be classified as:
 Equidistant time-series data which are the measurements made at regular intervals
(hourly, daily); the reported values may be instantaneous, accumulated or averaged.
 Cyclic time-series data are the data measured at irregular intervals of time but the time
sequence is repeated regularly. For example, the observation of at many places river stage
is measured daily at 08:30 and 17:30 hrs.
 Values of non-equidistant data series are observed when some specified event takes
place. For example, in a tipping bucket rain gauge, the bucket tips after a certain depth of
rain has fallen and the value is recorded.

2.2.2 Space-Oriented Data

Space-oriented data comprise of the information related to physical and morphological
characteristics of catchments, rivers (cross-sections, profile, bed characteristics, networks), soil
maps, lakes and reservoirs data (elevation-area variation), etc. Traditionally, such data are stored
in the form of paper maps and manually analyzed. The modern trend is to use a Geographical
Information System (GIS) to input, store and analyze such data. Different types of information,
such as topographical and land use of an area, are stored in a GIS in different layers of a map
which can be overlaid and analyzed.

2.2.3 Relation-Oriented Data

Such data comprise of information about mathematical relationships established between two or
more variables. A mathematical relationship between two or more variables is established for
many purposes, such as data validation, filling-in missing data, etc. The variables themselves
may form a time-series but their relationship is of interest here. The relationship may be
expressed in mathematical, tabular, or graphical form. The stage-discharge rating, spillway rating
table and the calibration ratings of various instruments are typical example of relation-oriented
data. More than one equation may be required to characterize the relationship which may change
with time.
2.3 Design of Hydrometeorological Data Networks
Data on temporal and spatial characteristics of water resources of a region are obtained by a
network of observational stations. Setting up a station requires investment for infrastructure
equipment, logistics, and for operation and maintenance. Scientific planning is necessary for
network design so that the desired results could be achieved with minimum cost.

Since the requirement of water resources data depends on their end use, it is difficult to
formulate general rules on network design. While designing hydrologic networks, the decisions
to be taken are:

i. the variables to be measured and the frequencies and duration of observations;

ii. the location of gauging stations;
iii. the instruments to be installed and methods of observation; and
iv. data observation and transmission system.

Since the hydrometeorological data networks are operated by a number of independent

agencies, a good coordination among them is important. This will reduce the expenditure and
improve data quality. Of particular importance is the coordination between hydromet, water
quantity and quality data networks.

2.3.1 Classification of Observation Networks

Hydro-meteorological observation networks can be classified in many different ways. Based on
spatial features, there are two types of networks:
a) areal networks, such as those for precipitation, and ground water levels, and
b) linear networks such as those for streamflow and river sediment.

Areal networks are established to get spatial characteristic of the variables over an area
while the linear networks are created for rivers, canals, etc. On the basis of purpose, the
networks can be classified in three categories: basic (to get the fundamental characteristics of the
variables of interest), specific (to gather data for some specific purpose, e.g., a reservoir project),
and temporary (which are in operation for a short period of time).

Another criterion for classification of hydro-meteorological networks could be the

purpose for which the network is established. The purpose of a basic network is to provide a
level of hydrological information at any location within its region of applicability that would
preclude any gross mistake in water resources decision making (WMO, 1994). In the early stages
of development of a network, the first step should be to set up a minimum network. Such a
network should consist of the minimum number of stations required to initiate planning for
utilization of water resources management in the region.
 Depending on the purpose, the hydrological observation station could be classified into
three types:
1) Primary stations: These are also termed as key gauging stations, principal stations or
bench mark stations and are maintained on long term basis to generate representative
flow series of the river system.
2) Secondary stations: These are essentially short duration stations which are operated only
for short time period to establish the flow characteristics of the river.
3) Special purpose stations: These are also termed as specific purpose stations or project
stations or temporary stations meant for projects and are discontinued when the purpose
is served. The purpose could vary from design, management and operation of the project
to monitoring and fulfillment of legal agreement between co-basin states. Many a times,
the primary as well as the secondary stations also serve as special purpose stations.

The Guide to Hydrological Practices (2008) published by World Meteorological

Organization (www.wmo.ch) contains useful guidelines to set up networks for various types of
data, and observe and analyze the data. Bureau of Indian Standards (www.bis.org) have also
published various standards for Indian conditions.

2.4 Precipitation Networks

The optimum density of a precipitation gauge network depends on the purpose for which the data
are to be used. For example, measurements of precipitation for flood forecasting require denser
networks as compared to that for rainfall-runoff modeling. WMO (2008) has recommended the
following (Table 2.1) as minimum network densities for precipitation stations.

Table 2.1 Recommended minimum densities of stations (area in km2 per station) [Source: WMO
(2008)].
Physiographic unit Precipitation Evaporation
Non-recording Recording
Coastal 900 9000 50000
Mountains 250 2500 50000
Interior plains 575 5750 5000
Hilly/undulating 575 5750 50000
Small islands 25 250 50000
Urban areas - 10–20 -
Polar/arid 10000 100000 100000

The optimum network should make it possible to derive required information with desired
accuracy. The optimum number of rain gauge stations (N) in a network is given by (BIS 1968):
 N = [Cv /p]2 (2.1)

where Cv = the coefficient of variation of the precipitation values of the existing rain gauge
stations, and p = the allowable maximum percentage error in the estimate of basin mean rainfall.
A typical value of p is 10 percent. Here, Cv is computed by

Cv = 100*s/Pm (2.2)

In which s is the standard deviation and Pm is the mean rainfall of the existing stations.
Obviously, a decrease in the percentage error would mean an increase in the number of gauges
required. Mukherjee and Kaur (1987) have proposed a modified form of eq. (2.1) by including
the mean correlation (r) of precipitation over the area

N = [Cv /p]2(1 – r) (2.3)

WMO recommends that the precipitation (amount and form) should be measured with an
accuracy of 3–7% and rainfall intensity with 1 mm/hr at the 95 per cent confidence interval.
Snow depth below 20 cm should be accuracy of less than 1 cm and depth above 20 cm should
not have more than 10% error. The recommended accuracy for evaporation range 2–5% and for
wind speed 0.5 m/sec.

Example 2.1: A catchment has 6 rain gauges and the annual rainfall at these has been measured
as 750, 540, 465, 493, 421, and 780 mm. Find out the optimum number of rain gauges for the
basin if the error of estimation is required to be kept below 10%.

Solution: For the data given, mean = 574.83 mm and standard deviation = 152.59 mm. Thus

Cv = 100*152.59/574.83 = 26.54

Hence, using eq. (2.1), the optimum number of rain gauges for the basin (N) is

N = (26.54/10)2 = 7.04

This means that 7 rain gauge stations are required in the basin and the existing network of 6 rain
gauges is slightly inadequate. It needs to be strengthened by adding one new gauge so that the
estimate of rainfall depth has stipulated accuracy.

2.5 Stream Gauging Networks

A network of stream gauging stations is established in a river basin to provide data required by
the hydrologists for planning, development and management of water resources of the basin. The
collected data also enables to estimate the principal characteristics of the hydrological regime of
the basin.

Every major stream should be gauged near its mouth and its major tributaries should also
be gauged as feasible. Naturally, gauging depends on the existing and likely development in the
basin. According to WMO, the first gauging station is selected at the most upstream location
where the drainage area is about 1300 km2. The second station is located at a point in the
downstream direction where the drainage area is approximately doubled. The WMO
recommendations for a minimum density of stream gauging stations are given in Table 2.2.

Table 2.2 Recommended minimum densities of stations (area in km2 per station) [Source: WMO
(2008)].
Physiographic unit Streamflow Sediments Water quality
Coastal 2750 18300 55000
Mountains 1000 6700 20000
Interior plains 1875 12500 37500
Hilly/undulating 1875 12500 47500
Small islands 300 2000 6000
Urban areas - - -
Polar/arid 20000 200000 200000

Stations are also established in the basin at the sites where significant changes in the
volume of flow are noticed, for example downstream of the confluence of a major tributary or at
the outflow point of a lake etc. In case a suitable location is not available downstream of the
confluence, the sites can be located upstream of the confluence, preferably on the tributary.
While establishing sites at the downstream of confluence, it should be ensured that no other
small stream joins the main river before the station so that correct assessment of the contribution
of the tributary to the main river is obtained.

The distance between two consequent stations on the same river may vary from about 50
km to several hundred kilometers, depending on many factors. The drainage area computed from
origin up to two consecutive observation sites on a large river should preferably differ by more
than ten percent so that the difference in quantities of flow at the two stations is significant.
Sometimes stations are set up due to hydrological significance, say, to determine the flow
contribution from a typical catchment.

A different approach is recommended for small independent rivers which flow directly
into the sea (for example, the rivers in Western Ghats). In such cases, the first hydrological
observation station is to be established on a stream that is typical of the region and further
stations are added to the network to cover the area and obtain information about the variability.
Stream in the area whose flows are low should not be avoided from the network. Absence of
stations from low flow streams may lead to wrong assessment of the surface water potential of
the area if it has been evaluated just on the basis of the data from the high flow streams. Thus,
great care is to be exercised to ensure that all distinct hydrological features are adequately
covered by the gauging network.
An existing gauging network needs periodic review. The developments that take place in
the basin like construction of new water resources development projects may warrant addition or
closure of the sites. Often the rivers are polluted by the discharge of affluents from industries. A
need may also arise to establish stations to monitor the quality of water in the river.
Regarding the accuracy desired in measuring river water depth and discharge, WMO
recommends that the water depth measurement should have accuracy of about 2%, velocity of
flow 2–5%, and discharge about 5%. Suspended sediment concentration should be estimated
with accuracy of 10%.

2.5.1 Network Design Process

Design of networks is not a one-time affair. Factors affecting network design go on evolving
with time and thus the networks also require periodic review and adjustments. Design of
networks to measure stream gauge and discharge involves the following steps:
1. Network design begins with collection of maps and background information about the
area/region. Usually 1:250,000 scale topographical maps of the river basin showing basin
boundaries will form the base map for the network design. Smaller scale maps are of
limited use because it is difficult to identify the location of stations relative to key
features. It is also important to use an updated map. Ideally, the following maps should
also be collected:
i. Existing precipitation and gauge-discharge gauging stations operated by various
departments.
ii. Location of existing and proposed water projects and command areas of
irrigation projects.
iii. Land use map, also showing forests, main industries and population centres.
iv. Communications map showing roads, rails, power transmission lines, canals, etc.
v. Map showing soil classification, geological formation and mining areas.
2. Define the objectives of the network: who will be the data users and what will the use of
data? What is the required frequency?
3. Evaluate the existing network and find out how well it meets the objectives?
4. Review existing database to identify gaps, ascertain variability in catchment behaviour.
5. Identify gaps and over-design (if any) in the existing network; new stations may be
proposed and existing stations deleted shifted.
6. Prioritise stations by following appropriate classification system.
7. Estimate average capital and recurrent costs of installing and maintaining different
categories of stations and overall cost of operating and maintaining the network.
8. Decide on approximate location of sites and carryout site surveys.
 9. Review revised network in relation to overall objectives and available budget; adjust it as
necessary.
10. Prepare a realistic implementation plan which is achievable.

These steps and related topics are further elaborated in the following sections.

2.5.2 Criteria for Location of Stations

With particular reference to India, location of stream gauging stations is influenced by the
following factors:
• Places where major rivers cross State borders;
• Locations of proposed dams/diversion/run-of-river schemes including diversions or off-
takes/joining points for (proposed) inter-basin water transfers link canals;
• Locations whose data may be needed for flood forecasting;
• Conservation areas and areas of ecological interest;
• Areas of water supply shortages;
• Areas expected to have significant land use change, e.g., de-forestation or re-forestation;

2.5.3 Evaluation and Adequacy of Networks

To evaluate the networks, the existing network and proposed new stations should be marked on a
1:250,000 map. The catchment area for each river gauging station could be estimated from the
basin maps (hard copy or in GIS). Scanning the network systematically, the following questions
need to be considered for each station:
• What purpose will the station fulfill?
• Does a better location exist nearby?
• Have any developments (e.g. dam construction) taken place or are likely which could
affect this station?
• How close are the nearest upstream and downstream gauging stations? Two stations
should not be very close unless there are specific reasons.
• Does any other organisation operate a gauging station in the vicinity? If yes, could the
data from that station serve the purpose expected from this station?

Based on the answers, stations which can be added, deleted or relocated are identified.

Financial Aspects
In addition to technical financial considerations are also important in network design because the
stations cannot be established without adequate money, equipment cannot be purchased and
operated, and staff cannot be hired. Hence, after the preliminary design of the network has been
completed, the expenditure to establish stations and the cost of operating them should be
estimated. These monetary requirements should match with the budget so that the proposed
network is sustainable. In case of deficit in the budget, the network should be re-aligned or
additional budget should be arranged. Stations in the network may be prioritized to best attain the
objectives, given the constraints. Table 2.3 gives a possible categorization of stations based on
their relative importance.

Table 2.3: A possible categorization of stations based on their relative importance

Category Priority Relative Importance

A High Sites of major water resources project(s): existing or proposed;
important rivers crossing state boundary, large basins that are still
ungauged, heavily polluted river.
B Medium Sites of medium scale water resources projects: existing or proposed,
medium size river crossing state boundary, operation of a medium
project, area with (potential) water quality problems. All of these
with an existing gauging site at some distance away.
C Low Minor irrigation project site, gauging station on small tributary, major
river but with nearby gauging sites.

Prior to removing/shifting an existing station in the network, the main users and beneficiaries of
the data should be consulted.

2.5.4 Site Selection Surveys

Once the objectives have been defined and the techniques for measurement/recoding water level
and flow measurement have been finalized, the site selection process can begin. To select the
most appropriate site for a station, site selection surveys are carried out. These surveys can be
divided into four distinct phases:
1. Desk study,
2. Reconnaissance surveys,
3. Topographic surveys, and other surveys.

By now, the target location for the gauging station will have already been identified on a
1:250,000 or similar map in earlier steps. However, 1:250,000 is too small a scale for final site
selection purposes. Large-scale topographic maps (1:50,000) should be checked to identify
possible sites within the target zone. Reconnaissance surveys should be undertaken by an
experienced hydrologist along with a person familiar with the area. As the hydraulic conditions
and river characteristics vary considerably from non-monsoon to monsoon season,
reconnaissance survey in both the seasons would facilitate correct decision on the suitability of
the site. When the establishment of site cannot wait that long, the suitability of the site for
hydrological observation could be decided after single inspection assisted by toposheet studies
and other relevant field investigations including measurements of width and depth.
 At sites of interest, ownership of the land and approach should be ascertained. The site
shall be accessible in all seasons and all weather. It is important to use updated maps since most
surveys were completed several decades ago and things may have undergone large changes.
Recent situation can be obtained from remote sensing images or internet sites such as Google
Earth.
Information on the historical high flood level should also be collected (by local enquiry
and / or by examining the available landmarks) during the inspection. An all weather accessible
site located in a straight uniform reach free from weeds, rock outcrop, pools and back water
effect with stable non-overflowing banks with flow confined to single channel normal to the
selected cross-section of measurement would be an ideal site for stream gauging.

On completion of the reconnaissance surveys, one or more locations are shortlisted for
further consideration. After this, field surveys are carried out and the cross-section of the
proposed site is surveyed. If artificial controls (e.g. a weir) are planned, it will be necessary to
survey the river for some distance upstream and downstream to ascertain the impact on flows
and water levels. It will also be important to understand what type of control exists and to make
sure that the location will not be impacted by variable backwater effect of any structure.

2.5.5 General site selection guidelines

The following are the general site selection guidelines. Specific recommendations for different
types of stations are mentioned subsequently. It is emphasized that an ideal location which
satisfies all requirements can be found in very few real-life cases. In practice, it is often required
to take measurements in non-ideal conditions.
1) The approach channel should be of uniform cross-section and free from irregularities and
the flow shall have a regular velocity distribution. This can most readily be provided by
having a long, straight approach channel. There should be straight, uniform, well defined
approach channel upstream of the measuring section to ensure parallel and non-turbulent
flow. For rivers less than 100 m wide, a straight approach of 4 times channel width
should be preferred. For rivers more than 100 m wide, a straight approach channel of
minimum 400 m is desirable. When adequate length of straight channel is not present, the
straight length upstream should be at least twice that downstream.
2) Sites where high sediment deposition or scouring occurs or those which are subject to
weed growth should be avoided, if possible.
3) Locations which are subject to high turbulence or wind effects should be avoided.
4) In needs to be ensured that there is no parallel by-pass channel, natural or man-made, on
the surface or sub-surface, around the station.
5) The channel bed should be solid, relatively smooth and free from obstructions and debris.
6) The control shall be sensitive, such that a significant change in discharge, even for the
lowest discharges, should result in a significant change in stage. Small errors in stage
 readings during calibration at a non-sensitive station can result in large errors in the
discharges indicated by the stage-discharge relationship.
7) The station should be located where the flood plain is at its narrowest and the out-of-bank
flood flow is the minimum. It is often not possible to locate a gauging station so that all
flood flows are contained within the river channel. At many locations, there is an
elevation after which out-of-bank flow occurs.
8) The banks of the river should be high and steep and free from larger vegetation. Some
vegetation is desirable since this helps maintain the stability of the banks.
9) River banks at the site should be well-defined, stable, and free from vegetation and other
obstructions.
10) Downstream conditions should preferably be stable. Sites, which are influenced by
downstream confluences with other rivers, river control structures, dams, tidal conditions
or heavy weed growth, should be avoided. Such downstream conditions should be taken
into account when designing the structure to assess the modular limit.
11) Factors such as unhindered access to the site in all seasons, availability of office
accommodation, living space for the observers, electricity and other services should also
be taken into account.
12) Enough land should be available near the site to install various instruments.
13) Human interference (out of curiosity or with malafide intention) with hydrometric
installations is a problem in India. This issue has to be given serious consideration during
the site selection process. For example if a choice has to be made between two
hydraulically similar sites, the final selection should be made in favour of the site which
has fewer problems due to human interference and law and order.
14) Sites with a tendency for formation of vortices, reverse flow or dead water shall be
avoided.
15) The measuring section should be away from obstructions (artificial and natural) and
control structures, e.g., dams, weirs.
16) Channel at measuring section should be free from weed growth, accessible at all times of
the year and under all flow conditions, and must be safe to gauge.
17) For a station to be sustainable, manpower and logistic support to operate and maintain the
installation are necessary. Local manpower with desired qualification and interest is
always helpful.

2.5.6 Criteria for Water Level Gauging Sites

Water level or river stage is the primary variable that is measured at stream gauging sites and
most frequent measurements pertain to river stage. For stage monitoring, the following additional
site selection guidelines apply.
1. Steep banks or sides are preferred; the location should be selected so that for manual
observation the gauge posts are readable over the full water level range.
 2. The stage measurement device should be installed as close to the edge of the stream as
possible. Sections subject to high velocities should be avoided to the extent possible
since drawdown effects can occur around the device.
3. To minimize the effects of turbulence and high velocities, water level measuring devices
can be installed in a suitable stilling bay at the bank.
4. It is desirable to have access to the site and gauge posts at all times.
5. The site should not a tendency to collect floating debris which may hinder working of
water level measurement device.

2.5.7 Criteria for Streamflow Measurement Sites

Current meter is a commonly used instrument and velocity area method is the preferred approach
to measure river discharge. A stage and discharge measurement station should have appropriate
conditions to install a stage measurement device and to measure discharge. The required features
of a good discharge gauging site are as follows:
a. The measurement section should be clearly visible across its width and unobstructed by
trees, aquatic growth or other obstacles.
b. There should be sufficient depth of flow across the whole cross-section:
c. Sites with mobile beds and bank shall be avoided. In some rivers, this is not possible and
the site may be chosen so that the bed and bank changes are minimised.
d. Ideally, flow should be confined to a single channel. When this is not possible, each
channel should be gauged separately to obtain the total flow.
e. The site shall be sufficiently far away from the disturbance caused by rapids and falls.
f. If the site is upstream of confluence of two rivers, it should be located sufficiently far
upstream so that it is beyond backwater and any disturbance due to joining of two rivers.
g. Velocities should be well in excess of the minimum required speed of the current meter
over the full flow range.

2.5.8 Criteria for Natural Control Sites

The factors that are important in selecting a good site with natural control are summarised below.
1. If possible a natural control should be selected where the relationship between stage and
discharge is substantially consistent and stable.
2. The control should be sufficiently far upstream of another feature or control structure to
avoid inconsistencies due to variable backwater effects. The channel should be stable.
3. The general course of the stream should be straight upstream and downstream of the site.
Ideally the measuring reach should be straight for about 2 - 3 times the river width or a
minimum of 400 m (whichever is less) both upstream and downstream of the site.
4. Stable (unchanging) controls should be available in the form of a bedrock outcrop or
other stable riffle for low flows and a channel constriction for high flows; a fall or
cascade, which remains un-submerged over the full range of stage is ideal.
 5. A pool (deeper water) upstream of the control is helpful because it ensures the recording
of stage at low flows and avoids/dampens high velocities at observing/ recording, device
during high flows.

2.5.9 Criteria for Artificial Control Sites

A variety of flow measurement structures are used and the choice depends on a variety of factors
including objectives, flow range, afflux, size and nature of the channel, channel slope and
sediment load, operation and maintenance, and cost. The applications and limitations of a
structure will determine where its use is most appropriate. Each structure has its own specific site
selection criteria. Some general criteria to be considered are described here.

1) Generally the use of artificial controls should be limited to small but important rivers (<
100 m wide) and for special investigations in artificial channels.
2) Existing structures may be adapted for the purpose of flow measurement, wherever
feasible.
3) The sensitivity of upstream area to increased levels should be assessed. For example, will
the installation of the structure cause a potential, increased risk of flooding.
4) A minimum length of straight approach channel of five times the maximum width of the
water surface is recommended for most structures, except for thin plate weirs where ten
times the maximum channel width is recommended. However, research has shown that
for triangular profile weirs accurate results can be obtained even if the weir is only twice
the channel width from an upstream bend.
5) Thin plate weirs are particularly sensitive to upstream velocity distribution.
6) Like all controls, it is essential that the structure creates a sensitive stage-discharge
relationship. In wider rivers, this can be a problem at low flows. Structures such as the
triangular profile flat “v” weirs provide such sensitivity.
7) The discharge coefficients of many structures vary when the velocity head in the
upstream approach channel becomes large in relation to the depth of flow. A
dimensionless number which describes this is the Froude number (Fr). To prevent water
surface instability in the approach channel the Froude number should generally not
exceed 0.5.
8) The design of the structure should be such as to minimise upstream sediment deposition
or downstream scouring. In rivers with high bed loads the use of structures which
significantly reduce the stream velocity is not recommended.
9) On rivers which are navigable or those which are important fish migration routes the use
of flow measurement structures should be avoided and some other form of flow
measurement considered.

2.5.10 Bureau of Indian Standards (BIS) criteria for selection of river gauging sites
The ideal requirements for a good gauging site as enunciated in the standard IS 1192-1981
"Velocity - Area methods for measurement of flow of water in open channels" are given below.
The accuracy of measurement of discharge by velocity area method is increased if the site is
selected considering these aspects.

The site selected should comply, as far as possible, with the following essential requirements:
a) The reach of the open channel at the gauging site shall be straight and of uniform cross
section and slope, as far as possible, in order to avoid irregularities in velocity
distribution. The length of the reach need not be more than 1600 m and should not be less
than 400m. When the length of the straight channel is restricted, it is recommended for
current meter measurements and the straight length upstream of the measuring cross
section should be twice that on the downstream.
(Note: In case of artificial channel, the minimum length of straight reach should
preferably be such as to give a drop in water level of 0.06 m. or the minimum length
should be equal to four times the width of the channel, whichever is larger.)
b) The depth of water in the selected reach shall be sufficient to provide for the effective
immersion of either the current meters or floats, whichever are to be used.
c) When near a confluence, the site, if located on a tributary shall be sufficiently upstream
preferably beyond the backwater effect; and if located on the main stream, upstream or
downstream of the confluence it shall be beyond the disturbances due to the tributary.
d) The site should be easily accessible at all times of the year.

In addition to the above requirements, the following points shall be taken into
consideration as desirable requirements in the selection of the gauging site.
a. The flow should be confined in a single channel and there should be no overflow as far as
possible. Where this is not possible, the site in which minimum number of channels exist
and the flood plain has minimum width should be preferred.
b. Where these requirements cannot be met (for instance- when in alluvial rivers the river
bed is changing during the period of measurement, or when, under flood conditions, the
river is not confined to a single channel in embankments), a gauging site shall be chosen
such that the bed change and/or overflow is minimum. Floodplain, if cannot be avoided,
shall be of minimum width, as smooth as possible, with a distinct channel, and clear of
bushes and trees. The flow in the over bank or floodplain section (s) shall be measured
separately and added, treating the whole as a composite section.
c. The site shall be remote from any bend or natural or artificial obstruction if disturbances
of the flow are likely to be caused thereby.
d. The orientation of the reach should be such that the direction of flow is as close as
possible normal to that of the prevailing wind.
e. Sites at which there is a tendency for vortex formation should be avoided.
 f. The site should, as far as possible, be free from trees and obstructions which may
interfere with flow and clear vision during observation.
g. The site shall be free from aquatic growth which is likely to interfere with the
measurement of depth and the current meter reading.
h. The site shall be away from the back water zone caused by any structure on the river.
i. The site should be sufficiently away from the disturbance caused by rapids and falls. etc.

2.5.11 World Meteorological Organisation (WMO) criteria for selection of site

The following are the WMO recommendations for selection of a site:
I. The general course of the stream should be straight for about 100 m upstream and
downstream from the site.
II. No flow bypasses the site as subsurface flow.
III. The stream bed is not subject to scour and fill.
IV. The banks are permanent and high enough to contain floods.
V. Unchanging natural controls are present in the form of a bedrock outcrop or other ruffle
for low flow and a channel constriction for high flow.
VI. Small pool is present upstream from the control at extremely low stages to ensure a
recording of stage at extremely low flow and to avoid high velocities
VII. A satisfactory reach for measuring discharge at all stages is available within reasonable
proximity of the gauge site.

2.5.12 International Standards for Hydrometry

International Organization for Standardization (ISO) has published a large number of standards
related to Hydrometry. A partial list is given in Table 2.4. For further details, their website
(www.iso.org) may be referred. A brief description of the provisions of relevant ISO standards is
given in Annexure A.
2.6 Errors in Hydrological Observations
Theoretically, the true values of hydrological variables cannot be determined by measurements
because errors of measurement cannot be eliminated completely (WMO 2004). Errors arise in
hydrometric measurements due to several reasons. Most common causes are:
 Faulty equipment, e.g., a current meter with worn-out or damages parts;
 Malfunction of instrument, e.g., slippage of float tape in water level recorder;
 Improper exposure conditions of the instruments e.g., a rain gauge surrounded by high
rise buildings.
 Observation errors by the observer, e.g., gauge misread;
 Wrong entry of data in records/computer; and
 Error in computation, e.g., mistake while converting current meter rotations to velocity.

Table 2.4 A partial list of ISO standards related to Hydrometry.

ISO Number Details
ISO 748:2007 Hydrometry -- Measurement of liquid flow in open channels using current
meters or floats
ISO 772:2011 Hydrometry -- Vocabulary and symbols
ISO 1070:1992 Liquid flow measurement in open channels -- Slope-area method
ISO 1100- Hydrometry -- Measurement of liquid flow in open channels -- Part 2:
2:2010 Determination of the stage-discharge relationship
ISO 1438:2008 Hydrometry -- Open channel flow measurement using thin-plate weirs
ISO 2425: 2010 Hydrometry – Measurement of liquid flow in open channels under tidal
conditions
ISO 3846:2008 Hydrometry -- Open channel flow measurement using rectangular broad
crested weirs
ISO 4359:1983 Liquid flow measurement in open channels -- Rectangular, trapezoidal and U-
shaped flumes
ISO 4360:2008 Hydrometry -- Open channel flow measurement using triangular profile weirs
ISO 4362:1999 Hydrometric determinations -- Flow measurement in open channels using
structures -- Trapezoidal broad-crested weirs
ISO 4373:2008 Hydrometry -- Water level measuring devices
ISO 4374:1990 Liquid flow measurement in open channels -- Round-nose horizontal broad-
crested weirs

It needs to be stressed that no statistical analysis can replace correct observations because
spurious and systematic errors cannot be eliminated by such analysis. Only random errors can be
characterized by statistical means.
 Errors
E in hydrologic meeasurements can be claassified in tthree categoories: system matic,
random, and spurious. Fig. 2.3 giives a graph
hical depictioon of errors. These are ddiscussed in ddetail
in the folllowing.

Fig. 2.3 Explanation

n of measurement errors [Source: WM
MO, 2008].

Systema atic Errors

A systemmatic error or
o bias is a systematic difference, either positiive or negattive, betweeen the
measured d value andd the true vaalue. System matic errors arise mainlly due to m malfunctioninng of
instrumen nt. Hence, if
i the instruuments and measureme nts conditioons remain unchanged, such
errors cannot be reduuced just by increasing the
t number of measurem ments. Systeematic errorss also
arise ofteen due to diifficult meassuring condditions, such as unsteadyy flow, meaandering andd bad
location of observattion stationss, and lack of knowleddge of obseerver. Such error shoulld be
eliminateed by propeerly adjusting g, repairing, or changinng the instruument, and by changinng the
measurem ment conditiions. For exaample, this can
c be done by straightenning the appproach channnel of
a stream--gauging secction. If the systematic
s error
e has a knnown value,, this should be accounteed for
appropriaately and errror due to this source should be considered zero. Systeematic errorrs are
generally
y more seriouus and the vaalidation proocess must bbe able to dettect and corrrect them.

Regarding
R prrecipitation, WMO (19882) listed thee following errors for wwhich adjusttment
needs to be made to get a near accurate
a estiimate of preecipitation frrom a measuured precipittation
report.

(a) error due to the system

matic wind field
f deformation above the gauge orrifice
(b) error due to the wettin
ng loss on th
he internal w
walls of the collector
 (c) error due to evaporation from the container (generally in hot climates)
(d) error due to the wetting loss in the container when it is emptied
(e) error due to blowing and drifting snow
(f) error due to splashing in and out of water, and
(g) random observational and instrumental errors.

The first six errors listed above are systematic and are listed in order of general importance. The
net error due to blowing and drifting snow and due to splash in and out of water can be either
negative or positive while net systematic errors due to the wind field and other factors are
negative. Since for liquid precipitation the errors listed at (e) and (f) above are near zero, the
general model for adjusting the data from most gauges takes the form

Pk = K (Pg + P1 + P2 +P3) (2.4)

where Pk = adjusted precipitation amount

K = adjustment factor for the effects of wind field deformation
Pg = the measured amount of precipitation in the gauge
P1 = adjustment for the wetting loss in the internal wells of the collector
P2 = adjustment for wetting loss in the container after emptying
P3 = adjustment for evaporation from the container

The data needed to make the adjustments include wind speed, drop size, precipitation intensity,
air temperature, humidity and other characteristic of the gauge site.

Random Errors
Random errors vary in an unpredictable manner, both in magnitude and sign, when repeated
measurements of the same variable are made under the same conditions. Random errors cannot
be eliminated, but their impacts can be reduced by repeated measurements of the variable. These
are equally distributed about the mean or ‘true’ value. The errors of individual readings may be
large or small, e.g., the errors in a staff gauge reading where the water surface is subject to wave
action. Usually, they compensate with time or are minimized by taking a sufficient number of
measurements. The uncertainty of the arithmetic mean computed from n independent
measurements is several times smaller than the uncertainty of a single measurement. The
distribution of random errors can usually be assumed to be normal (Gaussian). For certain cases,
normal distribution can or should be replaced by other statistical distributions. These errors can
be identified by a statistical-outlier test that gives a rejection criterion.

In measuring rainfall, random errors could arise due to spilling of the water when
transferring it to the measuring jar, leakage into or out of the receiver, observational error etc.
The others random errors which could be due to observer include:
 i. misreading and transposing digits,
ii. misrecording because of faulty memory,
iii. recording the data at the wrong place on the recording sheet,
iv. making readings at improper interval,
v. incorrect dating of the report,
vi. incorrectly reading or communicating the data to a reporting centre, etc.

It appears, therefore, that computerization and automation may be solution to reduce the
error. However, even without human intervention chances of erroneous reading in case of
precipitation may be possible because of
i. evaporation from gauge,
ii. overflowing gauge,
iii. mechanical or electrical mal-functions.

Spurious errors
These arise due to human mistakes or instrument malfunction or some abnormal external cause.
Reported data appear to be clearly in error. Sometimes the errors become obvious, for example,
wrong placement of decimal and the data can be easily corrected in such cases. In other cases,
the concerned measurements may have to be discarded. For example, an animal may drink water
from the evaporation pan and introduce errors in the data. Sometimes, such errors may be readily
detected but it may not be easy to correct them.

2.6.1 Sources of Errors

After understanding the types of errors, the next obvious question will be about their sources. It
would not be possible to list all likely sources of error because there are different instruments and
measuring methods and each of these will have their own sources of error. Some typical sources
of error were given by (WMO 2008):
a) Datum or zero error originates from the incorrect determination of the reference point of
an instrument, for example, staff-gauge zero level, difference between the staff-gauge
zero and the weir-crest levels;
b) Reading (or observation) error results from the incorrect reading of the value indicated by
the measuring instrument. This error is normally attributed to neglect or incompetence of
the observer. It could also arise, for example, due to bad visibility, waves, or ice at the
staff gauge;
c) Interpolation error is due to inexact evaluation of the position of the index with reference
to the two adjoining scale marks between which the index is located;
d) Error due to wrong assumption or neglect of one or more variables needed to determine
the measured value (for example, assuming a unique stage-discharge relationship during
periods of unsteady flow when discharge depends on slope as well as stage);
 e) Hysteresis;
f) Insensitivity error arises when the instrument cannot sense the small change in the
variable being measured;
g) Non-linearity error is that part of error whereby a change of indication or response
departs from proportionality to the corresponding change of the value of the measured
quantity over a defined range;
h) Drift error is due to the property of the instrument in which its measurement properties
change with time under defined conditions of use, for example, mechanical clockworks
drift with time or temperature;
i) Instability error results from the inability of an instrument to maintain certain specified
metrological properties constant;
j) Out-of-range error is due to the use of an instrument beyond its effective measuring
range, lower than the minimum or higher than the maximum value of the quantity, for
which the instrument/installation has been constructed, adjusted, or set (for example,
unexpected high water level);
k) Out-of-accuracy class error is due to the improper use of an instrument when the
minimum error is more than the tolerance for the measurement.

It may be emphasized here that uncertainty in measurement has a probabilistic character.

Therefore, one can define an interval in which the true value of the variables is expected to lie
with a certain confidence level. If measurements are independent, then the uncertainty in the
measurements can be estimated by taking a minimum of a large number (say > 25 observations)
and calculating the standard deviation. A problem in applying statistics to hydrological data is
that many hydrological variables are assumed to be independent random variables. For many
hydrologic variables, this assumption is not strictly valid. For example, short-term river flows are
correlated.

2.6.2 Secondary Errors of Measurement

Many hydrological variables are not directly measured but are estimated form the measured
values of several variables. For example, discharge passing through a site may be estimated by
the stage at that site or discharge at a weir may be computed as a function of a discharge
coefficient, characteristic dimensions and head. If the individual components and their errors are
assumed to be statistically independent, the resultant uncertainty (also referred to as overall
uncertainty) can be calculated from the uncertainties of the individual variable.
Let a quantity, Q, be a function of three measured quantities, x, y and z, and the
uncertainty in Q be denoted by eQ. Further, let the uncertainties in variables x, y and z be given
by ex, ey and ez, respectively. We can estimate the uncertainty eQ by applying the Gauss error
transfer theorem (WMO 2008):

2 2 2
 Q   Q   Q 
(eq )  
2
e x    e y    ez 
 x   y   z 
 (2.5)

where ∂Q/∂x, ∂Q/∂y and ∂Q/∂z are the partial differentials of the function expressing explicitly
the relationship of the dependent variable with the independent variables.

2.7 Validation of Hydrologic Data

As described above, measured raw data may have errors. Errors may also arise in data entry,
during computations and (hopefully very rarely), from the mistaken ‘correction’ of ‘right’ data.
Reliability of the data determines whether they are suitable in various applications or not. Use of
erroneous data may do more harm than good and may introduce an error of unknown order in
any decision that has been taken by the use of such data. It is, therefore, necessary that the data
are checked for any possible error and these are removed before the data are used in analysis,
design, and decision making. Thus, the need for data validation or quality control arises because
field measurements are subject to errors. Data validation aims at detecting and removing these
likely errors and inconsistencies in the data.

Data validation is the means by which data are checked to ensure that the corrected
values are the best possible representation of the true values of the variable. Data validation
procedure includes primary and secondary data validation. Validation of hydrologic data must
never be considered as a purely statistical or mathematical exercise. Staff involved in it must
have a background in hydrology and must understand the field practices. To understand the
source of errors, one must understand the method of measurement or observation in the field and
the typical errors of given instruments and techniques. Knowledge of the method of
measurement or observation influences our view of why the data are suspect.

Basically, data validation is carried out:

 to correct errors in the observed values where possible,
 to assess the reliability of data even though it may not be possible to correct errors, and
 to identify the source of errors to ensure that these are not repeated in future.

The input variables in an analysis may be directly measured (e.g., rainfall) or they may be
derived using a relationship with one or more variables (e.g., discharge that has been obtained
from a rating curve). In the latter case the error in the variable (discharge) depends both on field
measurements and the error in the relationship. An error may also be introduced if the
relationship is no longer valid or the values are extrapolated outside the applicable range.
Validation involves different types of comparisons of data and includes the following:

Single series comparison

 between individual observations and pre-set physical limits;
  between sequential observations to detect unacceptable rates of change and deviations
from acceptable behaviour (most readily identified graphically); and
 between two measurements of a variable at a single station, e.g., daily rainfall from a
daily gauge and an accumulated total from a recording gauge.

Multiple stations/data
 between two or more measurements at nearby stations, e.g. flow at two sites along a
river; and
 between measurements of different but related variables, e.g., rainfall and river flow.

2.7.1 Levels of Validation

Validation of data is best done soon after observation and at the observation station because
secondary or related information to support validation is readily available. However, data
validation at observation sites may not be always possible due to logistics and the lack of trained
personnel. Validation of hydrological data can be grouped in three major categories: a) primary
validation, b) secondary validation, and c) hydrological validation.

Note that none of the procedures of data validation are absolutely objective and there is
no guarantee that all the errors will be captured and removed. They are basically tools to screen
out suspect data which are to be further examined by other tests and corroborative facts. When it
is ascertained that a particular value is incorrect, an alternative value that is likely to be closer to
the true value is substituted. Since each hydrological variable has distinct characteristics, it is
necessary that specific validation techniques be designed for each variable. Further validation a
pure statistical exercise; the properties and behavior of the variable under consideration should
always be kept in mind.

2.7.2 Primary Validation

Primary data validation is done to highlight and, if possible, correct those data which are not
within the expected range. Primary validation involves comparisons within a single data series or
between observations and pre-set limits and/or statistical range of a variable or with the expected
behavior of the generating process. Sometimes, information from a few nearby stations may also
be pooled. If it is not possible to definitely conclude that the suspected value is erroneous, such
value is not changed but is flagged indicating that it is doubtful. All data which have been
flagged as suspicious during primary validation are again screened later on the basis of additional
information.

2.7.3 Secondary Validation

After primary validation secondary validation of data is taken up to for expected spatial behavior
of the variable as inferred from neighboring observation stations. It is assumed that the variable
under consideration has spatial correlation within small distances. This assumption must be
supported by the underlying behavior of the process under examination. Checks applied at this
stage basically examine if the data at the station under consideration is spatially consistent with
the data of the surrounding stations. The spatial validation and consistency check is carried out
using the data of key stations which are known to be of good quality.

When hydrological variables have a high auto-correlation, such as ground water levels, or
the data has high correlation with neighboring stations, the validation and data correction can be
carried out with a higher level of confidence. However, processes such as convective rainfall
show a great temporal and spatial variability. It is difficult to ascertain the behavior of such
processes with the desired degree of confidence. Based on available information and statistical
properties, if it is not possible to conclude whether the suspected value is erroneous or not, such
value is not changed but is flagged as doubtful. All doubtful data are further validated on the
basis of additional information.

2.7.4 Hydrological Validation

Here the basic idea is to correct erroneous data by the use of hydrological knowledge consists of
comparing data of correlated variables at nearby stations to identify inconsistencies between the
time series or their derived statistics. This test can be applied to a measured variable (water level)
or to derived variables (flow, runoff) and is usually done through regression or simulation
modeling.

Ideally all the hydrological data should be subjected to hydrological validation. For
historical data to which no (or few) checks have been applied, hydrological validation provides
an effective check on the quality and reliability of records. Thorough hydrological validation
requires a high level of professional expertise and can be time consuming. Required man power
and time may not be available always. Therefore, this validation may be applied selectively.
Finally, the validation may be able to identify a particular section of record/ data item that is
unreliable, but it may not always be possible to correct the values.

2.7.5 Validation of Climatic Data

Climatic data are known to have high spatial correlation depending upon topography. Validation
is mainly concerned with spatial comparisons between neighboring stations to identify anomalies
in recording at the station. Methods of validation can be classified in two groups.

(i) Single station validation tests for homogeneity,

(ii) Multiple station validation.

2.7.6 Single series tests of homogeneity

Single series testing for homogeneity will normally only be used with long data sets. Series may
be inspected graphically for evidence of trend and this may often be a starting point. However,
statistical hypothesis testing can be more discriminative in distinguishing between expected
variation in a random series and real trend or more abrupt changes in the characteristics of the
series with time.

Trend analysis (time series plot)

A series can be considered homogeneous if there is no significant linear or curvilinear trend in
the time series of the climatic element. The presence of trend in the time series can be examined
by graphical display and/or by using simple statistical tests. The data are plotted on a linear or
semi-logarithmic scale with the climatic variable on the Y-axis and time on the X-axis. The
presence or absence of trend may be seen by examination of the time series plot. Mathematically
one may fit a linear regression and test the regression coefficients for statistical significance.

Trend generally does not become evident for a number of years and so the tests must be
carried out on long data series, often aggregated into annual series. Trend may result from a wide
variety of factors including:
 Change of instrumentation
 Change of observation practice or observer
 Local shift in the site of the station
 Growth of vegetation or nearby new buildings affecting exposure of the station
 Effects of new irrigation in the vicinity of the station (affecting humidity, temperature
and pan evaporation)
 Effects of the urban heat island with growing urbanisation
 Global climatic change

The presence of trend does not necessarily mean that part of the data are erroneous but
that the environmental conditions have changed. Unless there is reason to believe that the trend is
due to instrumentation or observation practices or observer, the data should not generally be
altered but the existence of trend noted in the station record.

Residual mass curve

A residual mass curve represents accumulative departures from the mean. It is a very effective
visual method of detecting climatic variability or other inhomogeneities. The residual mass curve
can be interpreted as follows:
 an upward curve indicates an above average sequence
 a horizontal curve indicates an about average sequence
 a downward curve indicates a below average sequence

2.7.7 Multiple stations validation

The simplest and often the most helpful means to identify anomalies between the data of
multiple stations by plotting time series of the data of stations on same or adjacent graphs. This
should generally be carried out in the very beginning, before other tests. For climate variables the
series will usually be displayed as curves of a variable at two or more stations where
measurements have been taken concurrently, for example atmospheric temperature, dry bulb
temperature, or sunshine hours. It is important to keep the same scale so that the visual
impression is right.

While examining the current data, the plot should include the past time series of
sufficient length to ensure that there are no discontinuities between one batch of data received
from the station and the next. This will ensure that the data are being entered against the correct
station and correct date/time.

For climatic variables, which have strong spatial correlation, such as temperature, the
graphs will generally run along closely, with nearly the same variation with the mean separation
representing some location factor such as altitude. Abrupt or progressive straying from this
pattern will be evident from the comparative plot, which would not necessarily have been
perceived at primary validation from the inspection of the single station. An example might be
the use of a faulty thermometer, in which there might be an abrupt change in the plot in relation
to other stations. An evaporation pan affected by leakage may show a progressive shift as the
leak develops. This would permit the data processor to delimit the period over which suspect
values should be corrected.

Comparison of series may also help in accepting of values which might have been
suspected in primary validation because they fell outside the warning range. Where two or more
stations display the same behavior, there is strong evidence to suggest that the values are correct.
An example might be the occurrence of an anomalous atmospheric pressure in the vicinity of a
tropical cyclone.
Comparison plots provide a simple means of identifying anomalies but not of correcting
them. This may be done through correlation or regression analysis, spatial homogeneity testing
(nearest neighbor analysis) or double mass analysis.

Residual series
An alternative method of displaying comparative time series is to plot the differences. This
procedure is often applied to river flows along a channel to detect anomalies in the water balance
but it may equally be applied to climate variable to detect anomalies and to flag suspect values or
sequences.

Let X1 and X2 be two variables. The difference series is computed as

Yi = X1,i – X2,i (2.6)

Regression analysis
Regression analysis is a very commonly used statistical method. In the case of climatic variables
where individual or short sequences of anomalous values are present in a spatially conservative
series, a simple linear relationship with a neighboring station may well provide a sufficient basis
for interpolation. Correlation and regression have been covered in detail in a later module in this
course.

In a plot of the relationship, the suspect values will generally show up as outliers but such
plots provide no indication of the time sequencing of the suspect values. One will not be able to
say whether the outliers were scattered or contained in one block. For seasonal data, the
relationship should be derived for a period within the same season as the suspect values. (The
relationship may change between seasons). The identified suspected values should be removed
before deriving the relationship, which may be applied to compute corrected values to replace
the suspect ones.

Double mass curves

Double mass curve analysis, cumulative plots of variable under consideration at one station and
surrounding stations may also be used to show trends or in homogeneities between climate
stations but it is usually used with longer, aggregated series. However, in the case of a leaking
evaporation pan, described above, the display of a mass curve of daily values for a period
commencing some time before leakage commenced, the anomaly will show up as a curvature in
the mass curve plot.
This procedure may only be used to correct or replace suspect values where there has
been a systematic but constant shift in the variable at the station in question, i.e., where the plot
shows two straight lines separated by a break of slope. In this case the correction factor is the
ratio of the slope of the adjusted mass curve to the slope of the unadjusted mass curve. Where
there has been progressive departure from previous behavior, the slope is not constant as in the
case of the leaking evaporation pan, and the method should not be used.

Spatial homogeneity (nearest neighbor analysis)

This procedure is most commonly used for rainfall but can be used for other variables also. Its
advantage for rainfall in comparison to climate is that there are generally more rainfall stations in
the vicinity of the target station than there are climate stations. The advantage for some climate
variables is that there is less spatial variability and the area over which comparison is permitted
may be increased. This topic is discussed in greater details in module 3.

Closure
The science of hydrology deals with immense volumes of data of a number of variables. Since
data collection is expensive process, it is necessary that the data collection campaigns are
carefully planned and executed. Errors may creep in the data due to various causes. Therefore,
before the data can be put to use, it needs to be screened and validated to remove these errors.
After the data has undergone quality control checks, it is ready for use in planning, design, and
operation.

Annexure A
2.A Definitions of terms related to measurement errors

Definitions of some important terms related to accuracy have been sourced from WMO (2008)
and are given below.

Accuracy: The word accuracy is generally used to indicate the closeness or the agreement
between an experimentally determined value of a quantity and its true value. It is the extent to
which a measurement agrees with the true value. An accurate result closely agrees with the
actual value for that quantity. In other words, accuracy tells us how close a measurement is to an
accepted standard. Precision describes how well repeated measurements agree with each other. It
tells us how close two or more measurements agree. It is worth mentioning here that precision
does not necessarily indicate anything about the accuracy of the measurements. An experiment is
considered good when it is both precise and accurate.
An experiment is said to have high precision, if it has small random error. It is said to
have high accuracy, if it has small systematic error. There may be four possibilities for
characterizing the obtained experimental data, as shown in Fig. 2.A.1: (B) precise and accurate,
(A) precise and inaccurate, (D) imprecise and accurate, and (C) imprecise and inaccurate. In
hydrological observation, the objective is to reduce both systematic and random errors as much
as possible. However, for economy of effort, one must try to strike a balance between these two
sources of error, giving greater weight to the larger of the two.

B
A
Precision

C D

Accuracy
Fig. 2.A1 Measurement of rainfall by four rain gauges. Gauge A is precise, inaccurate; gauge B
is precise, accurate; gauge C is imprecise, inaccurate; gauge D is imprecise, accurate. The
innermost circle indicates the true value.

Confidence interval: The interval which includes the true value with a prescribed probability and
is estimated by using the sample data.
Confidence level: The probability that the confidence interval includes the true value.
Error: The difference between the result of a measurement and the true value of the quantity
measured. This term is also used for the difference between the result of a measurement and
the best approximation of the true value, rather than the true value itself. The best
approximation may be a mean of many measurements.
Expected value: The best approximation of the true value, which may be a mean of many
measurements.
Measurement: An action intended to assign a number as the value of a physical quantity in stated
units. The result of a measurement is complete if it includes an estimate of the probable
magnitude of the uncertainty.
Precision: The closeness of agreement between independent measurements of a single quantity
obtained by applying a stated measurement procedure several times under prescribed
conditions.
Accuracy has to do with closeness to the truth, precision has to do only with closeness together.
Precision of observation or of reading is the smallest unit of division on a scale of measurement
to which a reading is possible either directly or by estimation.
Random error: That part of the error that varies in an unpredictable manner, in magnitude and in
sign, when measurements of the same variable are made under the same conditions.
Range: The interval between the minimum and maximum values of the quantity to be measured,
for which the instrument has been constructed, adjusted or set. It can be expressed as a ratio of
maximum and minimum measurable values.

2.11 REFERENCES
BIS (1968). Recommendations for establishing network of raingauge stations. Bureau of Indian
Standards, New Delhi.
Herschey, R. W. (Editor). (1978). Hydrometry Principles and Practices. John Wiley and Sons,
New York.
Herschey, R. W. (Editor). (1995). Streamflow Measurement. Chapman & Hall, E & F N SPON,
London.
ISO (1982). Liquid flow measurement in open channels – part 2: Determination of stage-
discharge relation, ISO 1100/2-1982(E). International Standards Organization, Switzerland.
ISO (1983). Measurement of Liquid Flow in Open Channels. ISO Standards Handbook 16.
International Organization for Standardization, Switzerland.
ISO (1996). Hydrometry — Measurement of liquid flow in open channels — Part 1: Guidelines
for selection, establishment and operation of a gauging station. ISO 1100-1:1996,
International Organization for Standardization, Geneva.
WMO(1988). Manual on water quality monitoring -- planning and implementation of sampling
and field testing. Operational Hydrology Report No. 27, WMO No. 680. World
Meteorological Organization, Geneva.
WMO(1989). Management of ground water observation programmes. Operational Hydrology
Report No. 31, WMO No. 705. World Meteorological Organization, Geneva.
WMO (2008). Guide to Hydrological Practices. Volume I: Hydrology – From Measurement to
Hydrological Information. WMO No. 168. World Meteorological Organization, Geneva.
 Module 3
Measurement and Processing of Precipitation Data

3.1 Measurement of Precipitation Data

3.2 Precipitation Gauges
3.2.1 Rain Gauges
3.2.2 Autographic rain gauge
3.2.3 Tipping Bucket Rain Gauge
3.2.4 Precipitation Measurement by Weather Radar
3.3 Measurements Related to Snow
3.3.1 Measurement of Snow Cover and Properties by Remote-Sensing
3.4 Processing of Rainfall Data
3.4.1 Internal consistency check
3.4.2 Scrutiny of Precipitation Data by Multiple Time Series Graphs
3.4.3 Correction and completion of rainfall data
3.4.4 Spatial Consistency Check
3.5 Spatial Averaging of Rainfall Data
3.5.1 Arithmetic Average
3.5.2 Thiessen Polygon
3.5.3 Isohyetal Method
3.6 Estimation of Missing Data
3.6.1 Normal Ratio Method
3.6.2 Distance power method
3.7 Disaggregation of Rainfall Data
3.8 Estimation of mean areal precipitation in mountainous areas
3.9 Storage of precipitation data in data bank
3.10 Rain Storm Analysis

Keywords: Measurement, Precipitation, Rain Gauges, Tipping Bucket, Weather Radar, Snow,
Processing, Spatial Interpolation, Thiessen Polygon, Isohyetal, Filling-in, Normal Ratio,
Disaggregation

The objectives of this module are:

• To explain the methods of measurement of precipitation data,
• To discuss techniques of validation and processing of precipitation.
• To explain the distribution of daily precipitation data into shorter duration. Methods of
mean areal precipitation estimation in plain and mountainous areas are explained with
illustrative example to enable participants to achieve the capability of carrying out the
analysis themselves.

3.1 MEASUREMENT OF PRECIPITATION DATA

The amount, intensity and spatial distribution of precipitation are important inputs in most
hydrological studies. The total amount of precipitation which reaches the ground in a stated
period is expressed as the depth to which it would cover a horizontal projection of the earth's
surface. If any part of the precipitation is snow or ice, its depth when melted is included (WMO,
1994). The unit of precipitation is length and daily amounts should be read to the nearest 0.1
mm. Weekly, fortnightly, and monthly amounts should, however, be read to the nearest 1 mm at
least. In India, daily observations of precipitation along with other meteorological variables are
made at 08:30 am.

3.2 Precipitation Gauges

Precipitation (which includes solid and liquid part) is measured using a gauge which consists of a
collector to delineate the area of the sample and collect it and a funnel which leads to a sample
storage device. The precipitation, after it is collected, is measured by transferring the contents to
a graduated measuring jar. Different types of gauges are used to measure liquid and solid (snow)
precipitation.

Since the size, shape and exposure affect the precipitation caught by a gauge, standard
gauges are used so that the observations can be compared and there is consistent use of data. The
Symon's rain gauge has been adopted as the standard rain gauge In India. To install the gauge, a
masonry or concrete platform is sunk into the ground and the gauge is placed on it such that the
rim of the gauge is perfectly levelled and is approximately 30 cm above the ground. In a non-
recording instrument, the precipitation measured at 8:30 am on any particular date is entered
against that date and it is understood that this precipitation has been received in 24 hours
preceding the time and day of observation.

3.2.1 Rain Gauges

Basically, there are three types of rain gauges:

 Standard or ordinary rain gauges (ORG) are manually read, commonly once a day.
 Self-recording rain gauges (SRRG) record the rainfall depth in the form of a continuous plot.
 Automatic rain gauges with data logger. Data are stored in digital form either as rainfall at
fixed interval or as timings for each event of rainfall of fixed depth.

An ordinary rain gauge or ORG (see Fig. 3.1) is a simple device consisting of a collector
and a funnel. A standard rain gauge consists of a circular collector funnel with a brass or gun
metal rim and the rainfall collection area of either 200 cm2 (diameter 159.5 mm) or 100 cm2
(diameter 112.8 mm). The funnel leads to a base unit, partly embedded in the ground and
containing, a polythene or glass collector bottle. The gauge is read once (usually) or twice daily
and any rain gathered in the collector is poured into a measuring graduated measuring glass
cylinder to determine rainfall depth in millimetres.
Since rainfall received by the ORG is measured and recorded manually, errors may be
introduced by the observer in many ways. Observer may read the measuring cylinder incorrectly
or may incorrectly note
n the amo ount in the data
d sheet. O
Observer maay also read the gauge aat the
wrong tim me or may noten the am
mount againstt the wrong date. Obserrver may alsso use the w wrong
2 2 2 2
measurinng glass (e.g g., 200 cm glass
g for a 100
1 cm gauuge or 100 ccm glass forr 200 cm gauge.
Hence prrecipitation will
w either bee half or twiice the true vvalue). Besiides human ccause, errorss may
also arisee due to insttrument faultt: if the gaugge rim is dam
amaged, the collection arrea is changged; if
the funneel is blocked, water willl not reach the collectiion bottle – it may oveerflow the riim or
(partly) lost
l by evap poration. Colllector bottlee may be brroken and gaauge might be leaking. Note
that errors due to most
m of these causes wiill be difficuult to detect, particularrly if only a few
records are
a present, unless a veery large errror has beenn committedd in reading or reportinng the
values. Errors
E in the observationss at a stationn can be easi ly detected iif there is a cconcurrent reecord
from a reecording rainn gauge at th he same or a nearby statioon.

Fig. 3.1 Standard no

on-recording rain gauge.

Three
T types of
o recordingg rain gauges are in genneral use: weighing typee, float typee, and
tipping bucket
b type. In weighing type instru uments, all the precipittation fallingg is continuously
recorded and it can measure
m all ty
ypes of preccipitation (liqquid or solidd).

3.2.2 Autographic
A c rain gaugee
Natural syphon
s rain gauge is frrequently useed to measuure short-perriod rainfalll. It consistss of a
circular collector
c funnnel and rain nfall recordiing mechaniism. The coollector has a gun metall rim,
2
200 mm diameter (314 cm area)) and is fixed d at 750 mm m above grouund level. Thhe funnel leaads to
a chambeer in which a float is lo ocated. Thiss float rises when rain w water enterss the chamber. A
syphon chamber
c is atttached to th
he float cham
mber. After 110 mm of raiin has fallen, the float risses to
a certainn level and thet syphon action is in nitiated. A ppen is attachhed to the ffloat and records
rainfall on
o a chart placed on a drrum in the formf of rise of the float. After waterr is siphonedd out,
the float returns to th
he original po osition.
 The
T drum mo oves with the help of a mechanical
m cclock such tthat it complletes one rottation
in 24 houurs. The horiizontal axis of the chart is marked w with hours annd the vertical axis repreesents
depth of rainfall. Thee chart is chaanged every day at the sset time and the observerr reads the hhourly
rainfalls from the ch hart and reco ords in data sheets. If noo rain has faallen on a giiven day, thee pen
will tracees a horizonttal line on th
he chart reprresenting zerro rainfall. DDuring the peeriods of raiinfall,
the pen produces
p a slloping line; higher is thee intensity of rainfall, steeeper will be the slope oof the
curve pro oduced by th he pen. Rainn water drain ned by the siiphon actionn may be colllected in a vvessel
and meassured when the t chart is changed
c and
d this will be a check for the total rainnfall.
Fig. 3.2 show ws a self recoording rain gauge assembbly.

Fig. 3.2 A self recorrding rain gaauge with chart mountedd on a drum ((top) and raiinfall recordeed by
the instru
ument on a chart.
c

Measurement
M t errors can arise due too instrumentaal or observver faults. Thhe funnel maay be
(partially
y) blocked soo that waterr enters in th
he float cham mber at a raate different from the raainfall
rate, lead
dings to erroors. If the flloat is wrongly adjustedd, syphon acction is initiiated at a raainfall
depth diffferent than 10 mm. In case of very intense
i rainffall, the rise of float andd siphon mayy take
place so frequently thatt individu
ual pen tracces cannot bbe distinguisshed. Furtheer, the clockk may
malfuncttion – it may y stop, or maay be either slow or fastt and the tim mings will bee incorrect. IIf the
float getss sticky with h the walls of the cham mber, rainfalll will not bbe recorded correctly. OOn his
part, the observer maay incorrectly y read the daata from the chart.
Commonly
C an n ORG is allso installed at the site oof recording station and the daily raainfall
recorded at the two stations maay be differrent due to different exxposure connditions, diffferent
levels off the rims, ettc. Traditionaally, the dailly ORG dataa is considerred to be moore correct wwhere
there is a discrepancy y between thhe two.

3.2.3 Tipping
T Bucket Rain Ga auge
A tipping g bucket rainn gauge con nsists of a circular collecctor funnel w
with a brass or gunmetaal rim
that direccts the rain into
i a pair of tipping buckets whichh sit on a kniife edge (Figg. 3.3). Afterr rain
water hass filled one side of the bucket
b with a small amoount (say, 0.0025cm) of raain, the asseembly
tips. An electrical pu ulse is generrated on eacch tilt and iss recorded tto provide ddata of rain ffallen
with timee. At the saame time, raain water beegins to fill tthe second sside of buckket, and so oon. A
data loggger records thet occurren nce of each tip along w with the timee of tip and the data maay be
download ded as per convenience
c and need. Since
S manuall input is noot used in daata observation or
recordingg, there are little
l chances of errors. The
T instrum ment can be eeasily calibraated and cheecked
in field by
b pouring a known amount of waterr in the colleector funnel.

Fig. 3.3 Tipping buccket recordin

ng rain gaugee.

Measurement
M t errors may occur when
n the funnel iis (partially blocked) so that water eenters
in the tip
pping bucketts at a rates which are different
d thaan the rate oof rainfall. IIf the bucketts are
damaged d or out of balance, theey may tip after non-sstandard rainnfall or the tipping maay be
incorrecttly recorded. A faculty reed switch may
m fail to rregister tips or may doubble register tthem.
The gaug d to some fault or lighhtning strike etc. Differennces may arise in
ge electroniccs may fail due
the rainfaall recorded at the daily rain
r gauge (ORG) due too of differennt exposure cconditions.
3.2.4 Precipitation Measurement by Weather Radar
Although rain gauges are in use for a very long time, they suffer from a limitation: a rain gauge
gives the point value of rainfall but in almost all hydrologic studies, spatially averaged values are
required. Weather radars have been developed to overcome this limitation.

The word radar stands for radio detection and ranging. Weather radar sends
electromagnetic waves in all directions. When these waves collide with raindrops, an echo of the
waves is produced which is caught back by the radar. The strength of echo or returning signal
depends on the intensity of the falling rain and the time taken by the echo to reach the radar
depends upon the distance of the rainfall field from the radar.

A radar is calibrated using the data of rain gauges and relationship between the radar
reflectivity Z (mm6/m3 or the sixth power of the diameter of raindrops in mm per cubic meter of
the atmosphere) and rainfall intensity R (mm/h), known as Z-R relationship is established. A
commonly used empirical equation is:

Z = 200 R1.6 (3.1)

The main advantage of the weather radar is that it can give the estimates of rainfall over
wide areas along with location and movement of storms. Weather radars can also measure
localized storms which may be missed by rain gauges. The range of radar depends on the
hardware; radars with range up to hundreds of km are common these days. The factors that
control the measurement of precipitation are its type, size of raindrops, width of the radar beam,
refraction of beam and atmospheric attenuation. Combined use of rain gauges and radars
provides much more improved rainfall estimates than what any single of them could provide.

A radar can measure precipitation over areas that are difficult to access and gives areal
coverage of rainfall distribution, it is possible to detect the movement of rainfall and most
systems are capable of automatic data processing. The spatial distribution of rainfall is
simultaneously displayed on a monitor. Due to these reasons, the use of weather radars in
precipitation measurement and warning is increasing. These have been found to be particularly
useful in flood forecasting and warning. On the negative side, the measurements from weather
radar are affected by echoes from the ground or ground clutter. In some cases, these radars have
been found to underestimate light rainfall and overestimate heavy rainfall.

Satellites are also being used to estimate precipitation over large areas and in near real-
time. Images from geostationary and polar orbiting satellites along with cloud top temperature,
shape, texture, and cloud history are used for estimation of precipitation. Another possibility is to
combine satellite images with radar data to obtain improved estimates.
3.3 Measurements Related to Snow
Many big rivers of the world are snowfed and their catchments receive a part of precipitation in
solid form or snow. Three variables related to snow are important in water resources studies:
snow depth, area of snow cover, and snow water equivalent. Such data are used to estimate water
content of the snowpack and are of immense use in planning of irrigation and hydropower
projects, reservoir regulation, disaster management, etc.

While measuring snowfall, care should be taken to measure only fresh snow. For this
purpose, a suitable patch on the ground is cleared beforehand or the top of the existing snow
surface is covered with some material (wood, tarpaulin, etc.) so that old and fresh snow can be
demarcated and the depth to this surface is measured. Note that if there is a layer of old snow, it
would not be right to compute the depth of a fresh snowfall from the difference between two
consecutive measurements of the total depth of (fresh and old) snow since old snow keeps on
settling continuously.

A snow pillow is a flat circular container that is filled with non-freezing liquid and is
used to measure the water equivalent of snow. The weight of the snow on the pillow can be
measured by hydrostatic techniques or pressure transducer. The depth of snow may also be
measured in a standard container of uniform cross-section after levelling the snow (without
compressing it). Snow pillows of various dimensions and materials are used to measure the
weight of snow that accumulates. The most common pillows are flat circular container of 3.7 m
diameter made of rubberized material filled with a non-freezing liquid. The pillow is installed on
the surface of the ground, flush with the ground, or buried under a thin layer of soil or sand. To
prevent damage to the equipment and to preserve the snow cover in its natural condition, it is
recommended that the site be fenced. Under normal conditions, snow pillows can be used for 10
years or more.
Hydrostatic pressure inside the pillow is a measure of the weight of the snow on the
pillow. Hydrostatic pressure can be measured by means of a float operated water-level recorder
or a pressure transducer. Data from snow pillows are most reliable when the snow cover does not
contain ice layers, which can cause bridging above the pillows.

In regions receiving heavy snow, commonly snow stakes are installed at different
(accessible) representative locations and are observed to find the depth of fallen snow. Snow
stakes should be installed such that they can be inspected easily from a distance. It needs to be
ensured that the site is the representative of the area and is free from human disturbance. Snow
stakes should be painted in white to minimize increase in melt of surrounding snow. The stakes
are clearly graduated in metres and centimetres and in inaccessible areas, these are provided with
crossbars so that the reading can be taken from a distance.

The water equivalent of snow is the depth of water that would be obtained when the snow
is melted. It can be determined by melting the snow collected in a gauge and measuring the melt
water in the same way as rainfall is measured. It is better to take several representative samples.
Alternatively, cylindrical samples of fresh snow are collected with a suitable snow sampler and
are either weighed or melted. Weighing-type recording gauge are used to determine the water
content of snowfall. During periods of snowfall, funnels of the gauges should be removed so that
the precipitation can fall directly into the receiver. Isotope techniques are also used to estimate
snow water equivalent. However, the instruments used are expensive and adequate safety
measures need to be observed.

In important catchments which receive considerable snow fall, snow surveys are made on
permanent snow courses. A snow course is a permanently marked line where snow surveys are
made. The snow courses should be carefully selected and should be representative of the
catchment area. These should be located in accessible areas where snow falls to the ground
without being intercepted by vegetation, and the site should be protected from strong winds.
During surveys, snow samples are taken and depths are measured at various places at a number
of points along the course.

3.3.1 Measurement of Snow Cover and Properties by Remote-Sensing

Snow related variables (area, water equivalent, etc.) can be measured by the use of remote-
sensing. Data from various bands of electro-magnetic (EM) spectrum such as: gamma rays,
visible and near IR, thermal IR and microwaves, can be used for this purpose. Among these
bands, the microwave band is the most suitable for snow related applications, followed by visible
and near infrared bands. Many algorithms and models have been developed to determine snow
covered area, snow depth, and snow water equivalent using satellite data. Remote sensing data
has many advantages: repeated coverage of the area, ability to get information from remote and
inaccessible areas, and cost-effectiveness.

Microwave portion of the spectrum includes wavelengths in the range of 1 mm to 1 m.

Microwave remote sensing has additional advantages that (depending upon the wavelength) it
can see through clouds, haze, light snow, and smoke. Active microwave sensors are those that
supply their own sources of energy or illumination. Passive sensors, on the other hand, are those
that sense the naturally available microwave energy within their own field of view. The potential
of both active and passive microwave region is similar. An advantage of active microwave over
passive microwave is considerably higher resolution. However, the analysis of active microwave
data is more complex than passive microwave.
Good relationships have been established between snow parameters and properties and
microwave emission. Remote-sensing data from satellites such as Indian Remote Sensing
Satellites (IRS), Landsat, Satellites pour l’observation de la terre (SPOT), etc. are currently being
used operationally to identify snow cover and in snowmelt runoff studies.
3.4 Processing of Rainfall Data
Before the precipitation data is stored (on computer media) for further processing, it is necessary
to carry out preliminary checks and scrutiny. Raw precipitation data commonly has gaps and
inconsistent values. Hence, it is essential to process the data and carry out validation before it is
used in further analysis. Processing of rainfall data consists of a series of steps and procedures.
The various steps and methodology involved in processing of precipitation data are briefly
described.

Processing of data is carried out with two major objectives: to examine the data for errors
and remove them, and to organize the data in a form that is appropriate for subsequent analysis.
The rainfall data may have following errors:
 Entries on the wrong day - shifted entries,
 Entries made as accumulations,
 Missed entries, and
 Rainfall measurement missed on days of low rainfall.

These errors are also termed as “improper registering of data”.

3.4.1 Internal consistency check

As the first step, the data should be checked for the reasonableness based on past experience and
statistics of the station/ region where the station is located. Some of the statistical values used to
check rainfall data includes the values of normal rainfall, highest observed rainfall, or rainfall
corresponding to 25, 50 or 100 year return period.

Example 3.1: The daily precipitation reported from a station is 358.6 mm and the statistics of
the reporting station are:

i. Normal monthly rainfall of the corresponding month : 350.0 mm

ii. Mean maximum 1-day rainfall ( x ) : 210.6 mm
iii. Standard deviation () of maximum 1-day rainfall : 51.0 mm
iv. Highest observed 1-day rainfall : 285.3 mm
v. Value of 1-day max. rainfall of 100-year return period : 300.0 mm
vi. Value of 1-day probable maximum precipitation : 370.8 mm

The reported daily rainfall value (358.6 mm) is more than the normal monthly rainfall of
the corresponding month and is, therefore, doubtful. This value is more than the mean maximum
1-day ( x ) and ( x +) which are 261.6 mm and 261.6 mm, respectively. The reported daily value
is compared with 1-day Probable Maximum Precipitation (PMP) value. Since it is less than the
PMP (370.8 mm), it is considered possible. At this stage, it is not advisable to reject this value. It
should be flagged and further checked by spatial consistency.
 The internal consistency or self consistency checks are applied by using statistical
information based on historical data of the station and current data in case of short duration
rainfall. An example of checking the data by the internal consistency is given here.

Example 3.2: Hourly rainfall data reported at a station are as follows:

Hours 1 2 3 4 5 6
Rainfall (mm) 8.0 10.8 85.8 28.5 19.8 15.0

The hourly rainfall reported during 3rd hour is suspected. Check its reasonability.

Solution: Although the hourly rainfall reported during 3rd hour is suspected, it could not be ruled
out. We examine the total rainfall for 1-3 hours to check the value in the 3rd hour. The 3-hourly
total rainfall was reported as 54.1 mm and this indicates that the rainfall in the 3rd hour could be
35.3 mm (assuming that the rainfalls in the 1st and 2nd hours are correct). When the 6 hour total
rainfall is reported as 117.4 mm, the value of 35.3 mm is confirmed for the 3rd hour. Further
checking for the erroneous value is carried out similarly.
This example shows that it is a good practice to report the hourly values as well as the
sum of block of hours so that possible errors in conveying the data can be detected and corrected.

3.4.2 Scrutiny of Precipitation Data by Multiple Time Series Graphs

Validation of precipitation data can be carried out by plotting time series of data of multiple
stations on the same graph. Such graphs can be drawn for hourly, daily, monthly and yearly
rainfall data. The validation of compiled monthly and yearly rainfall totals helps in bringing out
the inconsistencies that are either due to a few very large errors or due to small systematic errors
which persist unnoticed for much longer durations. A comparison of daily rainfall at multiple
stations is shown in Figure 3.4. This type of variation at the two stations appears to be normal
given the spatial structure of rainfall.
Figure 3..4: Comparisson of daily rainfall at multiple
m statioons.

Scrutiny by tabulatioons of daily rainfall

r seriees of multiplee stations
To scrutiinize the datta, rainfall data
d of vario
ous rain gauuge stations are tabulated side by siide in
different columns in a table or sp preadsheet. A careful exxamination oof the table hhelps in reveealing
any anommalies that may
m be preseent in the datta and whichh may be diffficult to seee in multiplee time
series graaphs in somee cases.

Checking g against datta limits for totals at lon

nger durationns
Many sy ystematic errors are so small that individuallyy they cannnot be easilyy noticed. IIf the
observedd data are addded for long ger time duraations then tthe accumulaated errors bbecome large and
the resullting time series should again bee checked against corrresponding expected liimits.
Therefore, the daily rainfall datta of each station shoulld be aggreggated to moonthly and yyearly
values annd then checcked against the maximu um monthly and yearly ttotals. If therre is a signifficant
differencce, the data are
a cross-cheecked and veerified with the availablle records off the same sttation
(previouss times) and at nearby sttations.

Test for means

m (t-Tesst)
The most common parametric
p teest used to check whetheer or not twoo samples are from the same
populatioon is the t-- test. The main assum mptions of this test arre: (i) the observations are
independdent, (ii) thee observations are draw wn from norrmally distriibuted popuulations, andd (iii)
these poppulations haave the samme variance. Hence this test is usefful to determ mine whetheer the
mean off the samplees are signifficantly diffeerent from eeach other and whetherr both the sseries
belong too the same population
p or
o not. Acco ording to thiis test, the ‘‘t’ statistic oof the samples is
determinned by:
 1 1
t  X1  X 2 S  (3.2)
n1 n2

where, X 1 and X 2 are the arithmetic mean of the two samples of size n1 and n2 respectively; S
is the unknown population standard deviation estimated from the samples variances s1 and s2 as:

(n1  1) s1  (n 2  1) s 2
S
n1  n 2  2 (3.3)

If the value of the statistic t is less than the tabulated value of Student’s distribution at some
chosen significance level α and (n1+n2-1) degrees of freedom then the hypothesis that “the means
of both the samples are not significantly different” may be accepted at the chosen significance
level.

Test for variances (F-test)

F-test is commonly used for testing whether or not the variances of two samples are significantly
different. According to this test, the F statistic of the samples is determined as:

F  s12 s22 (3.4)

If the computed F is less than the tabulated value of F distribution at some chosen significance
level α, and n1-1 and n2-1 degrees of freedom then the hypothesis that “the variances of both the
samples are not significantly different” may be accepted at the chosen significance level.

Statistical tests such as t-test and F-tests are discussed in Module 10 in detail.

3.4.3 Correction and completion of rainfall data

After primary and secondary validation of rainfall data, various errors and missing values in data
series can be identified. Such incorrect and missing values can be replaced wherever possible by
estimated values based on other observations at the same station or at neighbouring stations.
Procedure adopted for correction and completion of data depend on type of error
observed/indicated and the availability of suitable data. Some typical cases are discussed here.

Rain gauge station failure: Depending on the nature and type of failure of rain gauge, data of
self recording rain gauge and/or the ordinary rain gauge can be utilised to correct and complete
the missing data.
Accumulated rainfall data: If the data of ordinary rain gauge is read after several days, the
recorded total represents accumulated rainfall over a period of several days. Such accumulated
totals can be distributed over the period of accumulation based on the rainfall recorded at the
neighbouring stations over the same period.
Long-term shift in rainfall data: Consistency of a rain gauge station data is examined to
determine if the conditions relevant to recording rainfall at a station have undergone a significant
change during the period of record. In case there is significant change inconsistency would arise
in the rainfall data of that station. This inconsistency would be noticed from the time the
significant changes have taken place. Some of the common causes for inconsistency of record
are:

i) Shifting of a rain gauge station to a new location,

ii) The surroundings of the station have undergone a marked change so that the exposure
conditions are changed,
iii) Occurrence of observational error from a certain date.

The double mass curve technique is used to check the consistency of a rainfall series.
This technique is based on the principle that the data are consistent when each recorded data
comes from the same parent population.

Double mass analysis is a graphical method to identify and adjust inconsistencies in a

station's data by comparing the trend of data of reference stations. In a double mass curve, both
axes are accumulated precipitation values (see Fig. 3.5). Usually the accumulated seasonal or
annual precipitation values of reference station(s) are taken as abscissa and those of the station
under test as ordinate. A change in the regime of the rain gauge, such as change in exposure and
the change in location is revealed by the change in the slope of the straight line. The older
records are adjusted by multiplying the precipitation values by the ratio of the slope of the later
period to the slope of the earlier period.
slope of later period
Adjusted precipitation  Raw precipitation * (3.5)
slope of earlier period

To carry out the test, a group of stations in neighborhood of the index station (for which
the consistency test is required) is selected. The data of the annual rainfall of the index station
(X) and also the average rainfall of the group of base stations covering a long period is arranged
in the reverse chronological order, i.e. the latest record as the first entry and the oldest record as
the last entry in the list. The accumulated precipitation of the station X and the accumulated
values of the average of the group of base stations are calculated starting from the latest record.
Individual cumulative rainfall values of the index station are plotted against mean cumulative
values for various consecutive time periods. A break in the slope of the resulting plot indicates a
change in the precipitation regime of the index station. The precipitation values at the index
station prior to the change of regime are corrected by using the initial slope of the line as follows:
 PX (i )  m  C A (i )  C X (i  1) (3.6)

where PX(i) is the corrected rainfall at the index station X during year i; m is the initial slope of
the line; CA(i) is the cumulative mean rainfall up to the year i; CX(i-1) is the cumulative rainfall
at the index station X till the year i-1.
As example, consider the double mass plots of Bhopal and Basoda (Figures 3.5 and 3.6)
which have been plotted using the annual rainfall data of eight other neighbouring stations. It can
be observed from Fig. 3.5 that the change in slope of the mass curve for Bhopal is insignificant,
whereas, the same is significant in the case of Basoda (Fig. 3.6). Hence, corrections are needed
for Basoda data.

80000

70000

60000
Cum. at Bhopal, mm

50000

40000

30000

20000

10000

0
0 10000 20000 30000 40000 50000 60000 70000 80000
Cum. areal average, mm

Figure 3.5: Double mass curve for Bhopal. X-axis is the accumulated mean precipitation at group
of stations and on the y-axis is accumulated precipitation at Bhopal.
 100000
Original Corrected
Cum. at Basoda, mm

80000

60000

40000

20000

0
0 20000 40000 60000 80000 100000
Cum. Average, mm

Fig. 3.6 Double mass curve for Basoda. X-axis has the accumulated mean precipitation at group
of stations and on the y-axis is accumulated precipitation at Basoda. Both original and corrected
curves have been plotted.
3.4.4 Spatial Consistency Check
Rainfall data exhibit some spatial consistency and this forms the basis of investigating the
observed rainfall values. An estimate of the interpolated rainfall value at a station is obtained on
the basis of the weighted average of rainfall observed at the surrounding stations. If the
difference between the observed and the estimated values exceed the expected limiting value,
such values are considered as suspect and are flagged for further investigation and ascertaining
the possible causes of departures.

Spatial consistency checks for rainfall data are carried out by relating the observations
from surrounding stations for the same duration with the rainfall observed at the station. This is
achieved by interpolating the rainfall at the station under question with rainfall data of
neighboring stations. The station being considered is called the test station. The interpolated
value is estimated by computing the weighted average of the rainfall observed at neighboring
stations. Ideally, the stations selected as neighbors should be physically representative of the area
in which the station under scrutiny is situated. The following criteria are used to select the
neighboring stations:

(a) The distance between the test and the neighboring station must be less than a specified
maximum correlation distance;
(b) too many neighboring stations should not be considered for interpolation; and
(c) to reduce the spatial bias in selection, it is advisable to consider an equal number of
stations in each quadrant.

Example 3.3: Rainfall reported at a group of five stations (see Fig. 3.7) is as follows.

Chando Kundla

Virpur Idar

Sabli

Fig. 3.7 Location of stations for spatial consistency check.

Station Kundla Idar Virpur Chandop Sabli

Rainfall (mm) 132.1 12.1 103.3 125.7 149.8

During the quality control process, the data at Idar is identified as doubtful. Check this
data for spatial consistency.
Solution: The rainfall at Idar is estimated using the distance power method and compared with
the observed value. From the four quadrants around Idar (Fig. 3.6), the station nearest from each
quadrant is selected for estimation of rainfall at Idar. Using the reference coordinate system, the
distance of each of the estimator stations from Idar is determined and the rainfall at Idar is
estimated.

S. N. Station Distance from Idar Di 1/Di2 Ri/Di2

(km)
1. Kundla 42 5.67*10-4 0.075
2. Virpur 39 6.57*10-4 0.068
3. Sabli 75 1.78*10-4 0.027
Total 14.02*10-4 0.170

Rainfall at Idar = [(Ri/Di2)] / [(1/Di2 )] = 0.17/14.02*10-4 = 121.25 mm.

Since the observed value is very much different from the estimated value, it is rejected
and replaced by the estimated value. Note that there is a possibility that the decimal point was
wrongly placed while recording the data at Idar.

3.5 Spatial Averaging of Rainfall Data

Precipitation observations from gauges are point measurements. However, in the hydrological
analysis and design, we frequently require mean areal precipitation over an area. A characteristic
of the precipitation process is that it exhibits appreciable spatial variation though the values at
relatively short distances may have good correlation.

Numerous methods of computing areal rainfall from point measurements have been
developed. While using precipitation data, one often comes across missing data situations. Data
for the period of missing rainfall could be filled using various techniques. Due to the spatial
structure of precipitation data, some type of interpolation making use of the data of nearby
stations is commonly adopted.
Let the precipitation data be available at n stations, spread over an area and Pi be the
observed depth of precipitation at the ith station. Using a linear interpolation technique, an
estimate of precipitation over the area can be expressed by
n
P * =  Pi W i (3.7)
i=1

where Wi is the weight of the ith station. The spatial averaging techniques differ in the method of
evaluation of these weights. Weights of an optimal interpolation technique are decided such that
the variance of error in estimation is the minimum.
The most commonly used methods are for Spatial Averaging of Precipitation Data:

(a) Arithmetic average,

(b) Normal ratio method,
(c) Distance power method,
(d) Thiessen polygon method, and
(e) Isohyetal method.

The choice of the method is depends on the quality and nature of data, importance of use and
required precision, availability of time and computer. Some of the commonly used methods are
described below.

3.5.1 Arithmetic Average

The simplest technique to compute the average precipitation depth over a catchment area is to
take an arithmetic average of the observed precipitation depths at gauges within the catchment
area for the time period of concern. The average precipitation is:
n

P W
i=1
i i

P= (3.8)
n

where P is the average catchment precipitation from the data of n stations, Pi is the precipitation
at station i, and Wi is the weight of ith station. If the gauges are relatively uniformly distributed
over the catchment and the rainfall values do not have a wide variation, this technique yields
good results.

3.5.2 Thiessen Polygon

The Thiessen Polygon method is based on the concept of proximal mapping. Weights are
assigned to each station according to the catchment area which is closer to that station than to
any other station. This area is found by drawing perpendicular bisectors of the lines joining the
nearby stations so that the polygons are formed around each station (Fig. 3.8). It is assumed that
these polygons are the boundaries of the catchment area which is represented by the station lying
inside the polygon. The area represented by each station is measured and is expressed as a
percentage of the total area. The weighted average precipitation for the basin is computed by
multiplying the precipitation received at each station by its weight and summing. The weighted
average precipitation is given by:
n
P =  Pi W i (3.9)
i=1

in which Wi = Ai/A, where Ai is the area represented by the station i and A is the total catchment
area. Clearly, the weights will sum to unity.
 An advantage of this method is that the data of stations outside the catchment may also be
used if these are believed to help in capturing the variation of rainfall in the catchment. The
method works well with non-uniform spacing of stations.

A major drawback of this method is the assumption that precipitation between two
stations varies linearly and the method does not make allowance for variation due to orography.
In this method, the precipitation depth changes abruptly at the boundary of polygons. Also,
whenever a set of stations are added to or removed from the network, a new set of polygons have
to be drawn. The method fails to give any idea as to the accuracy of the results. If a few
observations are missing, it may be more convenient to estimate the missing data than to
construct the new set of polygons.

Example 3.4: For a catchment, the rainfall data at six stations for July month along with their
weights are as given in Table 3.4. Find the weighted average rainfall for the catchment by using
the Thiessen polygon method.

Solution: Using the observed rainfall and station weight, weighted rainfall at each station is
computed. Summation of these values gives the weighted average rainfall for the catchment. The
computations are shown in Table 3.1.

Table 3.1 Estimation of the mean areal rainfall by the Thiessen polygon method.

S. N. Station Station Rainfall Weighted rainfall

Name weight (mm) (mm)
1. Sohela 0.06 262.0 15.7
2. Bijepur 0.12 521.0 62.5
3. Padampur 0.42 177.0 74.3
4. Paikmal 0.28 338.0 94.6
5. Binka 0.04 158.0 16.1
6. Bolangir 0.08 401.6 12.6
Weighted catchment rainfall 275.8
 Thiessen Polygon plot for Betwa up to Basoda (SEP-1995)
Sironj Sl. Station Name Weight
1 Basoda 0.0849
2 Berasia 0.1678
3 Bhopal 0.2573
4 Gairatganj 0.0578
5 Raisen 0.1685
Basoda 6 Sehore 0.0220
7 Sironj 0.0804
8 Vidisha 0.1611

Berasia

Vidisha

Gairatganj
Raisen
Bhopal
Sehore

Avg. rainfall: 170.801 (estimated using a combination of 8 raingauges)

Fig. 3.8 The Thiessen polygon method for computing the mean areal rainfall.

3.5.3 Isohyetal Method

The isohyetal method employs the area encompassed between isohyetal lines. Rainfall values
are plotted at their respective stations on a suitable base map and contours of equal rainfall,
called isohyets, are drawn. In regions of little or no physiographic influence, drawing of
isohyetal contours is relatively simple matter of interpolation. The isohyetal contours may be
drawn take into account the spacing of stations, the quality, and variability of the data. In regions
of pronounced orography where precipitation is influenced by topography, the analyst should
take into consideration the orographic effects, storm orientation etc. to adjust or interpolate
between station values.

Computers are being used to draw isohyetal maps these days, by using special software.
As an example, the isohyetal map for an area is shown in Fig. 3.8. The total depth of
precipitation is computed by measuring the area between successive isohyets, multiplying this
area by the average rainfall of the two ishohyets, and totaling. The average depth of precipitation
is obtained by dividing this sum by the total area. The average depth of precipitation (Pi) over
this area is obtained by:
n

P
i=1
i Ai
P= n
(3.10)
A
i=1
i

where Ai is the area between successive isohyets and Pi is the average rainfall between the two
isohyets.
Fig. 3.8 The
T isohyetaal method for computing
g the mean arreal rainfall..

Table. 3.2 Estimation

n of mean arreal catchment rainfall byy Isohyetal M
Method

Iso
ohyetal rangee Averrage value Areaa Volume
(mm) (mm)
( (km2 ) (105m3)

52
21.0 - 500.0 510.5
5 70 357.4
50
00.0 – 300.0 400.0
4 530 2120.0
33
38.0 – 300.0 319.0
3 100 319.0
20
00.0 – 300.0 250.0
2 20800 5200.0
15
58.0 – 200.0 179.0
1 28200 5047.8
56000 13044.2
13044.2
Averaage catchmen
nt rainfall =  2232.9mm
5600

Examplee 3.5: Using the point raainfall data for

fo a catchmeent, isohyetaal lines weree drawn as shhown
in Fig. 3..8. The area enclosed by
y each isohyeet was calcuulated as giveen in Table 3.3. Computte the
average catchment
c raainfall.

Solution
n: For each issohyet, the average
a value is worked out (the maxximum obseerved rainfalll was
108 cm and
a the minimum 38 cm m). This, multiplied by thhe area encloosed by that isohyet givees the
volume of
o rainfall foor that isohy
yet. Now thhe volumes ffor differentt isohyetals are summedd and
divided by the area of the catchment to get average catchment rainfall. The computations are
shown in Table 3.3.

Table 3.3 Estimation of mean areal rainfall by the isohyetal method.

Isohyet Average Area Net area Rainfall

value (cm) value (cm) enclosed (km2) volume (km2-
(km2) cm)
105 106.5 0.79 0.79 84.14
100 102.5 1.52 0.73 74.83
90 95 2.57 1.05 99.75
80 85 3.47 0.90 76.50
70 75 4.50 1.03 77.25
60 65 5.18 0.68 44.20
50 55 5.39 0.21 2.20
< 40 39 5.41 0.02 0.78
Total 5.41 459.65
Average catchment rainfall = 459.65/5.41= 84.96 cm
3.6 ESTIMATION OF MISSING DATA
Data for the period of missing rainfall data could be filled using estimation technique. The length
of period up to which the data could be filled is dependent on individual judgment. Generally,
rainfall for the missing period is estimated either by using the normal ratio method or the
distance power method.

3.6.1 Normal Ratio Method

In the normal ratio method, the rainfall PA at station A is estimated as a function of the normal
monthly or annual rainfall of the station under question and those of the neighboring stations for
the period of missing data at the station under question.
n

 NR
NR A
* Pi
i=1 i (3.11)
PA=
n

where Pi is the rainfall at surrounding stations, NRA is the normal monthly or seasonal rainfall at
station A, NRi is the normal monthly or seasonal rainfall at station i, and n is the number of
surrounding stations whose data are used for estimation.

Example 3.6: A catchment has four rain gauge stations A, B, C & D. Normal monthly rainfall at
these stations is known. The observed rainfall at the stations B, C & D for a storm event is
known and is given in the following table. Find the missing rainfall at station A.

Solution: The ratio of normal rainfall at station A to normal rainfall at station i or NRA/NRi has
been calculated and is given in table below.

Station  A B C D
Normal Rainfall (mm) 331.3 290.8 325.9 360.5
Event Rainfall (mm) ? 98.9 120.5 110.0
NRA/NRi 1 1.14 1.02 0.92

The estimated rainfall at station A is:

1.14 * 98.9 + 1.02 * 120.5 + 0.92 * 110.0
PA = = 112.3mm
3

3.6.2 Distance power method

The rainfall at a station is estimated as a weighted average of the observed rainfall at the
neighboring stations. The weights are equal to the reciprocal of the distance or some power of
the reciprocal of the distance of the estimator stations from the estimated stations. Let Di be the
distance of the estimator station from the estimated station. If the weights are an inverse square
of distance, the estimated rainfall at station A is:
 n

P /D
i=1
i
2
i
(3.12)
PA = n

 1/ D
i=1
2
i

Note that the weights go on reducing with distance and approach zero at large distances. A major
shortcoming of this method is that the orographic features and spatial distribution of the variables
are not considered. The extra information, if stations are close to each other, is not properly used.
The procedure for estimating the rainfall data by this technique is illustrated through an example.
If A, B, C, D are the location of stations discussed in the example of the normal ratio method, the
distance of each estimator station (B, C, and D) from station (A) whose data is to be estimated is
computed with the help of the coordinates using the formula:

Di2 = [ (x - xi)2 + (y - yi)2] (3.13)

where x and y are the coordinates of the station whose data is estimated and xi and yi are the co-
ordinates of stations whose data are used in estimation.

Example 3.7: Using the data of Example 3.6, estimate rainfall at station A using the distance
power method.

Solution: Since the coordinates of the stations are known, their distances from station A can be
calculated. The weights 1/Di2 are then computed for each station and the rainfall at station A is
estimated as follows:

Distance from 1/Di2 Rainfall Pi Weighted rainfall

Station
station A (mm) Pi*(1/Di2) (mm)
B 28.0 1.29*10-3 98.9 125.6*10-3
C 17.7 3.19*10-3 120.5 384.6*10-3
D 42.5 0.55*10-3 110.0 60.5*10-3
Total 5.01*10-3 570.7*10-3

570.7 10 3
Rainfall at station A =   113.9 mm
5.0110 3

3.7 Disaggregation of Rainfall Data

Many hydrological applications, such as flood forecasting, require rainfall data of shorter
duration, whereas the network of recording rain gauges (providing short duration data) is small
in comparison to that of daily rain gauges. Hence, it is often necessary to disaggregate (or divide)
the daily rainfall data into shorter time intervals (usually one hour). The observed daily rainfall is
distributed in hourly values to follow the same pattern as the observed hourly rainfall. The
information of short interval rainfall is used together with the information of daily rainfall from
nearby non-recording (daily) gauges. A common method to do this is the mass curve method.

A mass curve is a graphical plot of accumulated rainfall at a station versus time. Mass
curves of accumulated rainfall at (non-recording) daily stations and recording stations can be
prepared by plotting the accumulated rainfall values against time for the storm duration under
analysis. Now the mass curves of the recording rain gauge stations are compared with those of
the non-recording stations to decide which recording rain gauge is representative of which of the
non-recording rain gauge for the purpose of distributing daily rainfall into hourly rainfall.

Assume that the daily rainfall is observed at 0800 hours. For converting the daily rainfall
into hourly rainfall, the hourly rainfall from 0800 hr to 0800 hr for consecutive days is
accumulated and the rainfall during each hour is expressed as a ratio of the total rainfall during
24 hours (0800 to 0800). These ratios for the identified station are used to distribute the daily
rainfall for the corresponding duration at the non-recording rain gauge stations.
The procedure for distribution of daily rainfall at non-recoding rain gauge stations into
hourly rainfall is explained with the help of an example.

Example 3.8: Daily rainfall data (mm) of four stations for the period 28th August to 30th August
1973 is given below:
Station name 28/08/1973 29/08/1973 30/08/1973
Pendra Road 65.3 80.8 1.1
Dindori 23.2 58.4 1.0
Ghansore 42.0 98.0 20.6
Niwas 30.4 203.0 33.4

Hourly rainfall data of two self-recording rain gauge (SRRG) stations (Mandla and Jabalpur) is
given in Table 3.5 for the period 27-29 August 1973. Disaggregate the daily rainfall data into
hourly values.

300 JABALPUR
Cumulative Rainfall

200

100
MANDLA
(mm)

0
27.8.73 28.8.73 29.8.73
 Fig. 3.9 Mass curve of hourly rainfall.

300
NIWAS

GHANSOR
200
PENDRA ROAD
Rainfall (mm)
Cumulative

DINDORI
100

0
27.8.73 28.8.73 29.8.73 30.8.73

Fig. 3.10 Mass curve of daily rainfall.

Solution: The hourly rainfall data at each of the two SRRG stations is plotted on a graph to
prepare the mass curve of hourly rainfall as shown in Figure 3.9. The daily rainfall data at each
of the four stations is also cumulated and plotted on a graph in Figure 3.10. The points are joined
to form the mass curve of daily rainfall. It may be seen that the mass curve of hourly rainfall has
many kinks as this has been prepared using hourly data. The mass curves of daily rainfall are
compared with those of hourly rainfall to determine which of the daily rain-gauge stations are
represented by which of the SRRG stations. Based on the match of curves, it was decided that
the daily rainfall stations at Pendra Road and Dindori are represented by Mandla while Jabalpur
represents Niwas and Ghansore stations.

To convert the daily rainfall into hourly rainfall, the hourly rainfall from 0900 hr on a day
to 0800 hr for consecutive days is cumulated and the rainfall during each hour is expressed as a
ratio of the total rainfall during 24 hours (0900 to 0800). These ratios are used to distribute the
daily rainfall for the corresponding duration at those rain gauge stations, which are represented
by the SRRG. The daily rainfall distributed for 1 day, i.e. corresponding to 28-29 August 1973 is
given in Table 3.6.

Remarks
It is very well recognized that the areal precipitation estimates derived on the basis of point
rainfall observations from a network of sparsely and unevenly distributed precipitation gauges
could only be regarded as an index of rainfall.

Radar sensed echo intensity reflected by precipitation could provide useful areal estimates of
precipitation over areas with few or no precipitation gauges. Precipitation observed by a dense
network of precipitation gauge specially set up for the purpose on a temporary basis could be
related to echo intensities as sensed by radar to develop reasonable relationships for the area of
interest.

Table 3.5: Hourly rainfall data of Mandla and Jabalpur

Hours Mandla Jabalpur

27/08/73 28/08/73 29/08/73 27/08/73 28/08/73 29/08/73
1 0.0 0.0 0.0 0.0 0.0 23.5
2 0.0 0.0 0.0 0.0 0.0 30.5
3 0.0 0.0 0.0 0.0 0.3 10.2
4 0.0 0.0 0.0 0.0 0.5 30.8
5 0.0 0.0 0.0 0.0 0.1 16.9
6 0.0 0.0 0.0 0.0 1.0 1.9
7 3.0 1.7 0.0 0.0 24.4 0.6
8 0.0 9.0 0.0 0.0 0.2 3.1
9 0.0 2.3 0.5 0.0 0.0 5.3
10 0.9 0.0 0.4 0.0 0.0 3.8
11 0.0 1.5 0.1 0.0 23.8 0.8
12 0.0 5.5 0.0 0.0 1.3 1.5
13 0.0 0.0 0.0 0.0 0.2 0.0
14 0.0 0.0 0.0 0.0 0.0 0.3
15 0.0 0.0 0.0 0.0 0.1 0.9
16 0.0 2.5 0.3 0.0 0.6 1.5
17 0.2 1.5 0.4 0.0 29.0 0.4
18 2.1 0.8 0.3 0.0 11.0 0.5
19 0.7 3.4 0.0 25.7 7.1 0.1
20 0.3 0.6 0.0 0.0 0.4 0.5
21 0.1 2.5 0.0 0.1 2.7 0.7
22 0.1 0.0 0.0 1.0 6.8 0.0
23 4.7 0.0 0.0 6.0 20.6 0.0
24 0.0 0.0 0.0 1.3 0.0 0.0

Table 3.6: Daily rainfall distributed into hourly rainfall (28-29 August 1973)

Hour Pendra Road Dindori Niwas Ghansore

28/8/1973 29/8/1973 28/8/1973 29/8/1973 28/8/1973 29/8/1973 28/8/1973 29/8/1973
1 0.0 0.0 21.5 10.5
 2 0.0 0.0 28.0 13.7
3 0.0 0.0 9.3 4.5
4 0.0 0.0 28.2 13.7
5 0.0 0.0 13.4 7.5
6 0.0 0.0 1.7 0.8
7 0.0 0.0 0.5 0.2
8 0.0 0.0 2.8 1.3
9 8.9 6.4 0.0 0.0
10 0.0 0.0 0.0 0.0
11 5.6 4.1 21.9 10.7
12 21.8 13.8 1.0 0.5
13 0.0 0.0 0.2 0.1
14 0.0 0.0 0.0 0.0
15 0.0 0.0 0.1 0.0
16 9.7 7.0 0.5 0.2
17 6.5 4.7 26.6 13.3
18 2.5 1.8 10.1 4.9
19 13.7 9.9 6.5 3.1
20 2.4 1.7 0.4 0.2
21 9.7 7.0 2.4 1.1
22 0.0 0.0 6.1 2.9
23 0.0 0.0 18.9 9.1
24 0.0 0.0 0.0 0.0

3.8 ESTIMATION OF MEAN AREAL PRECIPITATION IN MOUNTAINOUS

AREAS
Precipitation data which exhibits appreciable spatial variation over relatively short distance is
often used as areal estimate for use as input in hydrological models. Several methods are
commonly used for estimating average precipitation over a specific area, such as a drainage
basin. The choice of the method is generally dependent on the quality and nature of data, the
importance of its use and required precision of the result.

Estimation of mean areal precipitation in mountainous areas by the conventional methods does
not yield the desired results because of the influence of orography, aspect, storm orientation, etc
on the precipitation in areas with pronounced orography. A better way of estimating areal rainfall
in mountainous regions is through the use of isopercental technique.

3.9 STORAGE OF PRECIPITATION DATA IN DATA BANK

In India, precipitation data collected by central and state organizations are stored in hard copy as
well as on computers. All the rainfall data collected by India Meteorological Department (IMD)
are transferred to computers by the office of Additional Director General of Meteorology
(Research), IMD, Pune. Users can procure precipitation and other meteorological data from IMD
Pune (www.imdpune.gov.in). In addition, processed precipitation data are also available at some
internet sites.

3.10 Rain Storm Analysis

While designing a dam, it is necessary that the outlet capacity is large enough to safely pass a
flood of certain magnitude. This critical flood is known as the design flood for the structure. The
type of the hydraulic structure is the main criterion to decide the design flood hydrograph. For
this purpose, the structures are classified as:

i) large or medium dams;

ii) medium structures, such as barrages, road and railway bridges; and
iii) small or minor structures, such as cross drainage works and minor irrigation tanks and
minor road bridges.

If long-term runoff data are not available, rainfall data which are generally available for a
longer period are used to estimate the design storm. This design storm is used with a suitable
rainfall-runoff model to obtain a design flood. The rainstorm analysis is the first step in the
design storm estimation procedure. The design storm (rainfall) is a magnitude of rainfall and its
distribution which is developed for the design of specific types of structures. It has three
components, namely, the rainfall amount, the areal distribution of rainfall, and the time
distribution of rainfall.
The Probable Maximum Precipitation (PMP) is defined as “theoretically the greatest
depth of precipitation for a given duration that is physically possible over a given size storm area
at a particular geographical location at a certain time of year”. The Standard Project Storm is
defined as that rain storm which is reasonably capable of occurring in the region of problem
basin. It is generally the most severe rain storm which has occurred in the region of the basin
during the period of available records.

The data on volumes of precipitation during severe storms is important for examining and
studying storms suitable for design purposes. Such information is generally presented in the
form of tables of the maximum average depth of storm precipitation over various standard area
sizes, such as 100 km2, 500 km2 etc. These data are known as the Depth-Area-Duration data and
they could be presented in tabular as well graphical form. WMO (2008) has described
procedures to estimate the probable maximum flood and standard project flood.
Closure
Undoubtedly, precipitation remains the most important and frequently used hydro-
meteorological in studies dealing with water resources. Precipitation is also the variable whose
longest measuring data series are available at a large number of stations. This module gives a
detailed coverage of acquisition and processing of precipitation data.

References
Jain, S.K., and Singh, V.P. (2003). Water Resources Systems Planning and Management.
Elsevier, Amsterdam.
Subramanya, K., Engineering Hydrology, Tata McGraw Hill Education Pvt. Ltd., New Delhi.
WMO(1982). Concepts and techniques in hydrological network design. Operational Hydrology
Report No.19, WMO No. 580, World Meteorological Organization, Geneva.
WMO(1982). Methods of correction for systematic error in point precipitation measurements for
operational use. Operational Hydrology Report No. 21. WMO No. 589. World
Meteorological Organization, Geneva.
WMO(1983). Guide to Meteorological Instruments and Method of Observation. WMO No. 8.
World Meteorological Organization, Geneva.
WMO (2008). Guide to Hydrological Practices. Volume I: Hydrology – From Measurement to
Hydrological Information. WMO No. 168. World Meteorological Organization, Geneva.
 Module 4
Measurement and Processing of Streamflow Data

4.1 Streamflow Data

4.1.1 Measurement of River Flows
4.1.2 Types of Station Control
4.1.3 Main elements of a streamflow gauging station
4.2 Measurement and Recording of River Stage
4.2.1 Non-recording river stage gauges
4.2.2 Depth Sounding and Suspension Equipment
4.2.3 Water Level Recorder
4.3 Direct Determination of Discharge
4.4 Velocity-Area Methods
4.4.1 Velocity Measurement by Floats
4.4.2 Measurement of Flow Velocity by Current Meters
4.4.3 Stage-Discharge Gauging Stations in Natural Channels
4.4.4 Discharge Measurements
4.5 Indirect Determination of Discharge
4.5.1 Estimation of Discharge by Slope-Area Method
4.6 Moving Boat Method
4.6.1 Determination of stream velocity
4.7 Discharge Measurement Using Artificial Structures
4.7.1 Weirs
4.7.1 Flumes
4.7.2 Short-Throated Flumes
4.8 Advanced Discharge Measurement Techniques
4.8.1 Dilution Methods for Measuring Discharge
4.8.2 Electromagnetic Method
4.9 Measurement of Discharge under Difficult Conditions
4.9.1 Special Problems in Streamflow Measurements in Arid and Semi-arid Regions
4.9.2 Measurement of Streamflow under Ice Cover
4.10 Closure
References
4.A International Standard ISO 1100
4.A.1 General requirements of a gauging station

Keywords: Streamflow Data, Station Control, River Stage, Depth Sounding, Water Level
Recorder, Velocity-Area, Float, Current Meter, Slope-Area, Moving Boat Method, Weirs,
Flumes, Dilution, Electromagnetic Method.

The objectives of this module are:

• To discuss the techniques of acquisition of river stage data,
• To discuss the techniques of acquisition of river discharge data,
• To discuss the advanced techniques of acquisition of river discharge data.
A stream is a flow channel into which the surface runoff from its catchment drains. Streamflow
or discharge is measured in units of volume per unit time (m3/s). The measurement of discharge
in a stream forms an important branch of "Hydrometry", the science and practice of water
measurement. This module deals with methods of measuring discharge in a river and processing
of streamflow data.

4.1 Streamflow Data

Stream flow records are the continuous data of flow passing through a particular section on the
stream. Since the measurements made at the gauging site may subject to various random,
systemic and spurious errors, raw data is to be processed to transform them into their most usable
forms through a variety of quality checks at appropriate stages to ensure data quality and
reliability. Processing of stream flow data is not a single step process; several steps are required
to produce reliable streamflow data.

For effective and efficient water resource assessment and management and for proper
river basin planning, development of flood forecasting system, etc. reliable, accurate, processed
and easily accessible data base containing data of hydrologic variables is a pre-requisite.

The most important hydrological data for management of surface water resources
pertains to streamflow. Streamflow has served as the lifeline for mankind and continues to do so.
Its importance is also relatively more, since this source is visible in contrast to ground water
which is hidden. Streamflow records are primarily continuous records of flow passing through a
particular location on the stream. Streamflow data are analysed to determine the magnitude and
variability of surface waters. These records are input in planning, design, and operation of
surface water projects and are also used in design of bridges and culverts, flood forecasting
systems, and flood plain delineation.

4.1.1 Measurement of River Flows

There are several methods for measuring river flows - the long-established and time-tested
velocity-area method, structural (use of weir, notch) and dilution techniques and recent
innovations such as the moving boat method, ultrasonic, and electromagnetic methods. Each of
these methods has certain advantages and limitations which makes their choice dependant on site
conditions and on the equipment and resources available with the gauging authority.

When the channel cross-section is regular and of a known shape, an estimate of the
discharge can be obtained by simply measuring the averaged flow either across the channel at
some depth or over depth in the center of the channel. Fig. 4.1 shows the classification of various
methods of discharge measurement.
 The sections where river measurements are carried out are known as stream gauging
stations. A network of these stations is established to collect data about river flows of a region.
The location of gauging sites primarily depends upon the purpose of data collection. If the site is
needed for a specific project, the general location will be in the vicinity of the project. However,
if a network of gauging stations is to be established to study the general hydrology of a region
and for planning and design of various projects than careful planning is required to identify
locations so that optimum information is obtained for the resources deployed in the data
collection.

River gauging stations are of three types: basic data stations, operational stations, and
special stations. The basic stations are established to collect data for a variety of uses, including
planning and design of projects, and to understand the hydrological characteristics of the region.
The operational stations collect data to run projects and issue forecasts. The objective of special
stations is to meet specific data needs that may arise in cases, such as research, project
investigation, special studies, legal cases, etc. Their operation is terminated when the specific
need is fulfilled. Sometimes, auxiliary stations are set up to augment the database from the
network or region. An auxiliary station may, for example, record only the peak discharge which
occurs at that site during a certain (usually limited) period.

Conventional Advanced
Methods Techniques

Direct Indirect

Velocity area methods Slop area method Moving boat method

using current meter
By stage - discharge Ultrasonic (Acoustic)
Velocity area methods relations
by using floats method
Measurement by Electromagnetic method
By Dilution methods
hydraulic structures

Fig. 4.1 Classification of methods of discharge measurement.

The number of gauging sites depends on the cost of installation and operation, the value
of the data, watershed size, degree of development, objective of data collection, accuracy,
hydrologic characteristics, etc. Some of these factors are interrelated. The streamflow data that
are of immense use in water resources are river water level or gauge and discharge. A continuous
observation of the river water level or stage may be made with comparative ease and economy.
At important stations, the stage is measured at short intervals and discharge is measured once or
twice each day. At less important stations, only stage measurements are made regularly.
A continuous measurement of discharge in a natural channel is comparatively difficult,
time consuming, expensive and requires special skills. Therefore, the discharge at a site is
measured less often and is estimated by indirect methods. Fortunately, there exists a relation
between stage and discharge at a section. This relation is termed as stage-discharge relationship
or rating curve. This relationship is used to transform the observed stages into discharges. Note
that the reliability of such discharge records is dependent on the reliability of stage data and the
correctness of the stage-discharge relation.

At many sites, the discharge is not a unique function of stage; variables other than stage
must also be simultaneously measured at such sites to correctly estimate discharge. For example,
if variable backwater occurs at a site, the data on water surface slope is required. The slope can
be measured by installing an auxiliary gauge downstream. The rate of change of stage is an
important variable where the flow is unsteady and channel bottom slopes are flat.

4.1.2 Types of Station Control

A permanent and stable stage-discharge relation is desirable from the point-of-view of
hydrologic measurements and to establish such a relation, the river channel at the gauging station
must be able to stabilize and regulate the flow at the station such that for a given stage, the
discharge past the station is always the same. The stability, reliability, and shape of the stage-
discharge relation are normally controlled by a section or a reach of the channel at or
downstream from the station and are known as the station control. Station controls can be
classified in four categories:
· section and channel controls
· natural and artificial controls
· complete, compound and partial controls
· permanent and shifting controls

When any change in the physical characteristics of the channel downstream to the control
has no effect on the flow at the gauging section, such control is termed as section control.
Natural or artificial local narrowing of the cross-section (waterfalls, rock bar, gravel bar) creating
a zone of acceleration (Fig. 4.2) are some examples of section controls. The section control
necessarily has a critical flow section at a short distance downstream.
Fig. 4.2 A section co
ontrol

A cross sectiion where non acceleratiion of flow occurs or w where the accceleration iis not
sufficientt to prevent passage of disturbances from the ddownstream m to the upstrream directiion is
called as a channel control.
c The length of th he downstreaam reach off the river afffecting the rrating
curve deppends on thee normal or equilibrium
e depth he andd on the eneergy slope S [L ∞ he/S, w
where
5/3 0.5
5 0.5 3/5
he follow
ws from Man nning’s Q =KmBhe S (wide rectaangular channnel) so he = [(Q/KmS ) ].
The length of channeel effective as
a a control increases
i witth dischargee.

An
A artificial control
c is sp
pecifically co
onstructed too stabilize thhe relationshhip between stage
and dischharge. Thesee include weeirs and flum mes, dischargging under ffree flow coonditions. Naatural
section controls
c incllude a ledgee of rock accross a channnel, the brinnk of a watterfall, or a local
constrictiion in width (including bridge
b openiings).

Natural
N contrrols can have a range off geometry aand stabilityy. Some natuural controlss may
have a siingle featuree such as a rock
r ledge across
a the chhannel at thee crest of a w
waterfall, theereby
forming a complete control.
c Thuus complete control
c goveerns the stagge-discharge relation oveer the
whole rannge of stagee experienced d. However, in many casses, station ccontrols are a combinatiion of
section control
c at low
w stages andd a channel control at hiigh stages annd are thus called comppound
or compllex controls.

Where
W the geeometry of a section and d the resultiing stage-disscharge relattionship doees not
change with
w time, it is described d as a stable or permanennt control. SShifting conttrols changee with
time thesse may be seection contro
ols such as boulder, gravvel or sand riiffles which undergo perriodic
or near continuous
c sccour and depposition, or they may bee channel coontrols with erodible bedd and
banks.
 The amount of gauging effort and maintenance cost to obtain a record of adequate quality
is much greater for shifting controls than for permanent controls. Since stage discharge
observations require significant effort and money, it is always preferred to select a gauging site
with a section or structure control. A complete control has many advantages: permanence, easy
installation and running of the gauging site, and favourable conditions for current meter
measurements. However, a complete control is not practicable in many cases and one may have
to be content with either channel control or a compound control.

4.1.3 Main elements of a streamflow gauging station

The main elements required at a streamflow gauging station are as follows:
a) a stage measuring device (manual observation) or a stage sensing and recording device;
b) a control section or reach;
c) a section suitable for measuring discharge;
d) set-up and devices necessary to measure discharge;
e) office building and place to keep equipment, spare parts, and accommodation, etc.

The International Standard Organisation (ISO) has brought out a large number of
standards dealing with measurement of liquid flow in open channels. Technical committee
TC113 of ISO deals with this topic. The publication ISO (1983) is a useful collection of
standards dealing with various aspects of streamflow measurement. WMO has also brought out
many publications related to streamflow measurement. Of course, the individual countries have
their own standards. The topic is covered in detail in Herschey (1986 and 1995) and Boiten
(2008). Site requirements for measurement of discharge using current meters or floats are given
in ISO 748.

Now, we will discuss measurement of river stage and discharge.

4.2 Measurement and Recording of River Stage

The terms stage and gauge height are interchangeably used to express the elevation of the river
water surface with respect to an established datum. As shown in Fig. 4.3, river stage is the
vertical difference between the water surface of a river relative to an established gauge datum.
The datum may refer to an arbitrary datum that is selected for convenience or the national
reference (e.g., the mean sea level or the datum of the survey of India or an arbitrary point,
slightly below the point of zero flow in the stream). To eliminate the possibility of negative
values of the gauge height, the datum selected should be below the elevation at which the flow is
zero. The gauge height is usually expressed in hundredths or thousandths of a meter. River stage
is a basic variable representing the state of a river.

River stage

Gauge datum
 Figuree 4.3 Cross section of a river
r with gaauge datum aand river stagge.

Usually,
U it iss difficult annd expensivve to measurre the dischharge of a rriver directlyy and
continuouusly. Therefore, one frequently
fr measures
m thee water stagge which iss easy and then
converts the stage vaalues to disch harges. Recoords of stagee are used w
with a stage-ddischarge rellation
to obtain
n the record ds of stream m discharge. Clearly, thhe reliabilityy of the disscharge recoord is
dependennt on the reliiability of th
he stage recorrd and the sttage-discharrge relation.

The
T water lev vel data is measured
m using a varietyy of equipmeent: staff gauuges, autographic
water lev
vel (chart) reecorders, an
nd digital typ
pe water levvel recorderss. These cann be classifiied in
two broad categories: (1) Non-reecording, maanual gaugess, (2) Recordding gauges.

4.2.1 Non-recordin
N ng river sta
age gauges
The simp plest way to measure riv ver stage is by
b the use of a staff gauuge which iss basically a scale
installed such that a portion of itt is always immersed
i inn the water. TThis gauge m may be a veertical
scale attaached to a co
olumn, pillaar, bridge pieer, or other sstructure thaat extends innto the low-w
water
part of thhe channel. Such staff (manual) gaauges are sim mple and innexpensive bbut must bee read
frequentlly.

Where
W the raange of wateer levels ex xceeds the caapacity of a single verttical gauge other
gauges may
m be instaalled in the linel of a cro
oss-section nnormal to thhe direction of flow (seee Fig.
4.2). Thee scales on suuch a series of stepped staff
s gauges should overllap by not leess than 15 ccm. A
ramp gau uge consistss of a scale marked on or securelyy attached too a suitable inclined surrface,
which clo osely follow
ws the contou ur of the rivverbank. Thee ramp gaugge may lie onn one continnuous
slope orr on more slopes. It should lie
on the line
l of a crross-section normall to
the
directioon of
flow.

A
staff or
ramp gauge is an inexpensive, simple, and reliable method of measuring water level. By using it,
water level can be measured by relatively unskilled staff. Ramp gauges amplify surges and
ripples but provides the opportunity to of a higher resolution. A staff gauge can only be used for
spot measurements. It is difficult to obtain readings in the field with a true resolution higher than
±5 mm.

Figure 4.4: Staff gauge – pictorial view (left) and markings (right)

Staff gauges can be either vertical or inclined. Vertical staff gauges are normally
porcelain enameled iron sections with a scale (with graduations of 5 or 10 mm) marked on or
securely attached to a suitable vertical surface graduated every 10 mm. The vertical staff gauge
is used as an inside reference gauge (if installed in a well), or as an outside gauge if installed in
the stream. Where the water level of the river varies over a large range, observations using a
single staff gauge might be difficult. In such cases, the gauge consists of stepped sections
installed at different locations in a line normal to the flow. Each of these stepped gages should
refer to the common datum and they should overlap by not less than 15 cm to ensure continuity
of readings and confirm their consistency with each other. Staff gauges are manually read,
generally each day in the morning in lean season and at (multi) hourly intervals during high
flows.

An inclined staff gauge is usually a graduated surface attached securely to a permanent

foundation. Rocky outcrops on river banks make good base for inclined gauges. When inclined
gauges are built flush with the stream bank, they are less likely to be damaged by floods, floating
debris, or drift than are projecting vertical staff gauges. Such gauges should be located as close
as possible to the measuring section, without affecting the flow conditions.

Manual gauges are to be read at fixed times whereas the recording gauge provides a
continuous data of the variation of stage. In manual observation, commonly the stage is read
daily in fair weather and (multi)hourly in monsoon season. In automatic recording stations, a
continuous record of stage is obtained by utilizing water level sensors interfaced with a analogue
recorder (chart) or a digital recorder (logger or telemetry).

The advantages of the non-recording gauge are low initial cost and ease of installation.
The disadvantages are the need for an observer (and even then data will observed at limited
times) and less accuracy. For a long-term operation, the advantages of a recording gauge
outweigh those of a non-recording gauge. Sometimes, an automatic and a non-recording gauge
are maintained together because the electro-mechanical recording gauge equipments are liable to
breakdowns. The values of stage may be required as a single instantaneous measurement, as a
short series of instantaneous measurements or as a continuous or practically continuous record of
the fluctuations of stage.
Even when a gauge recorder is used, the observer should visit the station from time to
time to ensure satisfactory performance of the sensor and recorder. Observer should note the
time and date of such checks along with the staff gauge and recorder values. It is essential that
the staff gauge itself is maintained such that it is safely accessible and legible to the observer.
Further, the observer should always visit the site following a major high flow event to ensure
continued measurement and recording of data.

4.2.2 Depth Sounding and Suspension Equipment

During field investigations, depth of water from surface to the river bed may have to be
measured. Depending on the velocity and depth of flow, either a sounding rod or a sounding line
is used for this purpose. A sounding rod is a graduated rigid rod with a base plate. It is used to
measure depths up to 5 to 6 m in flows with velocities up to 2 m/s. For measurements by either
sounding rod or wading rod, the rod must be held in a vertical position. For measurements by
sounding line, appropriate weights must be attached to keep it as close as practicable to vertical.

For smaller depths and velocities, a wading-rod is used; for greater depths, a sounding
line is used. A sounding rod should be as lightweight as possible but sufficiently strong to
withstand the force exerted by flowing water without undergoing significant deflection or
vibration. It should remain straight and vertical during use; should not cause significant heading
up of water and should not penetrate into the channel bed.

To make observations in flowing water with the help of suspension equipment, the
measuring equipment must be placed at the point of measurement in such a way that it does not
cause appreciable disturbances, irrespective of the depth of water and velocity of flow.

4.2.3 Water Level Recorder

As the name suggests, a water level recorder (WLR) is an instrument which senses and records
water level. It basically consists of a time element and a gauge height element which together
produce a time-series of water levels. The time element is controlled by a clock while the gauge
height element is activated by a float or a pressure actuated system. These recorders can be
classified as either analogue type or digital type, depending on the way the data are recorded.
The analogue type recorders produce a graphic record of fluctuations of the water level with
respect to time.

The water level recorders are generally of shaft-angular-input type, and the angular
rotation of the shaft is recorded. The depth of water surface is sensed for automatic recording by
a float in a stilling well which follows the rise and fall of the water level. A gas-purge system
that transmits the pressure head of water in a stream to a manometer is known as a bubble gauge.

A water level recorder gives a continuous record of the water level on a chart from which
values are manually extracted at desired intervals. The data from a digital water level recorder
are stored in an electronic memory and these are downloaded to a computer. The data can either
be recorded at equal intervals of time, usually at (fraction of) an hour interval, or at only those
instants when there is a change in water level by more than a pre-set amount.

(a) Float-type water-stage gauges

A float-type gauge consists of a float which is installed in a stilling well, a graduated tape or
wire, a counterweight, a pulley and a pointer. The tape or wire runs over the pulley and is kept
taut by the action of the counterweight. In this way the float that positions the tape with respect
to the pointer senses the stage fluctuations. As shown in Fig. 4.5, this device records the motion
of the float as it rises and falls with river water level. Motion of float and counterweight are
translated to move a pen or stylus while the chart moves at right angles and thus a continuous
record is produced. The stilling well protects the float and other accessories from floating debris
as well as moderates the rapid fluctuations in the stream level. Such gauges are appropriate for
streams with narrow, incised gravel-bed channels, so that the stilling well can be located close to
the stream. For wide and sandy channels, the stilling well has to be located on the stream bank
some distance away. Pipes connecting the stilling well to the stream are vulnerable to blockage
by siltation and the water level in the stilling well may not faithfully rise and fall with river water
level.

A float gauge can provide a direct record of river stage and no external energy source or
battery is required. It provides almost uniform resolution throughout the range and good
accuracy. But being a mechanical device, it is subject to errors from hysteresis and friction.
Construction and maintenance of stilling well is expensive, particularly in places where rivers
carry large amount of sediments.

A chart recorder produces a continuous hard copy record. Quality of the recording
mechanism affects the accuracy of the chart record. Subsequently, the data is manually converted
to digital format; this process is labour insensitive, time consuming, and may introduce errors.
Figure 4..5 Float typee recorder

(b) Bubble gauges

Bubble gauges
g recorrd the presssure required d to maintaain a small flow of gass from an oorifice
submergeed in the streeam and thiss is an indicaator of the w
water level inn the river. F
Fig. 4.6 show
ws the
gauging arrangemen nt by using bubble
b gaugge. The advaantage of suuch gauge iss that is doees not
need a co
ostly stilling arrangemennt like the onne needed forr float operaated gauges.

Figure 4.6: Bubbler ggauge

4.3 Direct Determination of Discharge
Discharge is the volume of water passing through a certain section in a unit time period. It is
commonly expressed in cubic metres per second (m3/s or cumec). The discharge at a site is a
function of the cross-section area and flow velocity. The cross-section area, in turn, depends on
river stage. At most stations, discharge is measured once a day; at important stations or during
high flows, it might be measured more frequently.
Discharge measurement techniques can be broadly classified into two categories: (i)
direct determination, and (ii) indirect determination. There are many methods under each
category. The important ones are discussed here.

In the direct determination methods either discharge itself is measured or some variable
on which discharge depends is measured. The commonly used methods are: velocity-area
methods, dilution techniques, electromagnetic method, and ultrasonic method. The first two are
described here.

4.4 Velocity-Area Methods

Discharge is the product of cross-sectional area and velocity of water. The velocity-area methods
involve measuring the flow area and velocity and these are multiplied to get discharge:

Q=v*A (4.1)

where Q = discharge [m3/s], v = velocity [m/s], and A = cross-section of flow [m2].

Since the velocity of flow at a cross-section varies laterally and with depth, it is not
enough to measure the velocity at a single point. Depending on the accuracy required, the width
of the stream is divided into a number of vertical portions (Fig. 4.7). In each of these portions,
the velocity is measured at one or more points along the depth to get a representative velocity in
that portion. The area of the individual portion can be easily calculated if the bed profile and
stage are known. The velocity may be measured by a conventional method (for example, float or
current meter) or by an advanced procedure, for example, the moving boat technique.
Fig. 4.7 Schematic sketch
s for a velocity-area
v a station.

Water
W particlles in a riverr cross-section travel at different veelocities. Thhey are subjeect to
friction as
a they com me into contact with th he sides andd bottom off the channel. Due to these
frictionall effects, waater flows faastest near thhe surface annd center off the channeel (away from m the
immediatte frictionall influences)). A typicall velocity pprofile is shhown in Figg. 4.8 where the
velocity varies as a parabola fro om zero at the channel bottom to a maximum m at (or nearr) the
surface. Also
A shown are the contours of equaal velocity inn typical riveer cross sectiion.
Fraction of Depth (%) 

Velocity (m/s)
Figure 4..8 Typical veelocity profile (top) and velocity disttribution (isoovels) in a cchannel (botttom).

4.4.1 Velocity
V Meaasurement by b Floats
A float iss a distinguisshable article that floats on the wateer surface, suuch as a woooden log, a pplastic
bottle paartly filled with
w water, or o branch off a tree. Surrface or neaar-surface flooats are useed for
streamfloow measurem ment. Norm mally, these are
a wooden cylindrical rods of nearrly 0.5 m leength.
They aree shaped succh that they float nearly y vertically with one thhird of the llength protruuding
above thee water surfaface. The floaats are paintted in bright colours for easy identiffication in m
muddy
or turbullent water. At
A times, a floating
f article such as a lump of vvegetation m may be usedd as a
float.

For a float measurement

m , two cross--sections suffficiently farr apart on a straight reaach of
river are selected. Thhe upstreamm and downstream cross sections shoould be suffficiently far apart
for accurrate assessm
ment of float travel time (3( to 5 timess the width oof the river or a minimuum of
20 seconds travel timme). If there is
i a structuree across the river (say a bridge), the upstream seection
should bee sufficiently
y far from th
he structure to
t avoid its eeffects. Bothh the cross seections (upsttream
and down nstream) shoould be cleaarly marked by placing m markers so that the exaact time wheen the
float crossses the cross-section caan be identiified. The uppstream chaannel cross ssection shouuld be
divided into
i a numbber (preferabbly an odd number)
n of eequal segmeents as practtically feasibble in
which thee floats will be placed.
When
W doing measuremen nts, an obserrver each is positioned aat upstream and downsttream
ends of the reach su uch that theey are visiblle to each oother. The ddownstream observer accts as
timekeepper and carrries a stop watch.
w Floaats are introoduced acrosss the stream m width a short
distance before the actual upstream cross-ssection so thhat they losee inertia andd travel witth the
velocity of water when they cross the upstream cross-section. Floats may be tossed from a bridge
or cableway; if there is no such facility than they can be thrown in the water from the river bank.
When the float crosses the upstream cross-section, the upstream observer gives a signal to the
downstream observer who notes the time taken by the float to cover the distance.

The velocity of the float is equal to the distance between the two cross-sections divided
by the time taken by the float to cover this distance. The mean velocity in the vertical is
determined as the float velocity multiplied by a coefficient which varies between 0.80 and 0.85.
This coefficient depends on the shape of the velocity profile of the river and the depth of
immersion of the float. Discharge for the segment is calculated as the segment area multiplied by
the segment velocity. Total discharge is obtained as the sum of segment discharges and mean
velocity of flow is computed as the total discharge divided by total area.

Floats can rarely be positioned at the desired location in the cross section. If these are
thrown manually, it is difficult to throw them very far and hence only the velocity of water near
the banks can be measured. Float measurements are not very reliable and this method is normally
restricted to an emergency or to measure high discharges when current meter is not available or
can’t be used. Hence, floats should be used only when it is not possible to use any other better
method.

4.4.2 Measurement of Flow Velocity by Current Meters

Current meter is the most commonly used instrument to measure the velocity of flowing water.
Accurate measurements of the velocity profile of the stream cross section are made by current
meters. A current meter consists of rotating element (rotor) whose movement is due to the
reaction of the stream current. The angular velocity acquired by the rotor is proportional to the
velocity of water. By placing a current meter at a point in a stream and counting the number of
revolutions of the rotor during a time interval, the velocity of water at that point is determined.
Current meters are of two types: those having a propeller rotating around a horizontal axis and
those having a series of conical cups mounted around a vertical axis. The normal range of
velocity which can be measured varies from 0.15 to 4.0 m/s.

Horizontal-axis meters consist of a propeller mounted at the end of a horizontal shaft

(Fig. 4.9). These are available in a range of propeller diameters. Propellers made up of plastic
have been introduced recently. These are cheaper and respond more quickly to changes in
velocity. Some current meters come with propellers of different pitch and diameter to suit
various flow conditions. The horizontal axis rotor with valves cause less disturbance to flow than
vertical axis rotors. Furthermore, due to axial symmetry with the flow direction, the rotor is less
likely to be entangled by debris than vertical axis rotors and the bearing friction is less compared
to the vertical axis rotors. The vertical axis rotor with cups or valves can operate in lower
velocities than the horizontal axis meters, the bearings are well protected from silty water, and a
single rottor serves fo
or a range off velocities.

In
n the propelller-type currrent meters (Figure 4.99), a propeller rotates abbout a horizzontal
axis due to force appplied by flowwing water. The revoluttions per timme interval aare recordedd. The
relation between
b revoolutions per second N of
o the currennt meter and the water veelocity v is ggiven
by follow
wing equation

v = a + b*N (4.2)

where b = constant ofo proportionnality and a = starting vvelocity or velocity requuired to overcome
mechaniccal friction. These consstants differ from one ccurrent meteer to the othher as a resuult of
manufactturing variattions as well as changee with time due to weaar and tear. Therefore, each
current meter
m should be recalibraated periodiccally.

Figure 4..9: Propellerr-type curren

nt meter

The
T velocity measuremen nts made byy using curreent meters arre usually cllassified in tterms
of the meeans used to o cross the stream
s durin
ng measurem ments, such aas wading, ccableway, brridge,
or boat. Wading
W is possible
p in sm
mall streamss of shallow
w depth onlyy; the currentt meter is heeld at
the requiisite depth below
b the su
urface by ann observer wwho stands inn the water.. In narrow well-
defined channels,
c a cableway is stretched from bank to bank weell above thhe flood levvel. A
carriage moving
m overr the cablewway serves ass the observaation platforrm. Bridges are advantaggeous
from the viewpoint of o accessibillity and tran
nsportation, aalthough theese are not tthe best locaations
from hyddraulic pointt of view. Th he velocity measuremen
m nt is perform
med on the doownstream oof the
bridge to
o minimize th he instrumennt damage due
d to drift aand knock aggainst bridgee piers. Boatts are
most satiisfactory for measuremen nts in wide rivers.
r
 The
T section liine at the gaauging site iss marked by permanent survey markkings. The ccross-
section along
a this seection line is
i determineed by surveyying with thhe help of ssounding rods or
soundingg weights. When
W the deppth of water is
i more or iff quick and aaccurate deppth measurem
ments
are needeed, an echo sounder
s is ussed.

A discharge measuremen
m nt requires determination
d n of sufficieent point vellocities in a river
cross secction to perm
mit computattion of an average veloccity in the sttream. The sstream is divvided
into a nu
umber of verrtical section
ns. The sum of cross-secctional area oof each sectiion multiplieed by
the averaage velocity of each sectiion gives thee total dischaarge:

n m 
Q total   Q i  (A i  v i ) (4.3)
i 1 i 1

where Qtotal
t = total discharge,
d mean
m = number of sections, Ai = cross area of secttion I, vi = m
velocity of
o section i, Qi = discharrge in sectio
on i.

A current meeter measuress the velocitty at a point.. However, the mean veelocity in eaach of
the selectted vertical segment
s is required
r to estimate discharge. The mean velociity in a vertiical is
determinned from velo ocity observvations at on
ne or more p oints in thatt vertical. C
Current meterrs are
held dowwn and posittioned at thee required lo ocation in fl owing waterr by soundinng weights. The
weights are
a connecteed to the currrent meter by y a hanger aand pin assem
mbly.

The
T number of velocity determinatio ons is limiteed to those w
which can bbe made witthin a
reasonable time. If the river stage
s is chaanging rapiidly, one shhould quickkly completee the
measuremment with a minimum chhange in watter stage.

Figure
F 4.10: Procedure for
f a currentt meter meassurement
 Usually there are national guidelines detailing how the vertical sections should be chosen
and in which depth the velocity measurements should be made to determine the mean velocity in
the vertical profile. These guidelines aim to ensure homogenous data sets. Two methods that are
frequently used to determine the mean velocity over the vertical profile are:

Two point measurement: The variation for most channels is such that the average of the
velocities at 0.2d and 0.8d below the surface equals the mean velocity in the vertical.
One point measurement: The velocity at 0.6d depth below the surface closely approximates the
mean in the vertical. The adequacy of these assumptions for a particular stream can be tested
by making a detailed vertical velocity measurement.

4.4.3 Stage-Discharge Gauging Stations in Natural Channels

When records of water level are to be used as a basis for computation of discharge, the relation
between water level and flow must be determined. In a stable channel with an appropriate control
feature which is stable and sensitive, a single relation may exist between water level and
discharge. In this case, the relation can be determined by taking discharge measurements
throughout the range of levels and flows required to be measured. Several techniques are
available for this purpose including, current meter gauging and float gauging, dilution gauging
transit time acoustic methods, the use of Doppler velocity meters (fixed hydro-acoustic
installations) or the Acoustic Doppler Current Meters. The frequency of any maintenance or
operational performance shall be such that the accuracy and timeliness of data provision meets
the users’ requirements.

4.4.4 Discharge Measurements

Discharge measurements using the above techniques shall be related to a stage reading taken at
the beginning and end of the discharge measurement and during the measurement if the stage is
changing rapidly or inconsistently. When sufficient numbers of discharge measurements have
been taken, a stage discharge relationship can be computed. Subsequent to the formulation of this
stage-discharge relationship, only occasional discharge measurements need be taken at flows in
the normal range to confirm the robustness of the relationship unless the site is subject to shifting
control conditions.
Opportunities should be taken to carry out discharge measurements in extreme events in
order to extend the stage discharge relationship. Discharge measurements made using the
velocity-area methods can be performed using rotating-element current meters, acoustic Doppler
velocimeters, or acoustic Doppler current profilers. These can be made by wading the stream or
small river with the meter mounted to a wading rod or by suspending the meter and a sounding
weight from a bridge, cableway or static boat. Velocity-area methods using floats is another
option when the presence of floating debris or very turbulent conditions precludes the use of
meters.
 Acoustic Doppler current profilers deployed from powerboats, remote-control boats, or
tethered rafts also can be used. Tethered rafts typically are deployed from bridges or cableways.
Another variation of velocity area methods is the slope-area method, which is typically used to
compute flood discharge indirectly by surveying the cross-sectional properties and water-surface
profile after the flood. Where a pre-surveyed cross-section is used for the purpose of discharge
measurement, then the section shall be checked following any major events e.g. over and above a
bank full flow.
4.5 Indirect Determination of Discharge
These methods make use of the relationship between the flow discharge and the water surface
slope and depth at specified locations. Two important indirect methods are the slope-area method
and use of flow measuring structures.

4.5.1 Estimation of Discharge by Slope-Area Method

Slope-area method is an approximate and indirect method of discharge estimation which is used
when measurement by more accurate methods, such as the velocity-area method, is not possible.
In this method, discharge is estimated by observing the water surface slope and cross-section
area. Although the accuracy of slope-area method is less compared to the velocity-area methods,
it is sometimes necessary to use this method because the magnitude of flows is so high and flow
conditions are such that the other methods of discharge estimation cannot be used. Such situation
may arise due to reasons like non-availability of equipment required for making current meter
measurements, inaccessibility of the site due to floods, presence of debris and floating matter in
the flood flow preventing the use of current meter, etc. In flashy streams, high floods may pass
without being measured due to their short duration. If the magnitude of such floods is to be
assessed after their occurrence, slope area method can be used. Discharge values should
estimated by slope-area method should not be used to establish rating curves.

A measuring reach of the stream is chosen for which the mean cross section area and the
slope of water are determined. The mean velocity is computed by using the open channel flow
formula such as the Manning's formula. Appropriate value of the rugosity coefficient is chosen
depending on the physical conditions of the channel. Discharge is computed as the product of the
mean velocity and the average cross-sectional area of the reach.

For the measurement reach, three things should be known: (i) The cross-sectional
geometry and properties at its ends, (ii) the value of Manning's roughness coefficient n, and (iii)
water-surface elevations at the end sections. The river reach should be straight (to the extent
possible) having stable bed and banks and uniform cross-section over a length of at least five
times the width of the channel. In any case, the length should not be less than about 300m. As far
as possible, the river reach should be long enough so that the difference between water levels at
the upstream and downstream gauges is not less than ten times the uncertainty in the difference.
The flow in the reach should be free from significant disturbances, draw-down or back-water
effect of any structure or tributary joining upstream or downstream and it should be contained
within its banks for all stages at which this method is used. If no uniform reach is available, the
reach should preferably be converging rather than diverging. The measurement site should be
easily accessible at all times.

A minimum of three cross-sections are generally desirable in the selected reach (two
should be at the ends and others in-between, one at the centre). Water surface slope is computed
from the gauge observations at either end of the reach; intermediate gauge(s) are used to confirm
that the slope is uniform throughout the reach. Cross-section area and wetted perimeter are
estimated at each cross-section. Manning’s n should be estimated only by a trained person.

Measurement of slope
The upstream and downstream water level gauges shall be read simultaneously. If there is only
one observer and this is not possible, the upstream gauge shall be read first, followed by the
downstream gauge, followed by a second reading of the upstream gauge. If the value of the
upstream gauge has changed by more than 10 mm between the two sets of readings then the
average of the two upstream readings should be used. Effort should be made to gauge readings to
at least the nearest cm or better.

Slope of water surface is computed from the average of gauge observations at either ends
of the reach. The intermediate gauge(s) are used to confirm that the slope is uniform throughout
the reach. When accurate gauges do not exist or have been damaged, flood marks on the banks
may be used to estimate the slope. In such cases, an effort should be made to locate and use as
many flood marks as possible in the reach with least possible delay after the flood.
If intermediate gauges are installed, the readings at these gauges should also be taken and
the slopes between the upstream and intermediate gauges, and intermediate and downstream
gauges estimated and compared with the slope obtained using the upstream gauges. If there is a
significant difference in the three slope estimates (say greater than 10%) then the validity of the
estimates should be ascertained.

Since slope-area method is frequently used to estimate peak discharges following a flood
event, observation of high-water marks after the passage of a flood is important. This involves
identifying debris marks and then topographic level surveys to find their levels. Several types of
high-water marks may be found, such as drift on banks, mud lines on trees and banks, and drift
in bushes or trees. Each high water mark should be rated as excellent, good, fair or poor; this
could help when interpreting the high-water profile. Sufficient wrack marks need to be identified
on both banks to fully define the water profile over the measuring reach. A graphical profile of
the high-water marks should be made so that irregularities in the profile can be removed.
Sufficient number of points along the measuring reach need to be used to define the surface
water profile.

Estimation of Discharge
Fig. 4.11 shows the longitudinal section of a river between two sections, 1 and 2. The head at a
cross-section consists of water surface elevation and the velocity head. The head loss is made up
of two parts: frictional loss and energy loss due to expansion or contraction.
Fig. 4.11 Channel reeach for the slope-area
s method.
m

First the fall (Z1 - Z2) ov

ver the measu
uring reach is estimatedd and then thhe water sloppe by
dividing the fall by th ( between sections 1 aand 2:
he distance (L)
Sw = (Z1 - Z2)/L
)

Friction slopee Sf can be computed by

V2 V 
2
( Z 1 - Z 2 ) +  1 - 2  (1 - k )
2 g 2 g 
Sf=
L (4.4)

where L is the reach length, k is the coefficieent for energgy loss; its vvalue is 1 forr contractionns and
0.5 for ex
xpansions. Estimate
E the hydraulic raadius for botth the upstreeam downstrream sectionns (R1
& R2) using
u the esstimated crooss-sectionaal areas (A)) and wetteed perimeterrs (P): R= A/P.
Accordin ng to Manninng's formulaa, the mean velocity
v in reeach 1-2 is calculated as

v1-2 = (1/n)R2/3
2
S1/2 (4.5)

where R is the hydraaulic mean depth and n iss Manning's roughness ccoefficient. If A is the ccross-
section area, then thee discharge Q is

Q = (1/n)AR2/3
2
S1/2 = K S1/2 (4.6)

Using
U the esttimated hydrraulic radius, conveyancce (K) of thee channel at the upstream
m and
downstreeam sectionss is computeed. Conveyan nce dependss on channell characterisstics. As the flow
in the reach may not be truly uniform, average conveyance of the reach is computed as the
geometric mean of the conveyances of the two end sections (K1 and K2):

1
K1  A1 R12 / 3 (4.7)
n

1
K2  A2 R22 / 3 (4.8)
n
Using the values of K1, K2 and the surface water slope, the following equation is applied
to provide a first approximation of discharge:
Q K K S 
1 2 f (4.9)

Using the value of Q, first estimates of the velocity heads at sections 1 & 2 are obtained
by:
2
V2 Q  1
VH1  1   1   (4.10)
2 g  A1  2 g

2
V2 Q  1
VH 2  2   2   (4.11)
2 g  A2  2 g

If VH1 < VH2, i.e., the velocity head difference is negative, then the reach is contracting
and the following equation can be used to obtain a second estimate of the slope of the energy
line:

Z1  Z 2    1V1   2V2 

 2 2


S  2g 2g 
(4.12)
L

If VH1 > VH2, i.e. the velocity head difference is positive, then the reach is expanding which
should be avoided. However, if an estimate is necessary, the following equation is used:

Z1  Z 2   0.5 1V1  2V22 

 2
 
 2g 2 g 
S (4.13)
L

Using the new estimate of the energy line obtain a second approximation of discharge (Q2). If
(Q2 - Q1)/Q1 > ± 1%, repeat the iteration procedure until the current approximation of Q is within
1 % of the previous estimate.
 The slope-area method can be used with some degree of accuracy in open channels with
stable boundaries, or in channels with relatively coarse bed material. This method may also be
used in other cases, such as alluvial channels including channels with over-bank flow or non-
uniform channel cross-sections, subject to the acceptance of large uncertainties involved in the
selection of the value of the rugosity coefficient, such as Manning's roughness coefficient n.

Cross-sectional area and wetted perimeter of stream

Generally average area of cross-section in the observation reach is taken as the mean of three
sections - two end sections and the central section. If for any reason, it is not possible to measure
more than one cross-section, the data of the central one only may be observed. The cross-
sections should be measured for each discharge observations at the same time (or as near as
possible) when the gauge observations are made. Rivers with rocks in the bed and banks and
carrying little sediment are least susceptible to changes in cross-sections and in such cases it will
be sufficient to observe the cross-sections before and after the floods.

If the reach is substantially uniform and there are insignificant differences in the cross-
sectional areas, A1, A2 .... Am at the chosen sections, the mean area of cross- section for the reach
of the stream may be taken as

A = (A1 + 2A2 + .... + 2Am-1+ Am )/ [2 ( m - 1)] (4.14)

Mean wetted perimeter for the reach may also be estimated similarly

Value of roughness coefficient

Where a reasonable value of roughness coefficient (also called rugosity coefficient) can be
determined from actual discharge measurements at the nearest lower stages by a more accurate
method, say the area velocity method using current meter, the value so obtained may be chosen.
In the absence of measured data, the values given in table 4.1 may be assumed for open channels
with relatively coarse bed material.

Table 4.1 Typical values of roughness coefficient for channels with relatively coarse bed
material
S. N. Type of bed material Size of bed material equivalent Rugosity coefficient 'n’
diameter in mm
1 Gravel 4 to 8, 8 to 20, 20 to 60 0.019 to 0.020, 0.020 to
0.022, 0.022 to 0.027
2 Cobbles and Shingle 60 to 110, 110 to 250 0.027 to 0.030, 0.030 to
0.035
 The major limitation of the slope-area method is the difficulty of a correct assessment of
the rugosity (roughness) coefficient n in the Manning's formula. This method can be used with
some degree of accuracy in channels with stable bed and banks having relatively coarse bed
material. This method may also be used in other cases, such as alluvial channels, subject to the
acceptance of larger errors involved in the selection of the value of the rugosity coefficient n.
This method is not suitable for large channels, rivers with very flat slopes and high sediment
concentration, and channels with significant curvature.

4.6 Moving Boat Method

On very large and wide streams, the conventional methods of measuring discharge by current
meter are frequently impractical or involve costly and tedious procedures. The moving boat
method can be very useful in such a situation. This method can also be used at sites where
unsteady flow conditions require rapid completion of measurement.

The moving-boat technique is a velocity-area method of determining discharge in which

data are collected while the observer is aboard a boat that traverses the stream along a pre-
decided path (Fig. 4.12). During the traverse across the river, an echo sounder records the bed-
profile of the cross-section and a continuously operating current meter measures the combined
stream and boat velocities. The angle between the current meter (which aligns itself in a direction
parallel to the movement of the water and the pre-selected path) is also measured. The velocity
that is measured at each of the observation points in the cross-section (vv in figure 4.14) is the
velocity of water past the current meter resulting from both stream flow and boat movement. It is
the vector sum of the velocity of water with respect to the stream bed (v) and the velocity of the
boat with respect to the stream bed (vb). Data are collected at many observation points in the
cross-section for each run.

The discharge is calculated in a manner similar to the conventional velocity-area method

by summing the products of the segment areas and average velocities. Since the current meter is
suspended about 0.5 m to 1.0 m below the water surface, a coefficient is required to adjust the
measured velocity to the average velocity in the vertical. Investigations on several rivers have
shown that the coefficient generally lies between 0.85 and 0.92. An accuracy of 5 per cent is
claimed in the moving boat method.
 Fig. 4.1
12 Moving boat
b equipment for disch
harge measur
urement.

The
T principaal differencee between a conventionnal measureement and the movingg-boat
measuremment is in thet method of computing water vvelocity. In the case of a conventtional
techniquee the mean velocity in the segmentts of a crosss-section of the stream is determineed by
point velocities meaasured at seeveral depth h is in the vertical. T
The moving-- boat technnique
measuress the velocityy over the width
w of a segment by suuspending thhe current m
meter at a connstant
depth durring the trav
verse of the boat
b across the
t stream. T The measureed velocity aand the depthh data
are used to determinee the discharrge.

The
T velocity observed at a typical ob bservation pooint in the crross-section (Fig. 4.13), vv, is
the veloccity of waterr past the currrent meter resulting
r froom both streaam flow andd boat movem ment.
It is the vector
v sum of
o the velocitty of water with
w respect to the stream m bed (v) annd the velocity of
the boat with respectt to the streaam bed (vb). The velocityy of water ccan be obtainned by meassuring
the anglee  between n the path off the boat an nd a verticaal vane whicch aligns itself in a direection
parallel to the movemment of the water
w past it.

Referring
R to Fig.
F 4.13, Th ocity v, perpeendicular to the boat patth (true courrse) at
he flow velo
each obseervation poin
nt 1, 2, 3,…, can be deteermined from
m the relationship
v = vv sin  (4.15)

here v is that compoonent of the stream velocity which iis perpendicuular to the ttrue course oof the
boat evenn though thee direction off flow may not
n be perpenndicular.

ation of Disccharge
Computa
After thee cross-sectiion has beenn selected, the
t width off the stream m is divided into an adequate
number ofo segments such that th here is not much
m variattion betweenn two adjaceent verticalss. Of
course, if
i previous measuremen nts have sh hown uniform mity of botth the crosss-section andd the
velocity distribution then fewer verticals may be takenn. It is betterr if no segmment carries more
than 5 to
o 10 percent of the total discharge. Fig.
F 4.14 shoows the crosss section off a river in wwhich
(n-1) verrticals are drawn.
d The velocity aveeraged overr the verticaal at each seection is knnown.
Considerring the totall area to be divided into
o (n-1) segm
ments, the tottal dischargee is calculateed by
the methood of mid-seection as:
n
Q = 
i= 1
( vi a i ) (4.16)

where Q is the total discharge, ai is an individual partiaal cross-secttion area, annd vi is the m
mean
velocity in that areaa. The areaa extends laaterally from
m half the ddistance from m the preceeding
observatiion vertical to half the distance
d to the
t next andd vertically ffrom the watter surface tto the
sounded depth.

Path
h of the
boat

Flow
w vv
vb

Observatioon
verticals v

Fig. 4.13 Moving bo

oat method of
o discharge measuremennt.
Fig. 4.14 Cross-section of a river with (n-1) verticals for measuring discharge.

The sampling data recorded at each observation point provide the necessary information
to determine the velocity of the stream. There are two methods to obtain this velocity, referred to
as method 1 and method 2. Method 1 consists of measuring the angle a between the selected path
of the boat and a vertical vane which aligns itself in a direction parallel to the movement of the
water. An angle indicator attached to the vane assembly indicates angle a.
The reading from the rate indicator unit in pulses per second is used in conjunction with a
calibration table to obtain the vector magnitude vv. Normally data are collected at 30 to 40
observation points in the cross-section for each run. Where practicable, automatic and
simultaneous readings of all required parameters may be recorded

4.6.1 Determination of stream velocity

By method 1 the stream velocity v, perpendicular to the boat path (true course) at each
observation point 1, 2, 3…., can be determined from the relationship

v = vv sin a (4.17)
The solution of equation (1) yields an answer which represents that component of the stream
velocity which is perpendicular to the true course even though the direction of the flow may not
be perpendicular.

Limitations of the method

i. The method is normally employed on rivers which are more than 100m wide and over 2m
in depth
ii. The number of segments should be at least 25.
iii. The width to be taken for each segment depends on the accuracy with which the velocity
in each segment can be measured. The distance between two observation points should be
sufficient to allow the observer to read the instruments and record the results. The
minimum speed of the boat should ensure that it can traverse the section in a straight line.
For the best results this speed should be of the same order as the velocity of the stream.
iv. The river should be sufficiently deep to allow to easily maneuver the boat. Instruments
may be damaged in shallow locations.
v. During the time that the boat traverses the stream the discharge should not change
appreciably.

Example 4.1: The data from a stream gauging station during a measurement campaign are given
below:

Distance from left water 0 1 3 5 7 9 10

edge (m)
Depth (m) 0 1.0 2.5 3.0 2.4 1.0 0
Average velocity (m/s) - 0.25 0.35 0.5 0.4 0.25 0

Compute the discharge in the river.

Solution: First, the average width and cross-section area of flow for each segment are computed.
The average width for the first segment (refer to Fig. 4.14), W1 , is

(1  2 / 2) 2
W1   2.0m
2 x1

The average width of the last segment, W N 1 is

(1  2 / 2) 2
W N 1   2.0m
2 x1

The average width of all other segments, Wi , is

(2  2)
Wi   2.0m
2

The cross sectional area of flow of each segment is

ai  d iWi

For each segment, the discharge is computed as

q i  a i vi

The calculations are shown in the table below.

Distance from Average Depth of Subsection Average Subsection

left water width, W (m) flow, d (m) area of flow velocity, v discharge
edge (m) (m2) (m/s) (m3/s)
0 0.0 0 0 0 0
1 2.0 1.0 2.0 0.25 0.5
3 2.0 2.5 5.0 0.35 1.85
5 2.0 3.0 6.0 0.5 3.0
7 2.0 2.4 4.8 0.4 1.92
9 2.0 1.0 2.0 0.25 0.5
10 0.0 0.0 0.0 0.0 0.0
The total discharge is the sum of segment discharges = 0 +0.5 +1.85 +3.0 +1.92 +0.5 = 7.77
m3/s.
4.7 Discharge Measurement Using Artificial Structures
The structures, such as notches, weirs, and flumes, are frequently used to measure flow in
laboratories. If the physical and hydraulic conditions at the site permit, a fixed, undeformable
structure may be constructed to measure river flow. A number of hydraulic structures are used to
measure flows in field conditions and these are listed here.
a) Thin-plate, sharp crest weirs consisting of a vertically set metal plate: v-notch or
rectangular;
b) Broad-crested weirs: made of concrete or masonry and are used for large discharges;
c) Triangular or Trapezoidal profile weirs;
d) Flumes: Rectangular, Trapezoidal, U-shaped, Parshall or SANIIRI. These have clearly
specified shape and dimensions and are made of concrete, masonry, or metal sheets
depending on their use and location. They depend primarily on the width constriction to
produce a control section.

4.7.1 Weirs
Weirs are used to control upstream water level (to divert water in another channel) or for
measuring discharge or for both. They produce a critical relationship between stage and
discharge by obstructing channel flow; the head over the crest of the weir can be related to
discharge. Weirs have a defined cross-section (either rectangular or triangular) and hence the
computation of discharge is simple. Flow conditions are more complex in flumes; a section of
critical, high-velocity flow is produced by a lateral smooth-walled constriction of the flow. In
both weirs and flumes, the influence of the water level downstream of the structure on flow
conditions at the upstream point of measurement is prevented. Figure 4.15 shows a typical weir
and flume.

Use of these structures in field conditions is limited by the ranges of head, debris or
sediment load of the stream, and the backwater effects produced by these. Since the size and cost
of the structures increase as the size (width and discharge) of the river increases, such structures
are usually built only in upper and middle reaches of rivers. Site requirements consist of a
reasonably straight (at least for a distance of five times the width) approach channel which
should be free of excessive sedimentation, weeds and other aquatic growth. The structure should
be rigid, water-tight, normal to the flow direction, and should be capable of passing high flows
without any damage to its body. The stage-discharge relation at the site depends on the
geometrical characteristics of such a structure.

Problems arise during the use of such structures on large rivers, particularly those that
carry appreciable quantity of sediments because the sediment may be deposited upstream of the
structure and change the discharge characteristics. Periodic inspections of the structure should be
carried out and any damage should be promptly attended.
 Figure 4.15 Typical
T weir and flume

Rating
R of the weir and flumes is carried c out on the bassis of laboraatory calibraation.
Howeverr, the rating should be checked
c afteer constructioon. The hyddraulic princciple that govverns
the use of
o weirs or flumes is thatt these structtures produc e a unique ccontrol sectioon in the floww. At
these stru
uctures, the river
r discharrge is a functtion of the hhead of waterr:

Q = f (H) (4.188)

ge (m3/s), an
where Q is discharg nd H is the head of waater (m) at thhe structuree. For weirss, the
equation becomes:

Q = K Hn (4.19)

where K and n are constants.

c Eq.
E (4.19) is applicable aas long as thhe downstreaam water levvel is
below a certain limitting water leevel known as
a the moduular limit. F
Free flow aree those flows that
are unafffected by thee downstreamm flow cond
ditions. For a two-dimennsional weirr, the discharrge is
estimatedd as

Q = Cd g b H1.5 (4.20)
where Cd is the discharge coefficient, g is the acceleration due to gravity, and b is the crest width
(m). The flow that is affected by tailwater conditions is known as drowned or submerged flow.
Discharge under drowned conditions is obtained by applying a reduction factor to the free flow
discharge.

4.7.1 Flumes
A flume is a flow measuring structure formed by a constriction in a channel. The constriction can
be either a narrowing section of the channel or a narrowing section in combination with a hump
in the invert. Critical conditions are produced providing sufficient amount of constriction. In
such cases, a unique stage-discharge relationship exists independent of the downstream
conditions.

Critical-depth flumes have an edge over a weir in situations when sediment material is
being transported by the flow. This sediment gets deposited upstream of the weir and affects the
discharge relation. The critical-depth flumes consisting only of horizontal contraction would
easily carry the material through the flume. Critical-depth flumes can be grouped into two main
categories, viz., long-throated flumes and short-throated flumes.

Constriction of long-throated flumes is sufficiently long (the length of the throat should
be at least twice the maximum head of water that will occur upstream of the flume) so that it
produces small curvatures in the water surface and the flow in the throat is virtually parallel to
the invert of the flume. This condition results in nearly hydrostatic pressure distribution at the
control section (where critical depth occurs) which, in turn, allows analytical derivation of the
stage-discharge relation. This gives the designer the freedom to vary the dimensions of the flume
to meet specific requirements. Such flumes are usually of rectangular, trapezoidal, triangular or
U-shaped cross-section. For a rectangular flume, the discharge of an ideal fluid is expressed as
3/ 2
2
Q  g bH 3 / 2
 3 (4.21)

here, H represents the upstream energy and b is the typical width dimension for the particular
cross-section shape of the flume. By introducing suitable coefficients, this equation can be
generalized to the following form
3/ 2
2
Q  g Cv Cs Cd bh3 / 2
3 (4.22)

where Cv = coefficient to take in to account the velocity head in the approach channels, Cs =
coefficient to take account of the cross-section shape of the flume, Cd = coefficient for energy
loss, and h = depth of water, upstream of the flume, measured relative to the invert level of the
throat (i.e., gauged head).
4.7.2 Short-Throated Flumes
In these flumes, the curvature of the water surface is large and the flow in the throat is not
parallel to the invert of the flume. The principle of operation of these flumes is the same as that
of long-throated flumes, i.e., the critical condition is created at the throat. However, analytical
derivation of the discharge equation is not possible since large curvature of flow produces non-
hydrostatic pressure distribution. Further, energy loss cannot be assessed. Therefore, it is
necessary to rely on direct calibration either in the field or in the laboratory to establish and
verify the discharge equation. The designer has limited freedom in choosing the dimensions of
the flume and has to select one of the standard designs to meet his requirements. Such flumes,
however, require lesser length and, hence, are economical compared to long-throated flumes.
One of the most commonly used short-throated flumes is the Parshall flume.

Parshall of USA designed a short-throated flume with a depressed bottom known as the
Parshall flume after his name. Parshall flumes have gives satisfactory service at water treatment
plants and irrigation projects. These flumes consist of a short parallel throat preceded by a
uniformly converging section and followed by a uniformly expanding section. The floor is
horizontal in the converging section; it slopes downwards in the throat and in the expanding
section, it is inclined upwards. The control section at which the depth is critical occurs near the
downstream end of the contraction.
4.8 Advanced Discharge Measurement Techniques
Conventional methods of flow measurement by the use of open channel sections or calibrated
structures are not always feasible because of channel size, unstable stage-discharge relationships,
or other physical, hydraulic, or environmental considerations (such as aesthetic, navigation and
fisheries impacts) and costs. Three relatively new methods of flow measurement in open
channels that have gained importance in recent times are:
 Dilution methods,
 Electromagnetic method, and
 Ultrasonic method

The first two are discussed here.

4.8.1 Dilution Methods for Measuring Discharge

Measurements with the current meter are difficult or impossible in boulder-strewn mountain
torrents. Chemical dilution method might be useful in such situations but it is restricted to those
streams where mixing occurs readily. Dilution techniques using chemical or fluorescent tracers
can be used in small and medium-sized mountain stream which do not have suitable reaches for
making a discharge measurement by using velocity-area methods or by constructing structures.
The dilution method of flow measurement, also known as the chemical method, is based on the
principle of continuity applied to a tracer which is allowed to completely mix with the flow.
Tracer material (like salt, fluorescent dye, radioactive material, or any easily measurable material
not present in the stream and not likely to be lost by chemical combination with materials in the
stream) may be used. Complete mixing of the tracer in the flow and accurate determination of
initial and final concentrations are essential.

A tracer is an ion or compound which is introduced into the flow to follow its behavior. A
known quantity of a tracer is introduced in the flow at an upstream section. At a downstream
cross-section of the reach, the concentration of the tracer is measured at regular intervals of time.

The reach selected for measurement should be such that there is no loss or gain of water
and the reach length should be sufficient to achieve complete mixing. The tracer can be
introduced in two ways: (1) constant rate injection, and (2) gulp injection. In the constant rate
injection method, the duration of injection should be such that a steady regime of concentration
is achieved for an adequate duration (about 10 to 15 minutes) in the sampling section.

In the constant injection rate, a tracer of concentration c1 is injected at the upstream end
of the stream reach at rate q. No flow should join the stream in the reach. Water samples are
taken at a downstream point and after an equilibrium concentration c2 is reached, the discharge Q
can be calculated as:
 c1 q + c0 Q = c2 (q +Q) (4.23)

c 2  c1
Q q (4.24)
c0  c2

c1 = concentration of the chemical dye in the constant rate injection

c0 = background concentration of the stream water upstream of the injection (often c0 = 0)
c2 = concentration of the chemical dye downstream (plateau value)
q = constant inflow rate
Q = discharge in the stream upstream of the injection.

In the integration method, a quantity of tracer of volume V and concentration C is added

to the stream. At the sampling station, the passage of the entire tracer cloud renders a
relationship between concentration and time. Common salt (NaCl) is frequently used as a tracer
and it can be detected with an error of 1% up to a concentration of 10 ppm (parts per million). A
cocktail of tracers may also be used.

Ideally the tracer used should not be absorbed by the sediments channel bank material
and vegetation. It should not chemically react with any of these, should not be lost by
evaporation and should be non-toxic. The tracer should not be very expensive and it should be
easy to detect it in small quantities.

4.8.2 Electromagnetic Method

The electromagnetic gauge operates on a principle similar to that of an electric dynamo. If a
conductor of certain length moves through a magnetic field, a voltage is generated between the
ends of the conductor. In the electromagnetic gauge, the magnetic field around the river is
generated by a coil, which is installed either above or below the river. The river is conductor in
this case which moves through the magnetic field. The movement of river water cuts the vertical
component of the Earth’s magnetic field and an electromotive force (EMF) is induced in the
water that can be measured by two electrodes. This EMF is proportional to the average velocity
in the river. The voltages generated are sensed by electrodes on either bank. The higher is the
water velocity, the greater is the voltage generated. Velocity in the entire cross section is
integrated to produce a mean velocity.
 Section 1
C1
Concentration

Section 2
C2

C0 Background concentration

Time 
Fig. 4.16 The concept of dilution method of discharge measurement with constant rate injection.

Figure 4.17 shows a sketch of an electromagnetic gauging station where the coil is
placed in the bed and the magnetic field is generated in the x direction, the EMF is in the y
direction, and the streamflow is in the z direction. Faraday’s law of electromagnetic induction
relates the length of a conductor moving in a magnetic field to the EMF generated. The
measures of induced probe voltage, coil current and depth are used to calculate flow as follows:

Flow = function of (probe voltage x depth x calibration constant/ magnetic field)

In practice, most river beds have significant electrical conductivity that will allow
electric currents to flow in the bed. From practical considerations, the induced field will be
spatially limited and electric currents flowing in the area outside the field will have the effect of
reducing the output potential. Both of the above factors have the effect of reducing the signal
and hence the voltage recorded. At an electromagnetic gauging station, it is necessary to
measure the conductivity of bed and water.
 Figure 4.17
7 Basic systeem of the eleectromagnettic method (S
Source: ISO 9213).

Earth's
E magnnetic field, battery
b effeccts of the prrobes in the water, and external eleectric
noise cau uses problem
ms in measurrements. To overcome thhese, the maagnetic fieldd produced bby the
coil is rev
versed at reg
gular intervaals. Measurem ments are taaken with thee field switchhed first onee way
and then the other.
A limitation of
o the electrromagnetic method
m is thhat it has to be calibratedd by referennce to
current meter
m gauginng or some other
o form off flow measuurement. Whhen the gaugge is construucted,
its perforrmance need ds to be veriified by com mparison witth the resultss of an alterrnative method of
flow gau uging. The electromagne
e etic river floow gauge esstimates thee mean velocity in the eentire
cross-secction but due to the assumptions an nd limitationns, the instrruments needd calibrationn and
adjustmeent.

Special site
s related consideratio
c ons
The folloowing are the
t special consideratio ons when seelecting the site for ann electromaggnetic
gauge:
a) Thee channel att the measurring section should be sstraight for at least threee times the river
widdth at the measuring
m seection, it sho
ould be sym mmetrical about the riveer centre linee and
preeferably of (n
nearly) trapeezoidal or recctangular shhape;
b) Thee upper limmit of channeel width useed is approxximately 30 m for sitess with below w-bed
meembrane. Th his restriction is based on construcction cost; thhe cost of constructionn will
inccrease signifiicantly with the size of thhe river;
c) Theere should be no majo or source ofo electricall interferencce near the site. Electrrified
raillways, electrrical power grid
g lines an nd radio transsmitters are the most likely sources;
d) Prooximity to heavily
h reinfforced concrrete and sheeet piling maay distort thhe magnetic field
andd should be avoided
a if poossible;
e) It is important to ensure that the membrane is particularly well protected against damage
by vessels moving in the river;
f) The conductivity of the watercourse being measured needs to be uniform throughout.

An electromagnetic gauge will only measure flows passing through the vertical magnetic
field generated by the coil. The velocity of any water passing outside the measuring reach will
register as negative, leading to errors in overall flow calculation. If full range flow calculation is
desired, careful selection of the site is required to ensure it is not by-passed. Training
embankments may be needed to contain the full range of flows to be measured.
Electromagnetic gauges are more suited to rivers where the flow range is relatively small
e.g., high base flows and not very high flows in monsoon. These gauges are particularly suited to
rivers in which considerable weed growth can occur; the gauge will also tolerate high levels of
suspended solids. Since this is a non-intrusive method of flow measurement, it is thus
particularly suited to measuring flows in channels with considerable fish movement.

Electromagnetic gauge are not suited to rivers where base flows can be very low relative
to the size of the channel. The uncertainties can be large when the depth of flow is very low. This
is because the offset, which is introduced in the calibration to allow for the difference between
bed level and the position of the coil, becomes highly significant relative to the depth of flow.
Thus, this method may not work well for many large Indian rivers or non-perennial rivers.

The accuracy of the electromagnetic method depends on the signal processing equipment
detecting and measuring small potentials sensed at the voltage probes. The electromagnetic
gauging station requires on-site calibration by current meter or other means and a relation
established between discharge and output.

4.9 Measurement of Discharge under Difficult Conditions

Harsh climatic condition (below freezing or very high temperatures), aridity high sediment
transport etc. pose spacing problems in river gauging. Weather conditions such as ice or frost can
impede the operation of the stage-sensing device or recorder thus introducing errors in the
observed data. Stilling well and inlet pipes need to be constructed in such a manner in these areas
that the system will remain in operation during periods of freezing temperatures.

Weed growth in a watercourse can change the hydraulic flow conditions in the measuring
reach, and hence the stage-discharge relationship. Seasonal vegetation development may make it
necessary to correct computed discharge data to follow the variation in the stage-discharge
relationship. This can be achieved by carrying out sufficient measurements to compute the
deviation. In some cases, the weed growth patterns may change in such a regular and repeatable
pattern that a seasonal stage-discharge relation may be defined.
 In the rivers that carry large volume of sediments, care has to be taken to prevent or
remedy significant siltation and choking of the inlet tube and/ or the stilling well itself. The risk
of siltation is very much dependent on local conditions. The stage-discharge relations of stations
operated in alluvial channels are subject to shifting-control conditions and typically more
frequent measurements are required to define a family of shift curves.

4.9.1 Special Problems in Streamflow Measurements in Arid and Semi-arid Regions

Special problems associated with the measurement of streamflow in arid and semi-arid regions
arise from the interaction of many climatic and geological factors. The degree of difficulty varies
from one region to another, depending on the combination of these factors. The main problems
associated with these regions are:

1. These regions mostly have inadequate infrastructure which makes movement and
communication difficult, time-consuming, unreliable, and expensive.
2. The harsh environment, dust, high soil erosion by strong winds, and the sediment carried
by rivers create problems for the conventional equipment. The end result is frequent
malfunctioning of equipment and loss of records.
3. The short duration and rapidity of onset of floods, coupled with (1), imply that the team
intending to measure such flows may miss them unless they are present at the site. The
duration of the peak flow may be so short that it is almost impossible to carry out
measurements.
4. Soft erodible beds make sounding operations difficult when scouring may occur beneath
the sounding weight as it touches the river bed.
5. The channel may shift laterally, isolating gauge and recorder well from the flow.
6. The bed level may vary due to scour and fill, particularly during the passage of a flood.
Such bed variations are difficult to measure and lead to errors in the application of cross-
sectional area rendering the resultant discharge value erroneous.

A detailed and careful planning of logistics, local knowledge, and experience are the key
factors in successfully tackling the river gauging problems of such regions. A combination of
conventional wisdom with new and appropriate technology by employing automation, telemetry,
and remote sensing is necessary to overcome such difficulties. Many equipment with minor
improvements could work well under arid and semi-arid conditions. Data logging systems, with
retrieval and telemetry have proved reliable under harsh environments.

4.9.2 Measurement of Streamflow under Ice Cover

If the Manning’s formula is used to compute discharge of a river, ‘n’ represents friction between
water and channel boundaries. In winters in cold regions, a sheet of ice forms on the channel
surface which imparts additional resistance to flow. The ice cover increases total friction by an
amount representing the difference between air and ice resistance. If the value of ‘n’ is derived
by considering the wetted perimeter as including the air and ice contacts as well as those of bed
and banks, and if the frictional resistance of the water-ice contact is greater than the frictional
resistance of the water-air contact, then ‘n’ will have a correspondingly greater value for
conditions of ice cover than for those of open channel.

A sizeable number of streamflow-gaging stations in the cold regions are affected by ice
during the winter. The procedure used by the U.S. Geological Survey for measuring discharge
under an ice cover involves averaging streamflow velocity at 0.2D and 0.8D, where D is the
effective depth of flow (measured vertically from the streambed to the underside of the ice
cover). Another approach is to use a coefficient to adjust point velocities at either 0.5D or 0.6D
to give the correct depth-averaged mean. A previous study to determine the streamflow
adjustment coefficients had found that there was considerable variability from station to station.
Hence, it is important that the coefficients for individual stations be determined.

On the basis of bias and root-mean-squared error, the first-visit complete-profile method
with velocity measured at 0.5D was recommended by Walker (1994) for implementation. In this
method, a complete set of vertical profiles for the first measurement of the winter season is used
to establish the adjustment coefficient for an individual site. Thereafter, the same adjustment
coefficient is used for subsequent measurements at that particular station and it was found to
result in nearly unbiased discharge measurements across the streamflow-gaging network.

4.10 Closure
The main objective of river gauging is to get as close as possible to the existing realities of the
field. The choice of measuring methods and the means of implementing them is a function of the
configuration of site, available resources, both human and material, and the degree of accuracy
expected. In difficult conditions, such as equipment breakdowns or dangerous situations,
estimations of speed, photographs taken, observing floodwater marks on permanent structures,
etc., are actions that can mitigate the absence of a true measurement and establish coherent limits
on further evaluation of a site. The selection and adaptation of gauging sites, competence of
personnel, equipment to be used for a job should all serve to advance a project toward optimum
measurement precision. The most important elements for good measurements are motivated,
well trained and properly equipped personnel. Not only the must the measuring equipment itself
be in good condition, but also the annex equipment must be in good working order.

REFERENCES
Boiten, W. (2008). Hydrometry, CRC Press, London.
Herschy, R.W. (2009). Streamflow Measurement, Routledge, London.
ISO 4369:1979. Measurement of liquid flow in open channels - Moving boat method.
International Organization for Standardization, Geneva.
ISO 6418:1985. Liquid flow measurement in open channels - Ultrasonic (Acoustic) velocity
meters. International Organization for Standardization, Geneva.
ISO 772:2011. Liquid flow measurement in open channels - Vocabulary and symbols.
International Organization for Standardization, Geneva.
ISO 748:2007. Liquid flow measurement in open channels - Velocity-area-method. International
Organization for Standardization, Geneva.
ISO 9213:2004. Measurement of total discharge in open channels -- Electromagnetic method
using a full-channel-width coil. International Organization for Standardization, Geneva.
Subramanya, K., Engineering Hydrology, Tata McGraw Hill Education Pvt. Ltd., New Delhi.
Walker, John F. (1994). Methods for Measuring Discharge under Ice Cover. Journal of
Hydraulic Engineering, ASCE, 120(11), 1327-1336.
WMO (2008). Guide to Hydrological Practices. Volume I: Hydrology – From Measurement to
Hydrological Information. WMO No. 168. World Meteorological Organization, Geneva.

Annexure A

4.A International Standard ISO 1100

The standard ISO 1100 gives guidelines for the establishment and operation of a gauging station
for the measurement of stage and/or discharge of a river or canal, reservoir, etc. Requirements
have been specified for measurement of stage and discharge in natural channels as well as for
stations with artificial structures. Some recommendations are given for measurements under
difficult conditions, e.g., under ice conditions. This section gives a summary of key
recommendations contained in ISO 1100.

4.A.1 General requirements of a gauging station

Before commencing work to establish and operate a gauging station, the following requirements
should be identified:
a) The range of levels and flows required to be measured;
b) The customer’s requirements for type and timeliness of data;
c) The allowable uncertainty in the results;
d) Other potential users of the data;
e) Life expectancy of the station and available budget.

In addition, other constraints should be identified including the following:

a) local environmental issues;
b) accessibility of the site under all conditions of flow and availability of electricity;
 c) stability of the watercourse embankments and bed;
d) information on any proposed hydraulic modifications planned for the future;
e) safety of the instruments and law and order issues;
f) influence of submergence of the gauging site due to downstream impounding structures;
g) aquatic weed growth in the watercourse.

Knowledge of the above requirements will ensure the provision of appropriate

measurement and recording facilities as well as an appropriate maintenance philosophy.

Water level (Stage) only gauging stations

The site selected for determination of stage should be selected according to the purpose for
which the readings are required. Accessibility of the site and the availability of an observer if the
gauge is to be non-recording are important criteria as is the availability of an appropriate power
supply and data communication capabilities if the gauge is to be recording.
Gauges on lakes and reservoirs are normally located near the outlet, but should be located
sufficiently far away from the zone where an increase in velocity causes a drawdown in water
level. Gauges on large bodies of water should also be located so as to reduce the effect of strong
winds which may cause misleading data.

Preliminary survey
A detailed examination of a large scale map of the area is required in the first instance; supported
by an aerial survey if necessary. This may be necessary if the area is not easily accessible by
other means. Aerial surveys and or satellite imagery can be used as a basis to select potential
sites. Enquiries should be made to determine if any plans exist for the modification of the river
reach which would modify the stream bed regime and impact the proposed gauging station.

Selection criteria
A list of potential sites shall be established with their advantages and disadvantages identified.
The establishment of the gauge zero shall be chosen so as to avoid negative readings. Thus it
shall be set well below the level of the control feature. This zero point shall be correlated with a
national datum through a station benchmark.

Stage measurement and recording

The reading of stage may be required as a single instantaneous measurement, as a short series of
instantaneous measurements or as continuous record of the fluctuations of stage. A vertical staff
gauge comprises a scale marked on or securely attached to a suitable and stable vertical surface.
Where the range of measurement required exceeds the capacity of a single vertical gauge, other
gauges should be installed on the line of a cross-section normal to the direction of flow. The
scales on such a series of stepped staff gauges should overlap by not less than 15 cm. A ramp or
inclined gauge consists of a scale marked on or securely attached to a suitable and stable inclined
surface, which conforms closely to the contour of the river bank.

Wire or tape weight gauge

A wire or tape weight gauge consists of a weight which is manually lowered until the weight
touches the surface of the water. The wire or tape may be wound on a drum attached to a
winding mechanism or it may be a hand reel. The wire or tape gauge can be equipped with
electrical contacts to improve measurements, when there is a large vertical distance between the
measuring point and the water surface.

Stage recording
Customer requirements shall dictate the method of recording stage. It may be that a single record
of stage taken daily and read manually will suffice. It is more common to provide a continuous
record of stage utilising water level sensors, such as floats, pressure transducers, and echo
sounders, interfaced with a digital recorder (logger or telemetry) or analogue recorder (chart).
When a recorder is used, visits by the observer should be made from time to time to ensure
satisfactory performance of the sensor and recorder.

Stage-discharge gauging stations

When records of water level are to be used as a basis for computation of discharge, the relation
between water level and flow must be determined. In a stable channel with an appropriate control
feature which is stable and sensitive, a single relation may exist between water level and
discharge. In this case, the relation can be determined by taking discharge measurements
throughout the range of levels and flows required to be measured.

Control section or control reach

A control section or control reach of a channel is a natural or artificial section or reach whose
physical characteristics can be measured and used to determine the relationship between stage
and discharge. In a control section any change in the stage downstream of the control does not
affect the stage upstream of the control. Whatever the discharge in the control section, a critical
stage can be determined.

Discharge measurements
Discharge measurements using the above techniques shall be related to a stage reading taken at
the beginning and end of the discharge measurement and during the measurement if the stage is
changing rapidly or inconsistently. When sufficient numbers of discharge measurements have
been taken, a stage discharge relationship can be computed (see ISO 1100-2). Subsequent to the
formulation of this stage-discharge relationship, only occasional discharge measurements need
be taken at flows in the normal range to confirm the robustness of the relationship unless the site
is subject to shifting control conditions.
 Discharge measurements made using the velocity-area methods can be performed using
rotating element current meters, electromagnetic current meters, acoustic Doppler velocimeters,
or acoustic Doppler current profilers. A variation of velocity area methods is the slope-area
method, which is typically used to compute flood discharge indirectly (see ISO 1070).
 Module 5
Measurement and Processing of Meteorological Data

5.1 Evaporation and Evapotranspiration

5.1.1 Measurement of Evaporation
5.1.2 Pan Evaporimeters
5.1.3 Processing of Pan Evaporation Data
5.1.4 Estimation of Lake Evaporation from Pan Measurements
5.1.5 Estimation of Evapotranspiration
5.1.6 Remote Sensing Measurements of Evaporation and Evapotranspiration
5.2 Temperature Data
5.2.1 Processing of Temperature Data
5.3 Relative Humidity
5.3.1 Processing of Relative Humidity Data
5.4 Measurement of Wind Speed
5.5 Sunshine Duration
5.6 Closure
References

Keywords: Evaporation, Evapotranspiration, Pan Evaporimeters, Temperature Data, Relative

Humidity, Wind Speed, Sunshine Duration

The objectives of this module are:

• to discuss techniques of acquisition, validation, and processing of evaporation and
evapotranspiration data;
• to discuss measurement of other meteorological data used in hydrology, namely,
temperature relative humidity, wind speed, and sunshine duration.

Besides precipitation, the other hydro-meteorological variables, such as evaporation,

temperature, humidity, wind speed and direction, and sunshine hours, are also important in
studies relating to water resources development and management. The acquisition and processing
of these data is discussed in this module.

5.1 EVAPORATION AND EVAPOTRANSPIRATION

Evaporation and transpiration constitute the primary abstractions of water from the hydrologic
cycle. It will be helpful to define these first. Evaporation is the process by which water is
changed from the liquid or solid state into the gaseous state due to the involvement of heat
energy. Evaporation is an important component of the hydrologic cycle, since a large fraction of
the precipitation is returned to the atmosphere by this process. Transpiration is a natural process
occurring in plants in which water is picked up from the soil moisture storage by the plant roots.
This water is used in photosynthesis and it is finally evaporated from stomata which are the pores
found in the leaf and stem epidermis of the plants and are used for gas exchange.
 Evaporation takes place mainly due to heat energy and plants play the main role in
transpirations. The combined effect of evaporation and transpiration is termed as
evapotranspiration (ET). In hot climates, the transfer of water to atmosphere by evaporation from
rivers, canals and surface-water storages (reservoirs & ponds) is of interest to hydrologists since
evaporation is a significant proportion of all water supplies. It is significant and is concern in the
sense that most of the water withdrawn from the sources for beneficial uses ultimately returns to
streams and aquifers and becomes available for reuse, while the loss of water due to evaporation
is entirely lost from the usable supply, at least for considerable time.
Over large land areas in India, about two thirds of the annual precipitation is lost to
atmosphere through evapotranspiration. In arid regions of India, evapotranspiration may be even
more significant and up to 90 per cent of the annual rainfall may be lost to the atmosphere.
Problem of evaporation is more serious in periods of droughts when aridity is high. Storage
reservoirs expose large water surfaces to direct radiation which causes evaporation and thus large
quantity of water may be lost, particularly during hot and dry season. During flood season,
reservoirs may lessen evaporation because water is confined to deep storages rather than thinly
spreading over wide flood plains.

The meteorological factors controlling evaporation are air and water temperature, wind
speed, atmospheric pressure, incoming solar radiation, humidity, and saturation vapor pressure
deficit. In addition, water quality, depth, soil type and nature also influence evaporation.
Meteorological factors that affect transpiration are essentially the same as for evaporation. In
addition, vapour pressure gradient, available soil moisture and plant properties also affect
transpiration.

There is an important difference between ET and free surface evaporation. Transpiration

is associated with plant growth (trees in a mature forest may grow slowly) and hence ET takes
place only when the plant is growing. Thus transpiration has diurnal and seasonal variations.
Consequently, ET will have seasonal variations. Evaporation also has large seasonal variations
due to changes in radiation received and in other meteorological inputs.

5.1.1 Measurement of Evaporation

Selection of a method for estimation of evaporation depends on the study area and its properties.
Evapotranspiration from small water bodies and soil can be estimated with reasonable accuracy.
However, direct measurement of evaporation and evapotranspiration from large water bodies or
catchments is not possible with present technology. For this purpose, indirect methods are
applied and they give acceptable results. Evaporation pans and lysimeters are used for this
purpose. For existing reservoirs and plots or small catchments, estimates can be made by water
budget, energy-budget, and aerodynamic approaches. These latter techniques are discussed only
from the point of view of instruments and observational requirements.
5.1.2 Pan
P Evaporiimeters
To estim
mate evaporaation from openo water bodies,
b evapporation of wwater from small pans filled
with water is measurred and it is assumed thaat evaporatioon from lakee and pan arre closely rellated.
The shappe of the panns could be either
e squaree or circular;; it may be m
mounted aboove the grouund or
sunk in the
t ground so s that the water
w i approximaately that off the groundd. In a large lake,
level is
pans may y be mounted d on floating
g platforms anchored
a at appropriate locations.
Two
T types off pans are use most com mmonly: thee United Staates Weatherr Bureau Claass A
pan and the Russian GGI-3000 pan; a devicce known ass the 20 m2 tank is alsoo used in Ruussian
Federatioon. The U.S.. Weather Bureau
B Class A pan is w widely used ththroughout thhe world annd has
been recommended by b WMO an nd the Interrnational Asssociation off Hydrologiccal Sciencess as a
referencee instrumentt. Performan nce of this pan has b een studiedd under a raange of clim matic
2
conditionns and wide limits of lattitude and elevation. Thee GGI-3000 pan and 20 m tank aree used
in the Ruussian Federation and some
s other countries w with differentt climatic coonditions ass they
possess reliable
r operrational quallities and a stable
s relatioonship with the meteoroological elemments
that influ
uence evapo oration (WM MO 2008). World
W over, the US Class A pan iis most com mmon
followedd by GGI 30 000 pan. WM MO sponsorred comparaative observvations (WM MO, 1976) oof the
2
Class A pan,
p the GG GI-3000 pan and the 20 m tank in seeveral counttries, which eventually lled to
some op perational reecommendations on thee suitability of these pans in diveerse climaticc and
physiographic condittions.

Inn India, the US

U Weather Bureau Class A pan is tthe standardd tool to meaasure evaporaation.
It is a cirrcular pan, 1.22 m in diaameter and 0.255
0 m deepp. The pan iss mounted on a white paainted
wooden stands (Fig. 5.1)
5 and is covered by a wire mesh to prevent looss of waterr due to birdds and
animals. The inner base b and sid
des of the paan are painteed white. A stilling well is placed iin the
pan and inside it is a pointer gauge. Measurrement mustt take accouunt of evapooration losses and
also gainns due to rain nfall.

Fig. 5.1 U.S. Weatheer Bureau cllass A land pan.

p
 The major difficulty in using a Class A pan for the direct measurement of evaporation
arises because of the use of coefficients to convert the measurements from a small tank to large
bodies of open water.

When installing evaporation pans it is important to ensure that the site of the pan is level
and free of obstruction. To the extent feasible, the ground cover should be maintained as close as
possible to the natural cover common in the area. Obstructions such as trees, buildings, shrubs or
instrument shelters should not be closer than four times the height of the object. The pan should
never be placed on a concrete slab, asphalt floor or gravel. The pan should be installed in such a
way that no object casts any shadow over the pan at any time. The recommended minimum size
of the plot where pan is installed should be 15 m x 20 m. The plot should be fenced to protect the
instruments and to prevent birds and animals from drinking the pan water. The fence should be
constructed so that it does not affect the wind structure over the pan. A barbed wire or iron mesh
fence is preferred. It may be necessary to prevent birds and small animals from drinking water by
using chemical repellants (these should not pollute water) in addition to a wire mesh. To estimate
the error introduced by the wire-mesh screen on the wind field and thermal characteristics of the
pan, readings from the protected pan should be compared with those of a standard pan at a
nearby site.

It is a good practice to install additional instruments along with a pan, such as an

anemometer to measure wind movement over the pan, a precipitation gauge, and thermometers
to measure temperature of pan water and surrounding air. The inner side of the pan is painted
white. The water level in the pan changes due to evaporation and rainfall.

On days without rain, at daily (or twice-daily) reading time, water is poured into the pan
using a graduated cylinder to bring the level precisely to the top of the pointer gauge. The
volume of water added is recorded and represents the depth of evaporation. On the days when
rain has fallen since the last observation, the rainfall may exceed evaporation and water may
have to be removed from the pan to bring the level to the hook level. The adjacent rain gauge is
used to measure the rainfall. It there is a forecast of heavy rainfall on a given day, a measured
amount of water may be removed from the pan in advance so as to avoid pan overflow.
The water level in the pan must be measured accurately before and after water is added.
The water level may be determined by means of a hook gauge consisting of a movable scale and
vernier fitted with a hook, enclosed in a stilling well in the pan. An alternative arrangement is to
use a float. A calibrated container is used to add or remove water at each observation so as to
maintain the water level to a pre-specified depth.

The amount of evaporation (E in mm) between two successive observations is the

difference in water level in the pan. It is corrected for any precipitation during the period:
 E = P  d (5.1)

where P is the depth of precipitation during the period between the two measurements, and d is
the depth of water added (+) to or removed (–) from the pan.

Pan evaporation provides an estimate of open water evaporation. Evaporation data from
pans are frequently used to estimate evaporation from water bodies, such as lakes and reservoirs
and evapotranspiration from an area. But a pan is a really small in size and its material is
different than a natural body. Hence its heat storage characteristics and air dynamics will be quite
different than a large water body. Therefore, evaporation from a pan will be higher than a large
open water body. An estimate of lake or reservoir evaporation (ER) is obtained by multiplying
the pan evaporation by a coefficient which is called the pan coefficient:

ER = KP Epan (5.2)

where KP is the pan coefficient; ER is the depth of evaporation from the reservoir and Epan is the
pan evaporation, both in mm/day. The value of pan coefficient depends on climate, geographical
location, season, size, and depth of the water body. This coefficient generally varies from 0.6 to
0.8. Lower values are typical of dry seasons and arid climates while higher values are
appropriate for humid climates. In the absence of better estimates, a value of 0.7 for the pan
coefficient may be used.

Some automatic evaporation pans are also in use. In these pans, water level is kept
constant by automatically releasing water into the pan from a storage tank or by removing water
from the pan in the case of precipitation. A record of the amount of water added to or removed
from the pan is maintained.

5.1.3 Processing of Pan Evaporation Data

Like any other instrument and measurement, errors may arise in pan observations. Errors may
crop up due to observer’s mistakes and instrument errors. Leakage from sides or bottom,
particularly from old pans is not uncommon.
Errors in the data may be detected as follows:

 The maximum value of evaporation in the region is used as the first check to screen
doubtful measurements.
 The observer may over-or under-fill the pan on a particular day. Such values may be
compensated for on the next day but the value for the current day will be wrong.
 Among the instrument errors, leakage is the most serious problem and it occurs usually at
the joint between the pan base and side wall. Small leaks are difficult to detect but may
have a significant systematic effect on evaporation totals over extended period such as a
 mouth. Where leakage has been detected, the pan evaporation records for a period of a
few months preceding the detection of leakage should be inspected carefully and marked
as suspect.
 Animals and birds may be able to drink water from the pan especially if the covering wire
mesh is damaged.
 Algae and dirt in the water will reduce the rate of evaporation.
 Errors arise in periods of high rainfall when the depth caught by the rain gauge is
different from the rain fallen in the pan as a result of splash or wind eddies.

To detect error in pan evaporation data which may arise due to the above reasons,
warning and maximum limits on evaporation are fixed. Pan evaporation measurements are less
accurate during the periods of heavy rains but during these periods the rate of evaporation is also
less. Excessively high evaporation values during days should be flagged as doubtful and should
be subject to further validation.

5.1.4 Estimation of Lake Evaporation from Pan Measurements

The evaporation from pans installed above the ground is influenced by the characteristics of the
pan. Sunken pans are subject to undetected leaks, accumulation of biomass and debris on the
water surface, and boundary conditions which are different from those of a large lake. Pans
exposed above the ground are subject to heat exchange through the sides and to other effects that
do not occur in lakes. Floating pans are subject to splash-in and splash-out of water, and are
costly to install and operate. Pans have much less heat storage than lakes and experience a
different annual cycle of evaporation, with pan-evaporation extremes occurring earlier than the
natural water body. Reliable estimates of annual lake evaporation can be obtained by multiplying
the annual pan evaporation by the appropriate pan-to-lake coefficient. The pan-to-lake
coefficient for a particular pan can be determined by comparison with actual lake evaporation or
with a large pan sunken to simulate a lake.
Variation of the pan coefficient with climate is determined by comparisons of
evaporation from large and small pans under different field conditions. On the basis of the
experiments conducted in Russia and other countries the pan coefficients were found to range
between 0.70 and 1.0.
Another method is the adjustment of the pan evaporation for heat gain or loss through the
sides and bottom. An example of this method is the technique in estimating evaporation by using
data from the Class A evaporation pan. In humid seasons and climates, the pan water
temperature is higher than the air temperature, and the pan coefficient may be 0.80 or higher. In
dry seasons and arid areas, the pan water temperature is less than air temperature, and the
coefficient may be 0.60 or less. A coefficient of 0.70 is assumed to be applicable when water and
air temperatures are equal.
5.1.5 Estimation of Evapotranspiration
Evapotranspiration from agricultural areas can be directly measured by installing a lysimeter.
Lysimeters are square or circular boxes, of area around 10 m2 and depth of several meters. These
boxes are installed in an area, say agricultural field and filled with the soil and crops, same as the
surrounding area, and flush with ground. These instruments are simple and, if all requirements
concerning their installation and observational techniques are fulfilled, are also quite accurate.

There is no single standard instrument for measuring evapotranspiration. General

requirements for the location of evaporation plots are as follows:
(a) The site selected for the plot should be typical of the surrounding area with respect to
irrigation, soil characteristics (texture, layering, genetical type), slope and vegetative
cover;
(b) The evaporation plot should be located beyond the zone of influence of individual
buildings and trees. It should be situated at a distance not less than 100 to 150 m from the
boundaries of the field and not more than 3 to 4 km from the meteorological station.
Lysimeters should be filled by soil taken from within a radius of 50 m of the plot, and the
soil and vegetative cover of the lysimeter should correspond to those of the plot.

Lysimeters are categorized in three groups according to their method of operation:

a) Weight based, which use mechanical scales to measure changes in water content;
b) Hydraulic based, which use the hydrostatic principle of weighing;
c) Volumetric based, in which water content is kept constant and evapotranspiration is
measured by the amount of water added or removed.

A weighing type lysimeter uses mechanical balance to determine the change in water
content of the control volume. The hydraulics-based equipment employs hydrostatic principles of
weighing, and in the volumetric based ones, ET is measured by the amount of water added or
removed from the control volume to keep constant water content. Note that lysimeters are
difficult and costly to install and maintain. In view of the difficulties in direct measurement of
ET, indirect methods are generally used to estimate it. This requires measurements of
meteorological variables which influence evaporation. Commonly, pan evaporation is multiplied
by a coefficient to get the crop ET.
Evapotranspiration can be estimated by the water budget or heat-budget methods; many
empirical formulae have been developed which are based on meteorological data. Food and
Agricultural Organization (FAO) of the United Nations have adopted the Penman-Monteith
(PM) equation as the standard technique to compute reference ET (Allen et al. 1998).

5.1.6 Remote Sensing Measurements of Evaporation and Evapotranspiration

Remote-sensing observations combined with ancillary meteorological data have been used in
obtaining indirect estimates of ET over a range of temporal and spatial scales (Schulz and
Engman, 2000). Recently there has been considerable progress in the remote-sensing of
parameters, including:
(a) Incoming solar radiation;
(b) Surface albedo;
(c) Vegetative cover;
(d) Surface temperature;
(e) Surface soil moisture.

Remote-sensing of several important parameters used to estimate evaporation is made by

measuring the electromagnetic radiation in a particular waveband reflected or emitted from the
Earth’s surface. Furthermore, remote-sensing has a potentially important role because of its areal
coverage in the spatial extrapolation process of ET.

Researchers are experimenting with satellite data (see, Bastiaanssen and others, 1998) to
estimate regional actual evapotranspiration. Remote-sensing of several important parameters
used to estimate ET is made by measuring the electromagnetic radiation in a particular waveband
reflected or emitted from the Earth’s surface. The soil moisture may be estimated using the
measurement of microwave properties of the soil (microwave emission and reflection or
backscatter from soil). However, there are uncertainties in such soil moisture estimates due to
previously mentioned factors such as surface roughness and vegetative cover.

Direct measurement of evapotranspiration by the eddy covariance is being attempted in

many countries. Recently in India, such towers are being established by CSIR and ISRO. Under
CSIR, 20 towers (termed as meteorological profilers for climate observation and modeling) have
been setup across India and are in operation. Similarly ISRO has installed 20 micromet towers
across various locations in India and these are in operation. It is high time that several more such
towers are established across India for a better assessment of actual evapotranspiration. The
satellite based approaches for estimating evapotranspiration are gaining interest in several parts
of the world. Using satellite products it is feasible to estimate the evapotranpiration at scales of
few sq. km. Hence by establishing a good ground network of observations will facilitate the
calibration and validation of satellite estimates.

5.2 Temperature Data

Temperature is a measure of the ability of the atmosphere and water to receive and transfer heat
from other bodies. Temperature of a water body is also an indicator of its quality, since it
influences the amount of dissolved gases and the rate of chemical and biological reaction and
activities. Since temperature affects evapotranspiration, snowmelt, and process associated with
changes in water quality, it is needed in many hydrologic studies. Temperature of air, snow pack,
soil, and water in rivers and reservoirs is of interest in these studies.

Temperature of a place depends on its geographical location, latitude (which determines

incoming solar radiation), altitude, time of the year and proximity to oceans. Temperature varies
primarily with the magnitude of solar radiation received at a place and follows diurnal and
seasonal cycles. It is influenced, at particular times by the exchange of air masses and by
cloudiness (which obstructs incoming radiation). Temperature decreases with altitude; normal
rates are approximately 0.6oC per 100 meters per 100 meters for moist air and 0.9oC for dry air.
The temperature of places near a large water body, such as sea, is moderated by its influence so
that the annual and diurnal range is smaller. Generally, temperatures at nearby places are
strongly correlated.

Temperature is periodically observed (typically once or twice daily) at standard times

using a set of four thermometers, located in an installation known as Stevenson screen (Fig. 5.2),
which provides a standard condition of ventilation and shade. The screen should face towards
north. A continuous record of temperature is obtained using a thermograph in which changes in
the temperature are recorded on a clock-driven revolving chart.
Manual observations of air temperature are made using thermometers. Four types of
thermometers: dry bulb, wet bulb, maximum, and minimum thermometers are used as described
in Table 5.1.

Table 5.1: Types of thermometers used to obtain temperature data

Type of Purpose
thermometer
Dry bulb To measure ambient air temperature.
thermometer
Wet bulb To measure temperature that will be attained by a volume of air if cooled
thermometer adiabatically to saturation by evaporation of water into it; all latent heat
has been supplied by the volume of air. This temperature is used to
calculate relative humidity.
Maximum To indicate the highest temperature reached since the last setting.
thermometer
Minimum To indicate the lowest temperature reached since the last setting.
thermometer

Two units are commonly followed in measuring and reporting temperature: Centigrade
and Fahrenheit. In India, centigrade is used more commonly. In the case of the dry bulb, wet
bulb and maximum thermometers, observations are of the position of the end of the mercury
column but in the case of the minimum thermometer, the reading is taken of the position of the
end of th
he dumb-belll shaped inndex farthestt from the bbulb. After tthe maximuum and miniimum
thermom
meters have been
b read, theey are reset using
u a standdard proceduure.

Fig. 5.2 Therrmometers in

nstalled in th
he Stevensonn screen.

The
T wet bulb b is so called
d because th he bulb of thhe thermomeeter is alwayys kept wet. Wet-
bulb temmperature is measured
m by
y using a theermometer w whose bulb is wrapped in cloth (callled a
sock). Th he sock is kept
k wet with h water via wicking or capillary acction. One ennd of the soock is
immersed d in a small cup of wateer and the otther surroun ds the bulb. It is to be eensured that there
is alwayss enough waater in the cu up so that thee sock is keppt wet. Wet-bbulb thermoometers shouuld be
shielded from radiattion as welll as adequatte ventilatioon must be maintained. An accuraccy of
0.3C is enough in most
m temperaature related applicationss.
Regarding
R th
he use of th hermometers, the naturee of intendeed use will determine w which
thermom meter is most suitable. Daata of the dry y bulb thermmometer is thhe most frequuently used input
in hydrologic studies and obserrvations of ambient
a air temperaturee are best caarried out w
with a
mercury--in-glass theermometer. Continuouss records oof temperatuure may bee obtained with
resistance or therm mocouple ellements. Fo or measurinng and reccording waater temperaature,
thermogrraphs, which h produce a continuou us record oof temperatuure are usedd. Usually these
comprisee of a mercu ury-in-steel sensing
s elem
ment immerssed in the waater, which is connectedd to a
recorder (WMO 200 08). Care sho ould be takeen in the insstallation of thermograpphs to ensuree that
measurem ments taken are represen ntative of th
he water tem mperature. Inn case temperature of thhe air
above a water body is needed, it i should bee measured aat 2 m abovve the waterr surface neaar the
centre off the water boody.

Temperature
T is also meaasured contiinuously usiing a thermoograph in w which changges in
temperatuure are recorrded throughh the use of a bi-metallicc strip. The ttemperature is registeredd on a
chart on a clock-driv ven revolvinng drum and d the measurrement (chaart) period m may be eitheer one
day or onne week. Th he observer extracts temmperatures at a selected interval froom the chartt. The
manuallyy observed reading on the dry bu ulb thermom meter is meeasured and recorded aat the
beginning and end of o the chart period and if these diff ffer from thee chart valuee, a correctiion is
applied to
o the chart readings at th
he selected in
nterval.
 To measure water surface temperature, several types of thermometers, such as mercury-
in-glass or mercury-in-steel (including maximum and minimum and reversing thermometer),
platinum resistance or thermistor elements with electronic circuit and meter or recorder and
thermocouple thermometers, are used. In general, the precision required for the measurement of
water temperature is ±0.1°C. However, in many circumstances precision of observation of
±0.5°C is adequate.

5.2.1 Processing of Temperature Data

Processing and validation of temperature data is based on location and site conditions and
comparison of observation between nearby stations. Errors in the temperature data may arise due
to many reasons such as:
 Observer commits error in reading the thermometer. Commonly he may make error of about
1oC or less which is difficult to detect. Sometimes error may be large. Reading errors are
common in old thermometers with faint graduation etchings.
 In the wet bulb thermometer an error will occur if the muslin and wick of the wet bulb are
not adequately saturated. There will also be an error if the muslin becomes dirty or is
covered by some oily substance. These defects will give a high reading of wet bulb
temperature.
 Error may arise in recording the thermometer reading.
 Thermometer fault which results in systematic errors in temperature:
 Breaks in the mercury thread of the dry, wet or maximum thermometer,
 Failure of constriction of the maximum thermometer,
 Break in the spirit column of minimum thermometer or spirit lodged at the top or bubble
in the bulb,
 Thermograph out of calibration and no correction made.

Thermometer faults usually result in persistent systematic errors in data. Most of the
common errors can be detected by setting up appropriate maximum, minimum, and warning
limits for a particular station and these may vary seasonally. For example, summer maximum
temperature at a place can be expected not to exceed 50°C and the winter maximum temperature
not to exceed 35°C. Of course, these limits will change with location. Other possible checks
include:
 Dry bulb temperature should be greater than the wet bulb temperature. On rare occasion
they may be equal.
 Maximum temperature should be several degrees more than minimum temperature.
 Maximum temperature measured using the maximum thermometer should be greater than
or equal to the maximum temperature recorded by the dry bulb during the interval. The
value of the maximum temperature will be set to the observed maximum on the dry bulb
if this is greater.
  Minimum temperature measured using the minimum thermometer should be less than or
equal to the minimum temperature recorded by the dry bulb during the interval. The
value of the minimum will be set to the observed minimum on the dry bulb if this is
lower.

5.3 Relative Humidity

The term that is commonly used in describing humidity is relative humidity (RH) which can be
defined as the ratio (in %) of mixing ratio to the saturation mixing ratio:

RH = 100 r/rs (5.3)

where the mixing ratio r = mv/md. Here, mv is the mass of water vapor and md is the mass of dry
air. The saturation mixing ratio rs is the ratio of the mass of water vapor in a volume of the
saturated air to the mass of dry air. The relative humidity (%) can be defined as the ratio of the
actual vapor pressure of the air to the saturation vapor pressure at the same pressure and
temperature. The relative humidity does not vary rapidly with time. Places close to sea have
higher RH (due to availability of moisture) and a smaller daily variation than inland locations.

RH is most commonly assessed by the joint measurement of dry bulb and wet bulb
temperatures (see section 5.2). From these measurements, the dew point temperature, and actual
and saturated vapor pressures may also be calculated. While the actual vapor pressure may vary
little during the day (except with the incursion of a new air mass), RH has a regular diurnal
pattern with a minimum normally coinciding with the highest temperature (when the saturation
vapor pressure is at its highest). It also shows a regular seasonal variation. RH is calculated from
the wet bulb depression (difference between dry and wet bulb readings) using a set of tables.

Relative humidity may also be measured continuously by means of hygrograph in which

the sensor is human/horse hair whose length varies with relative humidity. The values of
humidity are registered on a chart placed on drum which revolves by a clock-driven mechanism
and the measurement (chart) period may be either one day or one week. The observer extracts
humidity at a selected interval from the chart. A manually computed reading from dry and wet
bulb thermometers is recorded at the beginning and end of the chart period and if these differ
from the chart value, a correction is applied to the corresponding chart readings.

Measurement errors for the dry and wet bulb thermometers are the same as for
temperature. If the wet bulb is not properly maintained, it will show a high value of the
temperature from the wet bulb thermometer and consequently RH will be estimated higher.
Errors in the hygrograph may arise from poor calibration.

5.3.1 Processing of Relative Humidity Data

During processing
p of
o RH data, temperaturre data shouuld be checkked carefullly. The dry bulb
temperatuure should be
b greater thhan or (rarelly) equal to the wet bullb temperatuure. Errors inn RH
may be detected
d by setting up upper and lower
l warniing limits apppropriate too the stationn and
season. The
T maximu um value forr RH is 100 0%. Daily seeries shouldd be graphiccally inspectted to
identify any
a anomalo ous values.

5.4 Meassurement off Wind Speeed

Wind speeds at a place
p are conntrolled by pressure grradients whiich depend on temperattures,
topographic features, land use, and
a water bo ody, etc. Speeed of windd shows widee variations from
place to place
p as well as with tim
me.

Wind
W speed is measured by using a cup typee anemomeeter (Fig. 5..3). Cups oof the
anemomeeter rotate due
d to the prressure exertted by windd movement.. The numbeer of rotatioons of
the anemmometer overr a time inteerval is displlayed by a ccounter and oor recorded by a data loogger.
This nummber indicatees the averag ge wind speeed over the time intervaal. Standard practice in India
is to meeasure the wind speed d over a th hree-minute period andd this is coonsidered ass the
instantan
neous wind speed at th hat time. Wind
W speed ffor the day is calculatted from coounter
readings on successive days. Observations
O are made daily in thee morning oor twice-daiily in
morning and evenin ng. Wind sp peed measurrements may ay be instanntaneous; theese may alsso be
accumulaative. By noting the cou unter reading
g at the beginnning and ennd of a periood, average wind
speed forr the time intterval can bee determined
d by dividingg this value bby the time iinterval.

Fig. 5.3 Cup type anemo

ometer

The
T direction n of wind is reported
r as 16
1 points of the compasss; it could bee shown eithher as
numericaal or as alph
habetic code. Note that wind
w directiion is not a variable in formulas ussed to
compute evapotransp piration. Thee wind directtion may inffluence evapporation at a place if the wind
is coming
g from a place whose hu umidity is sig
gnificantly ddifferent.

Typical measureme
m ent errors
Errors in
n wind speed
d might arisee as the obseerver may wrrongly read and report tthe reading oof the
counter. Fault may develop
d in th
he instrumennt due to weear and tearr or poor maaintenance. W Wind
speed maay have largee variation with
w time and d distances aand hence itt is difficult tto set up lim
mits to
screen su
uspected valuues. Wind sp peeds in exccess of 100 kkm per hourr should be cchecked careefully
with otheer meteorolo
ogical variables.

5.5 Sunshine Duratiion

The data of sunshinee duration is an input in estimation oof evapotrannspiration ass well as in w
water
quality studies. The potential maximum
m sunnshine durattion dependss on latitudee and seasonn; the
unshine hourrs vary due to clouds, fog, etc. The aamount of bright sunshinne in urban areas
actual su
may be reduced by attmospheric pollution
p and
d smog.

The
T instrumeent commonlly employed d in India forr observatioon of the sunnshine duratiion is
the Camp pbell Stokess sunshine reecorder (Fig. 5.4). It is a glass spherre mounted on a sectionn of a
sphericall bowl. The sphere focu uses sun’s rays on a caard held in tthe grooves of the bow wl and
graduated d in hours. The sun’s rays burn th he card dayy when the sun is shinning. The caard is
changed daily after sunset. Note that, the sunnshine recordder uses the movement oof the sun innstead
of a clocck to form th
he time basiis of the reccord. Differeent grooves in the bowl must be ussed in
winter suummer and the
t equinoxees, taking diffferent card types. The llengths of buurnt traces oon the
sunshine card indicaate the sunsh hine duratioon. Sunshinee duration ddata at requiired resolutiion is
tabulated
d from the caard.
The
T use of th he sun as a timing
t devicce avoids tim ming errors. Errors mayy arise in thee data
from the wrong placcement of ch hart. The errrors may alsso creep in while the obbserver notees the
duration of sunshine from the chaart. The valuues below zeero or greateer than the m
maximum posssible
sunshine hours for th
he location shhould be lag
gged for furthher checkingg.

Fig. 5.4 Sunshine ho

ours recorderr.

Typical measuremen
m nt errors
The sunsshine recordeer is simple instrument in
i principle aand the use oof the sun raather than a clock
as a timee base avoid
ds timing errrors. Potential errors maay arise from
m the use off the wrong chart
which may
m result in the bum reaaching the edgee of the chart, beyonnd which it is not registtered.
Mistake may be made by the observer while extraction of information from the chart.

Error detection checks for sunshine data include:

 Hourly sunshine values greater than 1.0 or less than 0.0 are not permitted.
 Sunshine hours before 0500 and after 1900 are rejected and hence daily sunshine hours
greater than 14 hours are marked as suspect.
Daily warning limits may be set seasonally based on the maximum possible sunshine for the
location and time of year.

5.6 Closure
This module discusses the methods of observation and processing of data pertaining to several
other variables which are used in hydrological studies. Although data of these variables are
important, unfortunately the network to measure these variables is very weak in our country.
Thus there is a need to strengthen the network to observe these variables and create a strong
database.

With a view to Standardize the observation of meteorological data, WMO (1988) have
issued a document listing recommended practices for meteorological observations.

References
Allen, R.G., Pereira, L.S., Raes, D., and Smith, M. (1998). Crop Evapotranspiration, Irrigation
and Drainage Paper No. 56, Food and Agriculture Organization, Rome, Italy.
Bastiaanssen, W.G.M. (1998). Remote sensing in water resources management: The state of the
art. International Water Management Institute, Colombo, Sri Lanka.
Engman, E.T. and R.J. Gurney (1991). Remote Sensing in Hydrology, Chapman and Hall, London.
Schultz, G.A. (1993). Application of GIS and remote sensing in hydrology. HydroGIS 93:
Application of Geographic Information Systems in Hydrology and Water Resources. IAHS
Pub. No. 211, 127-140.
WMO (1988). General Meteorological Standards and Recommended Practices. WMO - No. 49,
World Meteorological Organization, Geneva, Switzerland.
WMO (2008). Guide to Hydrological Practices. Volume I: Hydrology – From Measurement to
Hydrological Information. WMO No. 168. World Meteorological Organization, Geneva.
 Module 6
Measurement and Processing of Water Quality Data

6.1 Basics of Water Quality

6.1.1 Water Quality Data
6.2 Water quality indicators
6.2.1 Physical Indicators
6.2.2 Chemical Indicators
6.2.3 Biological Indicators
6.3 Types of Pollutants
6.4 Sampling Surface Waters for Quality Tests
6.4.1 Samplers and Sample Containers
6.4.2 Sampling Labelling and Preservation
6.5 Laboratory Analysis of Water Quality Samples
6.5.1 Determination of Biochemical Oxygen Demand
6.5.2 The BOD Changes
6.5.3 Inferences Concerning BOD
6.5.4 Factors Affecting the Oxygen Demand Rate
6.5.5 Chemical Oxygen Demand Test
6.6 Sediment Related Data
6.6.1 Processing of Sediment Data
6.6.2 Sediment Rating Curve
6.6.3 Processing of Bed Material Data
6.7 Water Quality Standards
6.7.1 Water Quality Index
6.8 Water Quality Monitoring in India
6.9 References

Keywords: Water Quality, indicators, Physical, Chemical, Indicators, Sampling, Biochemical

Oxygen Demand, Sediment, Standards, Index

The objectives of this module are:

• to discuss techniques of acquisition, validation, and processing of water quality data;
• to discuss about the processing of sediment related data,
• to describe water quality standards and index in India.

6.1 Basics of Water Quality

Clean water is a clear, colorless, and odorless substance. The term ‘water quality’ denotes the
physical, chemical and biological characteristics of water. Water quality variables show the
status of water vis-à-vis the requirements for needs such as drinking, industrial, agricultural,
energy generation, ecological, etc. This term is most commonly used with reference to a set of
standards; in general “water of good quality” meets the desired standards. The physical
parameters of quality include turbidity, solids, electrical conductance, color, odor, and
temperature.

Pure water rarely occurs in nature. In fact, water commonly contains a variety of
constituents. Quality of water changes due to natural and human actions. Although rain water is
pure, it may interact with the matter present in atmosphere and become polluted. After falling on
the ground, water dissolves and carries impurities present on the ground. Infiltrating water comes
in contact with soil and rocks and some of these are dissolved and transported by water. Natural
waters often interact with organic materials. Human activities produce municipal, agricultural,
mining, and industrial waste. Activities such as recreation, social and religious rituals also cause
pollution.

By the time the precipitation water reaches the surface of the earth, it has already
collected a number of substances. Gases may have been absorbed or dissolved in it and smoke
and dust particles may have been picked up. If the atmospheric water has picked up radioactive
or acidic material, it may not be clean and suitable for some uses.
Precipitation water is subject to further changes of quality after hitting the earth surface
and when it travels on or below the ground, depending upon the place where precipitation falls.
Water reacts with carbon dioxide released by bacterial action on organic matter and acquires
acidic properties. In the acidic conditions, many soil and rock constituents are dissolved in water.
Anthropogenic actions influence the quality of water in significant (and mostly harmful)
ways. Dumping of municipal and industrial waste after little or no treatment in natural waters
results in drastic deterioration of the water quality of natural sources. Agricultural related
activities such as irrigation, use of fertilizer, pesticides, herbicides, etc., may lead to diffuse
pollution of both surface waters and ground water. Irrigation return flows contain chemicals and
also tend to increase salt contents in the receiving water. Construction schemes, such as those
connected with river training, flood control, low flow augmentation, etc., considerably influence
the quality regime. Unscientific mining activities often degrade water quality.
A big range of parameters are used to characterize waters quality. Largely the water
quality management objectives and the previous history of the water body will determine
selection of parameters. Some parameters, however, are of special importance and deserve more
attention. The commonly used parameters are discussed in the following sections. An integral
part of water resources management is that water quality standards are formulated, and
implemented, by doing measurements and monitoring.

Common measurements to estimate the presence of physical impurities and their quantity
in water are turbidity, solids, electrical conductance, color, odor, and temperature. Temperature
measurements help understand the value of density, viscosity, vapor pressure, oxygen saturation
value and rates of biochemical degradation. The test for residue is important in the management
of sewage treatment to indicate the physical state of the principal constituent. The solids present
in the dissolved form impact electrical conductivity of the water.
Monitoring objectives determine which variables are to be measured. Different variables
are measured at different frequencies which depend on the rate of changes in the variable as well
as its intended use. Typically, studies are carried out to collect and define base line or
background information, suitability of water for a designated use, or environmental impact
studies.

6.1.1 Water Quality Data

Water quality variables can be categorized in various groups: organic matter, major and minor
ions, toxic metals, nutrients and sediment data. The biochemical oxygen demand, chemical
oxygen demand and dissolved oxygen show how much organic matter of what type is present in
water. The toxic metals that may be present in water include mercury, lead, arsenic and nickel.
The nutrients, such as nitrogen, phosphorus and potassium, are important from the point of view
of growth of fauna.

Generally water quality variables are not measured continuously because of logistical
reasons. Taking water samples, storing and transporting them to the laboratories, and testing
them are expensive. Therefore, sampling is done at regular (but large) time intervals and
sometimes at irregular intervals. Important water quality variables or those which change rapidly
with time are measured more frequently than those which change slowly. For very few water
quality variables, samples are analysed at the observation station. Most analyses require
sophisticated equipment and are, therefore, carried in a laboratory. A wide range of measurement
techniques have been developed for water quality variables. Here, the discussion is categorized
according to physical, chemical, and biological variables.
After the water quality data have been assembled in a data storage system, the next step is
to interpret the data with respect to specific questions and inputs in management decisions. The
most frequently asked questions are:

 What is the water quality at the location of interest?

 What are the water quality trends for the station or region; is the quality improving or
getting worse?
 How do certain parameters relate with one another ?
 What is the total mass loading of materials ?

6.2 Water quality indicators

We discuss here three categories of indicators: physical, chemical, and biological.

6.2.1 Physical Indicators

The most important physical properties of water are temperature, density, viscosity, specific
weight, and vapor pressure. Most of the physical properties vary with temperature and such data
are widely available in literature (see McCutcheon et al. 1993). Clear natural water is
transparent. Turbidity is measured by determining light transmission. The test has little meaning
in relatively clear waters but is useful in defining the drinking water quality.

Solids can be classified by their state, chemical characteristics, and size and distribution.
Solids can be either dissolved (including colloidal and small suspended particles) or suspended
(including settleable). Particles passing a membrane filter with a pore size of about 1.2 micron
are considered dissolved and those retained on the filter are considered suspended.

The total solids content is the sum of dissolved and suspended solids. The amount of total
dissolved solids (TDS) in water is an important indicator of its quality for drinking, irrigation and
industrial use. TDS affects dissolved oxygen concentration and also influences the ability of a
water body to assimilate wastes. Dissolved solids affect ionic strength of water and thereby
impact mobility and transformation of metals. TDS also affects the growth and decay of aquatic
life.

To determine TDS, a sample of known volume is dried in an oven and the weight of the
residue divided by the volume of the sample, gives TDS which is normally expressed in mg/L.
Wide variations in the TDS concentration are observed in natural waters. TDS concentration in
rain water is below 10 mg/L; in river water it may be of the order of hundreds of mg/L.

Temperature: Temperature is important in water quality studies. Chemical and biochemical

reaction rates increase with temperature. Gas solubility decreases and mineral solubility
increases with temperature. The growth and respiration rates of aquatic organisms depend on
temperature. Most organisms reproduce only when temperature is in suitable ranges.

Temperature of river water generally follows the long term average ambient temperature
of the area. Temperature of water in a lake varies with depth. Cold-water species reside in deep
waters while warm-water species are found in shallow regions near the edges.

Color: The color of water depends on factors such as the type of dissolved material and the
suspended particles. Colors associated with water are many times the result of colloidal
suspension, e.g., tea. Most colors in natural waters result from dissolved material coming from
decaying plants. Many industrial wastes are colored.

Odor: Pure water has no odor. Odor indicates some toxic pollution of water. Water may smell
due to the presence of decaying organic matter. The decaying organic matter may accumulate in
bottom where conditions are suitable for the anaerobic bacteria. Sources of the organics include
plant debris washed into streams, dead animals, microorganisms, and wastewater.

Hardness: Water is classified as soft when the carbonate concentration is below 50 mg/L and is
called as very hard when it exceeds 180 mg/L. According to the guidelines of the World Health
Organization, the hardness of water should not exceed 500 mg/L.

6.2.2 Chemical Indicators

Two type of chemical matter may be present in water (a) inorganic and (b) organic matter. The
chemical tests of water quality determine the presence of specific ions. Indicators such as
alkalinity and hardness are also used to define water quality.

(a) Inorganic Matter

Natural water contains a variety of dissolved salts in solution. The principal chemical
constituents in most waters are Cations (Calcium, Magnesium, Sodium, and Potassium) and
Anions (Bicarbonate, Sulfate, Chloride, and Nitrate). Potassium is usually present in small
amounts and is derived from soil minerals, decaying organic matter, and ashes of burned plants.
Nitrate may also be present in small amounts. Besides, many minor ionic species, derived from
the contact of water with various mineral deposits, are also present.

Silica (SiO2) is the main non-ionic mineral found in natural surface water. Its presence
creates problems especially in industrial applications, where it causes scaling. A variety of heavy
metals may also be found. Important among these are arsenic, barium, cadmium, chromium,
lead, mercury, selenium, silver, zinc, and cyanide. These constituents are of concern primarily
because of their toxicity to micro-organisms, plants, and animals. Typically these constituents
come from the discharge of improperly processed industrial wastes, and high concentrations are
often found in wastewater sludges.

For a solution, pH varies from 0 (maximum acidic) to 14 (maximum basic); pH of a

neutral solution is 7. All geochemical reactions are affected by pH. Anthropogenic sources of
acidity include acid deposition and acid mine drainage (e.g. from coal mines). In addition, some
sources of acidity arise naturally due to the oxidation of sulfide bearing ores.

(b) Organic Matter

Decomposition of organic matter draws upon the oxygen present in water and may render it
unsuitable for aquatic life. In this way, organic matter is important for the health of a water body.
Organic matter is characterized by BOD (biochemical oxygen demand) and COD (chemical
oxygen demand). BOD is the most important indicator of pollution by organic matter. Similarly,
a direct measurement of dissolved oxygen is an important indicator of the health of a water body.
A low value of DO indicates high pollution by organic matter.
 Most organic compounds are combinations of carbon, hydrogen, oxygen, nitrogen,
phosphorous, and sulphur. The principal organic compounds found in natural waters and
wastewater includes proteins, carbohydrates, and lipids. Large presence of organic compounds in
water is harmful from health, treatment, and ecological standpoints. Some organic compounds
may be carcinogenic or may cause mutation in humans and other living forms at extremely low
concentrations, e.g. surfactants, pesticides and agricultural chemicals, organic solvents, etc.

Toxic metals and other elements may exist naturally in water. These metals include
Copper (Cu), Chromium (Cr), Mercury (Hg), Lead (Pb), Nickel (Ni), Cadmium (Cd), and
Arsenic (As). In many instances, concentrations of these have increased due to human activity.
Arsenic has been detected in many water sources in eastern India.

6.2.3 Biological Indicators

Domestic waste may contain a variety of pathogenic organisms, including viruses, bacteria,
protozoa and helminths. Tests for all these organisms are costly and time consuming. Most
commonly sample is tested for Escherichia coli (E. Coli) whose presence indicates the potential
for other pathogenic organisms. Presence of E coli is expressed as the most probable number
(MPN)/100 ml. Important microorganisms of concern in water and wastewater include bacteria,
fungi, algae, and viruses.

Bacteria are single cell protists. Most bacteria can be grouped into four general categories:
spheroid, rod, curved rod or spiral, and filamentous. Fungi are aerobic, multicellur,
nonphotosynthetic, heterophic, eucaryotic protists. Most fungi obtain food from the dead organic
matter. Fungi play an important role in the breakdown of organic materials in terrestrial and
aquatic environments. Consequent to decomposition of organic materials, carbon dioxide is
released to the atmosphere and nitrogen to the terrestrial environment.
6.3 Types of Pollutants
Pollutants can be categorized in five classes: Pathogenic micro-organisms, Organic matter,
Nutrients, Dissolved solids, and Trace compounds.

Faecal pollution is present in large numbers in faeces of individuals. Coliform bacteria

are indicators of faecal pollution and are expressed as most probable number, MPN/100mL. The
presence of the coliform group of bacteria determines the bacteriological quality of water.
Escherichia coli is the most widely known member of the group and is present in large numbers
in domestic and municipal sewage. Its presence in water indicates sewage contamination and
likely presence of pathogenic micro-organisms of human origin. The presence of coliforms in
water is detected by observing their growth in special culture media and making a statistical
inference about their number is reported as MPN /100 ml.
Although tests are available for specific pathogenic micro-organisms, there is no way of
knowing which pathogenic organism may be present in a sample. The cost of testing for all
pathogenic organisms is large. Furthermore, in an aquatic environment the die-off rate of E. coli
parallels that of pathogenic organisms. Hence, E. coli is frequently used as an indicator for
pathogenic organisms. E. coli is exclusively of faecal origin. Some coliform bacteria are normal
inhabitants of soil and water. Results of bacteriological analysis may be reported as total
coliforms and faecal coliforms.

Organic matter depletes dissolved oxygen (DO) through microbial decomposition. Deficit
of oxygen or anaerobic conditions change ecological regime: fishes may die, water emits foul
odours and unsightly conditions develop. BOD test measures aggregate organic matter in terms
of oxygen equivalence.
Organic matter is classified as that biochemically decomposable and that which cannot be
decomposed. Organic matter is characterized by BOD, COD and volatile solids. BOD test
measures biochemically decomposable or degradable organic matter, the other two measure total
organics. It may be pointed out that the three parameters reflect the aggregate organic
constituents. The BOD test measures oxygen equivalence of organic matter. The equation for
BOD decomposition reaction may be written as:

Micro-organisms + Organic matter + O2
CO2 + H2O (6.1)

Bacteria utilize various types of waste organic matter as their food to obtain energy. They
decompose the organic matter to carbon dioxide and water in presence of oxygen. The amount of
oxygen consumed is taken as a measure of the quantity of organic matter. The test is carried out
in a laboratory representation of the aquatic environment; a water sample is added to oxygenated
water and the loss of oxygen measured over a period of time.

Dissolved Oxygen: The DO amount in a surface water body is an important indicator of its
‘health’. The presence of oxygen in water is necessary to maintain favourable conditions for
growth and reproduction of a normal population of fish and other aquatic life. The absence of a
low level of DO in surface waters indicates pollution by organic matter. Under such conditions,
organic matter is decomposed by anaerobic bacteria and obnoxious end-products, such as
hydrogen sulphide, ammonia, etc. are produced
Generally, the dissolved oxygen concentration is near saturation in pristine rivers.
However, concentrations of gases change throughout the day. Usual concentration of CO 2 in
river water could be up to 10 ppm. BOD is generally 1 to 2 ppm for unpolluted river and. pH
values are generally around 7 (neutral). Turbidity is very high in flood season. Quality of ground
water is superior to that of surface water in some respects. For example ground water is free from
suspended solids and objectionable colour. But at the same time, ground water generally contains
higher dissolved solids.

Nutrients are those chemicals which contain elements, such as nitrogen (N), phosphorus
(P), carbon (C), sulphur (S), calcium (Ca), potassium (K), iron (Fe), manganese (Mn), boron (B),
and cobalt (Co). These elements are essential for the growth of living beings. Some of these are
required only in very small quantities and are called micro-nutrients. Plants require relatively
large amounts of C, N, and P. They obtain C from carbon dioxide and N and P from soil or
water. N and P may be limiting in the aquatic or land environment; that is the concentration of
one or other of these species may dictate the biomass of plant species which can survive in a
particular water body.
Nutrients such as nitrogen and phosphorus are essential elements for growth of plants and
animals. Both organic and inorganic forms of these are important. In water quality studies, the
nitrogen-containing compounds of interest are organic nitrogen, ammonia, nitrite and nitrate.
Phosphorus is of importance in aquatic environments. Phosphorous enters into water bodies
through fertilizer, industrial waste and rocks. Phosphorous is vital for all organisms and in many
cases it is the nutrient that limits productivity. Municipal and some industrial wastewaters
contain N and P. When such wastes are added to water bodies, it may result in accelerated
growth of algae (algal blooms or eutrophication). When the algal cells die, it increases the
decomposable organic matter in the water body resulting in consumption of oxygen and the
quality of water degrades.

Nitrogen accounts for about 80% of the gases present in the atmosphere and maintains
equilibrium of nitrogen concentration in open water bodies. Human activities influence nitrogen
in surface water in several ways: A) Nitrogen may be present in wastes discharged into surface
water bodies; B) Runoff from agriculture areas contains nitrogen; C) Nitrogen is also present in
the exhaust of automobiles and industries. The nitrogen solubility in water is very less; of the
order of 15 mg/L. Nitrogen is frequently present as nitrate and ammonia. The usual concentration
of nitrate in streams is 0.5 to 3 mg N/L and of ammonia about 3mg N/L. Runoff from
agricultural areas may have considerably higher concentration of nitrates. Domestic sewage
contains about 15-100 mg/L of the total nitrogen. Nitrification, which is oxidation of ammonia
and nitrite to nitrate, consumes dissolved oxygen in water. Excessive presence of nitrogen in
water can lead to eutrophication.

A large variety of substances exist in water in their ionic form. The major cations are
(positively charged ions) comprise calcium (Ca++), magnesium (Mg++), sodium (Na++) and
potassium (K+) and the associated anions include sulphate (SO4--), bicarbonate (HCO3-) and
chloride (Cl-). The cations having two positive charges are responsible for the hardness of water.
The aggregate salts are measured as total dissolved solids (TDS). Waters having less than 1500
mg/L TDS can be considered fresh waters.

A number of toxic metals and organic compounds may be added to water through
anthropogenic activity. Some metals which are toxic even in small concentrations are cadmium
(Cd), copper (Cu), chromium (Cr), mercury (Hg), lead (Pb), nickel (Ni) and zinc (Zn).

6.4 Sampling Surface Waters for Quality Tests

Sampling is performed to collect a small volume of water to determine the parameter(s) of
interest. Samples should be small enough in volume so that they can be transported conveniently
but should represent the whole water body. The concentration of all pertinent constituents in the
sample should be the same as in the source.
The sampling sites should be clearly located and easily accessible if periodic sampling is
necessary. Only if the samples are taken frequently from the same location, changes in water
quality with time can be determined with confidence. The sampling site should be representative
of the water body. Samples of water taken from the banks are not likely to be representative.
Bridges are good places to collect river water samples since the sample can be taken from the
main current. If there is no bridge in the vicinity of the location, boats are the next best
alternatives. The sample should be stored such that the parameter(s) of interest do no
significantly change before the sample is analysed.

6.4.1 Samplers and Sample Containers

To cover the range of parameters which need to be sampled and analysed a variety of sample
containers are required. Bottles used for collecting microbiological samples must be thoroughly
washed and sterilised before taking the samples. About two to three litres of sample would be
sufficient for most physical and chemical analyses. The simplest water sampler is a bottle
attached to a string. To lower a plastic or glass bottle in a body of water it is necessary to use a
sufficiently heavy holder so that the bottle may drop rapidly to the required depth. Such a holder
designed to carry a bottle to collect samples for dissolved oxygen measurement is shown in
Figure 6.1.
Figure 6.1 Dissolved oxygen sampler

When water from a particular depth is to be collected an arrangement like Meyer’s sample bottle
may be used, Figure 6.2. After the bottle is lowered to the desired depth a slight tug removes the
stopper.

Figure 6.2 Meyer’s sample bottle

Alternatively, if the DO sampler is lowered quickly to the desired depth, it will collect sample
which would be representative of the water at that depth. Small amount of water which gets in
the bottle from other depths while the sampler is being lowered is likely to flow out when water
overflows into the sampler.

6.4.2 Sampling Labeling and Preservation

When collecting the samples weather conditions and the state of water body, etc. at the
sampling site at the time of sampling should be noted. This information will be useful in
analysis and interpretation of data.
Immediately after sampling, the sample bottles should be labeled and given a unique
code number. Information on the label should include:
• Sample code number, location and the person who has collected the sample,
• Date and time of sampling,
• Pre-treatment or preservation carried out on the sample.
• The location of the sampling point should be recorded in terms of district, tehsil
and village, as well as geographical co-ordinates (latitude and longitude). If the
station has a unique identification number, it adequately identifies the sampling
point.
 Samples for BOD and bacteriological analyses should be stored at a temperature below
4°C and in the dark. In the field this usually means placing them in an insulated cool box
together with ice or cold packs. In the laboratory, samples should be kept in a refrigerator.
Samples for DO measurement should be chemically fixed.
If samples collected for COD analysis cannot be determined on the day of collection
they should be preserved below pH 2 by adding concentrated Sulphuric acid. Samples which
are to be analysed for the presence of metals should be acidified to below pH 2 with
concentrated nitric acid.

6.5 Laboratory Analysis of Water Quality Samples

Variations of parameters over a period of time are studied and trends are obtained for each
parameter at various locations. The results of the tests carried out are compared with the trends
obtained. The classification of the river waters of the station is carried out based on the results of
various water quality parameters using the classification chart of U.S. Salinity diagram for
classification of irrigation waters. The results are also compared with the various tolerance limits
for each parameter prescribed by the BIS codes, before publishing Water Quality Books.

6.5.1 Determination of Biochemical Oxygen Demand

When organic matter is deposited in the aquatic environment, it degrades through the action of
micro-organisms. To degrade the organic matter, micro-organisms take up the oxygen dissolved
in water. Consequently, the amount of dissolved oxygen (DO) in water is depleted at a rate
which depends on the temperature and the properties of material discharged. Generally, simple,
non-toxic organic chemicals are degraded faster than complex molecules and thus can exert their
oxygen demand more quickly. As more and more oxygen is depleted, it leads to degradation of
water quality since aquatic plants and animals need oxygen to survive and grow and will die if
adequate quantity of oxygen is not available. Shortage of DO also produces foul odours. The
term oxidation originates from a reaction with oxygen (where the substance that was oxidised
loses electrons to oxygen).
The biochemical oxygen demand (BOD) is the most frequently reported indicator to
assess the state of water bodies affected by organic pollution and depletion of oxygen. The rate
of BOD reaction depends on the temperature, the population of bacteria, and the amount of
organic matter present in the sample. The BOD test measures aggregate polluting organic matter
in a water sample and uptake of oxygen by recreating in the laboratory the same environment
that prevails in nature. Of course, not all environmental conditions can be recreated and so the
BOD test merely gives an indication of the likely pollution.
The standard BOD test is conducted over a 5 day period at 20 °C. In India, the test may
also be conducted over a 3 day period at 27 °C (IS 3025, Part 44, 1993) because the average
ambient indoor temperature in India is around 27 °C. This temperature is higher than 20 °C
which is the case for the UK where the test was originally developed. Besides, the 27 °C 3-day
BOD value is comparable to the 20 °C 5-day BOD value. Another reason behind adopting the 27
°C test in India is that in the event of power failure during the test the impact will be less
pronounced as the incubator containing the samples will be at nearly the same temperature as the
laboratory. As a result, heat exchange will be small and the desired sample temperature will be
maintained for longer periods.
Often the test is carried out over five days at 20 ˚C and is designated as ‘BOD 5’. The
Indian Standard BOD test involves assessing the loss of dissolved oxygen in a sample incubated
for three days at 27 ˚C. This is done by comparing the DO concentration of a sample which has
been incubated for three days with the DO concentration of the sample before incubation. The
loss of DO over the three day period is taken as the BOD (or BOD3) of the sample. Clearly, the
three day test will normally yield lower BOD values than a five day test. When the temperature
is increased from 20 ˚C to 27 ˚C, the reaction proceeds at a faster rate and hence the difference
between the 3 day and 5 day tests is small.

As an example, Table 6.1 gives DO values in BOD bottles, containing identical samples
when incubated at 20 and 27 ˚C over a period of 5 days. It also gives the cumulative oxygen
uptake values. Note that the BOD3 27 ˚C and BOD 5 20 ˚C values are nearly the same.

Table 6.1 DO values in BOD bottles, containing identical samples when incubated at 20 and 27
˚C over a period of 5 days.
Day DO, mg/L 20 Cumulative DO DO, mg/L 27 ˚C Cumulative DO
˚C consumed, mg/L, 20 ˚C consumed, mg/L, 27 ˚C
0 8.1 0 8.1 0
1 6.5 1.6 5.5 2.6
2 5.0 3.1 3.8 4.3
3 3.9 4.2 2.5 5.5
4 3.3 4.8 1.7 6.4
5 2.5 5.6 1.4 6.7

6.5.2 The BOD Changes

Figure 6.3 shows change in BOD over a number of days for two samples. Here sample 1
represents a water body which exerts the oxygen demand rapidly as seen from steep slope of the
curve in initial stages. The sample 2 exerts oxygen demand slowly as seen by the flatter slope of
the curve. Sample 1 could be containing organic material, untreated sewage effluent, or some
material which can be easily oxidised. Sample 2 represents a behavior which is typical of
effluents containing complex organic materials such as industrial chemicals, solvents, or fats
which are difficult to oxidise. The point where the curve asymptotically becomes horizontal and
the BOD value stabilizes is also known as the ‘ultimate BOD’ of the sample. Depending upon
the organic material content of sample, it may take 10 days or even more to arrive at this point.
Twenty-day BOD (BOD20) is frequently considered to be the ultimate BOD.
 




 



Figure 6.3 BOD curves for two samples: Sample 1 – easily degradable organics, Sample 2 –
difficult to degrade organics.

After a sample of water containing organic material is taken from a water body, the
oxidation reactions in the sample will continue thereby altering its BOD. To avoid this, the
samples should be analyzed as soon as possible after collection. If a delay in the analysis is
expected, samples should be stored around 4 to 5 ˚C which greatly reduces the rate of oxidation
reactions and the BOD does not change significantly.
6.5.3 Inferences Concerning BOD
The BOD test relies on growth of the micro-organisms degrading the organic matter in the
sample. Hence, right conditions to allow the microorganisms to grow without undue
environmental stress should be created. If the pH of the sample is too low (below 6.5) or too high
(above 7.5), the BOD analysis may be affected by the ability of bacteria to grow. Therefore, if
the pH of the samples is below 6.5 or above 7.5, it should be modified by the addition of acid or
alkali so that a pH is within the range 6.5 to 7.5.

Toxic material present in the BOD sample also inhibits the analysis and hence the sample
must be ‘seeded’ with bacteria so that the analysis proceeds normally. Organic compounds are
not the only materials which, when discharged to the aquatic environment, have an oxygen
demand. Ammonia, either free or when released from nitrogen containing organic compounds, is
also oxidised in rivers resulting in depletion of dissolved oxygen. The oxidation of nitrogen
compounds, carried out by nitrifying bacteria (a process known as nitrification) in the BOD
sample can be suppressed by the addition of an inhibitory chemical so that only the BOD
resulting from the oxidation of carbon compounds is determined. If such a chemical is not added,
the resulting BOD may be a combination of the oxygen demand caused by both carbonaceous
matter and ammonia based material in the sample.

Algae have the ability to produce oxygen by photosynthesis. Hence, if the sample
contains algae, they must be stored in the dark.

Example 6.1: Calculate the oxygen demand exerted by a sample of industrial waste whose
ultimate BOD is 256 mg/L, in 1, 2, 3 and 4 days. Assume that 25% of remaining demand is
exerted each day.

Solution: Table below shows the computations of BOD. First column shows the day number and
the second the BOD remaining in the sample. Column 3 gives the BOD exerted for the day and
the last column is the cumulative BOD. The BOD at the end of day t is the BOD at the beginning
of a day t less the BOD exerted on day t.

Time Remaining BOD mg/L BOD exerted for the day Cumulative BOD exerted
day mg/L mg/L
0 256 0 0
1 192 64 64
2 144 48 112
3 108 36 148
4 81 27 175
6.5.4 Factors Affecting the Oxygen Demand Rate
A number of factors control the rate of oxidization of organic matter in the aquatic environment.
These are the composition of the material, water temperature and the concentration of micro-
organisms. Generally the rate of the BOD reaction increases with temperature.

The concentration of micro-organisms present in the water body also affects the rate of
the BOD reaction. Normally there are sufficient numbers of suitable micro-organisms in a
natural water body to allow the BOD reaction to proceed. In some cases, for example, when
water contains chemicals which are toxic to bacteria, a few micro-organisms may be there to
carry out the oxygen demand reaction. If the BOD of such a sample were to be determined, it
would give a value much lower than that given the concentration of organic matter in the sample.
To get the correct results, the sample must be ‘seeded’ with suitable bacteria. Usually a small
amount of settled raw sewage is used as a source of seed. In case the waste is toxic, an
acclimated seed must be developed.

6.5.5 Chemical Oxygen Demand Test

When organic polluting matter is discharged in the aquatic environment, it normally takes up
dissolved oxygen during its degradation thus reducing the oxygen available for the respiration of
fish and other aquatic life. The chemical oxygen demand (COD) analysis measures the maximum
amount of oxygen that can be consumed by the organic matter in a sample of water. The test is
based upon the fact that all organic compounds, with a few exceptions, can be oxidised by the
action of strong oxidising agents under acidic conditions.
However unlike BOD, in the COD test the oxygen demand is determined by means of a
chemical test. A strong chemical oxidizing agent is added and it is ensured that virtually all
organic matter within the sample is oxidized during the analysis, some of which may not be
susceptible to bacterial decomposition. The amount of oxygen used during the test is then
calculated. Some naturally occurring organic compounds such as celluloses, or many synthetic
petrochemicals, are either non-decomposable or are degraded very slowly by bacteria.
A limitation of the COD test is its inability to differentiate between biologically
oxidisable and biologically inert organic matter. Further, COD does not provide any evidence of
the rate at which the biologically active material would be stabilised under the existing
conditions. Hence, COD values are greater than BOD values and may be much greater when
significant amounts of biologically resistant organic matter is present. Nevertheless, the COD
test gives a rapid indication of the amount of oxygen a sample will demand when released to the
aquatic environment.

After a sample of water containing organic matter has been collected, it is possible that
natural biochemical reactions within the sample will act to reduce the COD. Therefore, samples
should be tested without delay. If delay in analysing the sample is unavoidable, the sample
should be preserved by acidification to pH = 2 or below.
 Since the COD test measures both the biologically degradable and biologically non-
degradable organic matter, the COD value of a sample is always higher than its BOD value.
Further, COD test measures the total oxygen equivalents of the organic matter while the BOD
test measures oxygen equivalents of organic matter which is degraded in 3 days. This 3 day
value is about 70 % of the ultimate BOD. For domestic wastewaters and many decomposable
industrial wastes, COD is about 2.5 times the BOD. For unpolluted surface and ground waters
this ratio is even higher.

The Total Organic Carbon (TOC) content of a water sample is a measure of all the
carbon, which is bound up in the organic matter in the water sample. TOC is related to BOD and
COD as most of the oxygen demand measured during these analyses is due to organic carbon.

Since the BOD of a sample can be related to both the COD and the TOC, it is possible to
estimate the BOD from either the COD or the TOC. For this purpose, it is necessary to establish
a relationship between these variables for a particular sampling point. This is best done by
carrying out BOD and COD or TOC analyses on the sampling point under various conditions to
establish a reliable relationship.

6.6 Sediment Related Data

The data of amount of sediment transported by a river is need in design and management of
water resources projects, flood control structures, bridges, canals etc. The movement of solids
transported in anyway by the flowing water is termed as sediment transport. The measurement of
sediment is an expensive process and hence sediment rating curves are widely used to indirectly
assess the sediment concentration as a function of river stage or discharge.

Sediment being carried away by river flow originates from various sources, including the
watershed, agriculture field river bed and banks. The sediment data collected includes the
particle size distribution for the sediment and the sediment transport as bed and suspended load.
Sediment transport rates depend on many factors, and it may less than the sediment transport
capacity because of sediment availability. The total sediment transport by a river is the sum of
the suspended load and bed load. The sediment might originate as the bed material load and the
wash load. Fig. 6.4 shows the classification of sediment as suggested by ISO (ISO 4363: 1993).
 Along Bed
Bed the bed load
material
Total
In load
Total load suspension Suspended (t
(origin)
load
Wash load in
suspension

Fig. 6.4 Definition of sediment load and transport (Source ISO 4363: 1993).

Sediment samplers are used to determine the concentration of sediment. The type of
sampler depends upon the need and the purpose. The suspended sediment samplers collect
samples of water and sediment mixture in a river. The sampler is lowered to the desired depth
and is filled with a sample. The sampler should be big enough such that an adequate amount of
sample, say at least 0.5 liter, is filled. A special class of samplers, known as the depth integrating
samplers, is used to obtain the vertical profile. The samplers are filled as they are lowered from
surface to bed and then raised up. The rate of lowering or raising should be constant in both
directions.
In flowing channel, samples are collected using scoop material sampler in shallow rivers
and low velocities. Dredge type bed material samplers are useful for higher depths and velocities.
In dry beds the samples are collected by scrappers after digging a pit of about 1 foot depth in
river bed. The samples collected are dried. Normally 3 to 5 samples are collected during pre-
monsoons monsoon and post-monsoon period. These samples are brought to laboratory for
analysis.

When doing the sampling, the width of the cross-section is divided into sub-areas,
depending on the variability of sediment concentration in the lateral direction. Specially designed
samplers are used to measure bed load sediments. Bed load samplers are lowered to the bottom
and are allowed to be filled up for 5-10 minutes. It is necessary to do sampling a wide range of
discharges so that a representative rating curve relating sediment and water discharge can be
established.
The concentration of sediments is commonly expressed in g/cm3 or kg/m3. The
international standards ISO 3716 and 4363 provide details of sediment sampling. The samples
are usually analyzed in laboratories. After allowing the sediments to settle down, water is
carefully removed from the container and the remaining sediments are oven-dried. The particle
size analysis is first carried out by sieving and then the finer sediments are analyzed using
hydrometer. The sediment transported as suspended load is classified in three categories,
depending on the particle size:
 The coarse fraction (particles above 0.2 mm diameter),
 The medium fraction (particles between 0.075 and 0.2 mm diameter), and
  The fine fraction (particles below 0.075 mm diameter).

Acoustic Doppler Current Profilers (ADCPs) are being increasingly used for streamflow
and suspended sediment measurements. These instruments measure either the attenuation of an
acoustic pulse due to suspended particles or the backscatter of the pulse by the particles. Remote
sensing data are also now being used to estimate suspended sediments in water. The reflectance
properties of water change depend on the concentration and the properties of suspended
sediments and these form the basis to estimate suspended sediments.

6.6.1 Processing of Sediment Data

The sediment data collected at each site is checked thoroughly and processed. An important
analysis is derivation of sediment rating curve.
6.6.2 Sediment Rating Curve
A sediment rating curve relates the suspended sediment concentration in a river with water
discharge . Commonly, the relation is of the following form:
C = a Qb (6.2)

where C is the suspended sediment concentration (mg/l), Q is the discharge (m3/s), and a and b
are constants. A typical sediment rating curve is shown in Fig. 6.5. Conventionally, discharge
and sediment concentration are plotted on a log-log graph paper and a straight line is drawn.
Usually, the power equation is log transformed, and linear regression is applied to estimate the
parameters. A least squares method can be used to obtain the values of parameters a and b.
Typically, exponent b in eq. (6.2) lies in the range between 2 and 3.

In a plot between, discharge and sediment concentration there will be a large scatter in
points. One reason behind this scatter is that soil erosion rates in a watershed are not the same
during different seasons of the year. If the scatter is large, it might be necessary to develop
separate rating curves for different seasons or according to streamflow generation mechanisms,
such as rainfall, snowmelt, etc. The data pertaining to rising and falling limbs of the hydrograph
may also be separated to improve the relationship.

A sediment rating curve is mainly applied to obtain the value of sediment concentration
for a given discharge. Along with the flow duration curve at a given location, the sediment rating
curve can also be used to estimate the amount of sediment transport over a period of time, say a
year. Another important use of sediment rating curve is in estimation of the impact of land use
changes and watershed management on sediment yield.

y = 0.1236x - 264.83
R² = 0.5833
Sediment Conc.

Discharge

Fig. 6.5 A typical sediment rating relationship.

6.6.3 Processing of Bed Material Data

Data pertaining to mean diameter, type and grade of sediment are helpful for correct appraisal of
the behavior of rivers such as braiding, scouring, sedimentation etc. This behavior depends on
flow velocity, slope of river bed, type of catchment, and quantum and type of bed material.
6.7 Water Quality Standards
Water quality standards may be classified as ambient water quality standards, specific water use
related standards and effluent water quality standards.
The Central Pollution Control Board has classified the inland surface waters into 5 categories - A
to E on the basis of the best possible use of the water as shown in Table 6.2. The classification
has been made in such a manner that the water quality requirement becomes progressively lower
from class A to class E.
A water body may be subjected to more than one organised use. The use demanding the
highest quality is the designated best use. A water body or stretch of river whose existing water
quality does not meet the designated best use criteria requires action to mitigate the situation.
Based on such analysis river action plans are formulated.

Table 6.2 Primary water quality criteria for various uses of fresh water
Designated best use Class Criteria
Drinking water source A 1. Total coliform organisms MPN/100mL shall be 50 or less.
without conventional 2. pH between 6.5 and 8.5
treatment but after 3. Dissolved oxygen 6 mg/L or more
disinfection 4. Biochemical oxygen demand 2 mg/L or less
Outdoor bathing B 1. Total coliform organisms MPN/100mL shall be 500 or less
(organised) 2. pH between 6.5 and 8.5
3. Dissolved oxygen 5 mg/L or more
4. Biochemical oxygen demand 3 mg/L or less
Drinking water source C 1. Total coliform organisms MPN/ 100mL shall be 5000 or
with conventional less
treatment followed by 2. pH between 6 and 9
disinfection 3. Dissolved oxygen 4 mg/L or more
4. Biochemical oxygen demand 3 mg/L or less
Propagation of wild life, D 1. pH between 6.5 and 8.5
fisheries 2. Dissolved oxygen 4 mg/L or more
3. Free ammonia (as N) 1.2 mg/L or less
Irrigation, industrial E 1. pH between 6.0 and 8.5
cooling, controlled waste 2. Electrical conductivity less than 2250 micro mhos/cm
disposal 3. Sodium absorption ratio less than 26
4. Boron less than 2mg/L
Source: Central Pollution Control Board, Govt. of India.

6.7.1 Water Quality Index

A number of attempts have been made to develop an aggregate index to denote quality of water
based on broad categories (physical, chemical, and biological). Within each category, a number
of quality variables may be employed and the suitability of a given water source for an intended
use depends on the magnitude of these quality variables. To describe water quality, it is useful to
employ a subindex of a quality variable to indicate the quality of the water on a zero (worst
quality) to unity (best quality) scale. In this context, a variety of subindices have been proposed
over the last two decades. Swamee and Tyagi (2000) conceived the following general
aggregation form:
k
 N

I = 1  N   si1 / k  (6.3)
 i 1 
where N = number of subindices, k = a positive constant, and si is ith subindex.

6.8 Water Quality Monitoring in India

India has 14 major rivers basins (each with a catchment area exceeding 20,000 sq. km) and these
account for 85% of the total surface runoff. Over the time, most of the rivers in India have
become grossly polluted in various stretches. Regular monitoring networks started coming up in
the seventies. Central Pollution Control Board (CPCB) is now operating water quality
monitoring networks on major rivers under Global Environmental Monitoring System (GEMS)
and Monitoring of National Aquatic Resources (MINARS) programmes either directly or
through State Pollution Control Boards (SPCB). Some state pollution control boards have their
own monitoring stations. In addition, the Central Water Commission (CWC) is monitoring water
quality at a number of stations on almost all the major rivers. The National River Conservation
Directorate (NRCD) has also started monitoring river water quality under its action plans.
The Central Ground Water Board (CGWB) is maintaining a countrywide network of
groundwater level and quality stations. Details are given at their web-site: www.gov.in.
Groundwater data are also collected by various state agencies and some pollution control boards.
It can be noted that a number of agencies are collecting water quality data. As
maintenance of water quality networks is expensive, there is a need to clearly state the objectives
of the programmes and optimize the effort where possible

6.9 Closure
Monitoring, modeling, and management of water quality of various sources of water has become
very important for India because many of our rivers and lakes are highly polluted. In fact, some
rivers are so polluted that they resemble a dirty water drain. A disturbing recent trend is that the
aquifers are also being polluted by seepage of polluted water from surface and rivers; some
industries are pumping polluted water in the aquifers to avoid treatment and escape detection by
enforcement agencies. Urgent actions are needed to restore water quality of the sources so that
the ecosystems remain in healthy conditions and provide sustainable benefits to the society.

6.10 REFERENCES
McCutcheon, S.C., Martin, J.L, and Barnwell, T.O. (1993). Water Quality. In Handbook of
Hydrology, Edited by D.R. Maidment. McGraw-Hill Inc., New York.
Reichel, G. (1998). Suspended sediment monitoring: Use of Acoustic Doppler Current Profiler.
In Encyclopedia of Hydrology and Water Resources, edited by Herchy, R.W., and Fairbridge,
R.W. Kluwer Academic Publishers, Dordrecht.
Swamee, Prabhata K. and Aditya Tyagi (2000). Describing water quality with aggregate index.
Journal of Environmental Engineering, ASCE, 126(5), 451-455.
WMO(1988). Manual on water quality monitoring -- planning and implementation of sampling
and field testing. Operational Hydrology Report No. 27, WMO No. 680. World
Meteorological Organization, Geneva.
WMO(1989). Management of ground water observation programmes. Operational Hydrology
Report No. 31, WMO No. 705. World Meteorological Organization, Geneva.
WMO (2008). Guide to Hydrological Practices. Volume I: Hydrology – From Measurement to
Hydrological Information. WMO No. 168. World Meteorological Organization, Geneva.
 Module 7
Acquisition and Processing of Ground Water and Other Data

7.1 Ground Water Data

7.1.1 Time Invariant Ground Water Data
7.1.2 Time Variant Ground Water Data
7.2 Ground Water Level Observation
7.2.1 Automatic Recording Instruments
7.2.2 Observation well network
7.2.3 Water Level Maps
7.3 Well Discharge Measurements
7.3.1 Aquifer Pumping Tests
7.3.2 Measurement of Spring Discharge
7.4 Validation of Ground Water Data
7.4.1 Field Validation of Ground Water Data
7.4.2 Data Entry Validation
7.4.3 Primary Validation
7.4.4 Secondary Validation
7.4.5 Tertiary Validation
7. 5 Sampling ground water from borewells for water quality analysis
7.5.1 Presenting Ground Water Quality Data
7.6 Reservoir and Lake Data
7.7 Demographic and Socio-economic Data
7.8 Agriculture Data
7.9 Land Use Data
7.10 Water Use and Demand Data
7.11 References

Keywords: Ground Water, Data, Time Invariant, Water Level, Pumping Tests, Validation,
Reservoir, Lake, Demographic, Socio-economic, Agriculture, Land Use, Water Use, Demand.

The objectives of this module are:

• to discuss techniques of acquisition, validation, and processing of ground water data;
• to discuss techniques of acquisition and processing of other data used in water resources
planning and management;

Geological material which can hold water consists of unconsolidated deposits or consolidated
rock. Water exists in the openings in this material. Water will flow depending upon how the
voids are interconnected. Void space may be saturated (totally filled with water) or unsaturated
(not completely filled with water). A geological formation that contains and yields sufficient
quantity of water is known as aquifer. An aquiclude is a formation through which inappreciable
quantities of water can flow.
An important hydraulic property that controls the movement of groundwater is the
hydraulic conductivity. Hydraulic conductivity is defined as the volume of water that will move
in a unit time under a unit hydraulic gradient through a unit area, which results in units of
velocity (distance per unit time). A related term, transmissivity, is defined as the hydraulic
conductivity multiplied by the aquifer thickness. It is important to note that the hydraulic
conductivity is a unit property, whereas transmissivity pertains to the entire aquifer. To
understand the ground water conditions in an area, it is necessary to have correct
hydrogeological information. The greater the volume and accuracy of this information, the better
will be the understanding of the aquifer response, water level changes, hydraulic gradients, etc.

7.1 Ground Water Data

The term “ground water data” is used to point to all the data that are required in assessment and
management of ground water. Broadly, ground water related data can be classified under two
headings: time invariant data and time-variant data. Aquifer properties, hydrogeological logs,
and well locations are the data which do not change with time and constitute time invariant data.
Piezometer levels, discharge, and water quality are time variant or time-series data which are
routinely observed and can be processed, validated, and stored using the techniques that have
been discussed in previous modules.
Ground water data can be classified in two broad categories: time invariant data and time
variant data.

7.1.1 Time Invariant Ground Water Data

Information about the nature of subsurface materials can be found in the records of construction
of wells, mines, tunnels, and similar activities. Of particular usefulness to groundwater studies is
the record of conditions encountered during the drilling of a well. This can be done either by the
driller or a geologist who monitors conditions at the site, by bringing the drill cuttings to the
surface and examining any core samples taken. A well log is a continuous record of the type of
material encountered during the drilling of a well. In addition, description of the well site is also
helpful.

Fence diagrams are used to represent stratigraphic data in three dimensions. They are
similar to cross sections, but rather than interpolating subsurface geology from a map, the
geology between stratigraphic sections or cores drilled into the subsurface is interpolated. Fence
diagrams (see Fig. 7.1) are very effective at demonstrating changes in geological formations and
stratigraphic relationships in a region. To construct a fence diagram, the location of each section
is marked on a paper in a same way as if it was a map, and a vertical scale chosen. A vertical line
representing the length of the section is drawn and the stratigraphic boundaries are marked along
the line. The next step is to choose pairs of sections between which to draw the “fence” based on
relative locations of sections and the lithologic and stratigraphic variations. Generally, those
sections are selected which will present the panel in the most useful orientation and will show the
widest variation in lithologic and stratigraphic relationships. Most sections are connected to two
other sections with panels; some of them may be connected to three and the sections on the edges
may be connected
c to
o only one seection. In so
ome cases, oone of the paanels will bee partially hiidden
behind annother one. Once
O all of the
t useful paanels are commpleted, the fence diagrram will show w the
three dim
mensional geometry of thhe various stratigraphic uunits.

7.1.2 Time
T Variannt Ground Water
W Data
The mosst common time variantt ground waater data arre the time series of w water levels in an
observatiion well and d withdrawaals from a well.
w An exiisting well m may be useed to observve the
water tab ble if the depth of th he well exteends sufficiiently below w the rangee of water level
fluctuatio
ons. Existing pumped wellsw can also
a be usedd for this purpose if thhe annular sspace
between the outer casing
c of th
he well and the pump column alloows smoothh passage oof the
measurinng tape or caable for meaasuring the water
w level. IIf an existinng in use weell is taken uup for
observatiions, measurrement shou uld be carried out after tthe cessationn of pumpinng and passaage of
sufficientt time so thaat the water level in thee well fully rrecovers. Liikewise, withhdrawal of w water
from any y other welll in the vicinity of th he observatiion well should also bbe stopped for a
sufficienttly long timee.

Fig. 7.1 Fence diagram

A time seriees of groun nd water daata typicallyy consists oof depth off water table or
piezomettric head meeasured at an
n observation
n well or pieezometer at ddifferent tim
mes. Time intterval
between the successsive data maay be uniforrm or non-uuniform. Annalyses of suuch a time sseries
provides valuable information on trend of the data, ground water potential and its use, and
estimates of aquifer parameters. Joint analysis of time series of water level and rainfall can
provide useful insight about recharge to ground water.
The water level in a well is most often observed manually by dipping a graduated tape
thereby producing a time series (at non-uniform) time intervals. In recent past, automatic water
level recorders are increasingly being used. In India, there are wide variations in frequency of
observations and ranging from three times a year to weekly observations.
Water withdrawal from the pumping wells are estimated by gauges, based on duration of
pumping, or from the quantity of electricity consumed. Of course, specification of the pump and
the pumping head should be known. The discharge of springs can be easily measured by
constructing a weir or V-notch.

7.2 Ground Water Level Observation

Ground water levels in observation wells can be measured either manually or with automatic
recording instruments. Till a few years ago, ground water levels were mostly measured
manually. Automatic recording instruments are becoming quite common these days.

The most common and simple manual method is by suspending a weighted line (for
example, a graduated flexible steel or plastic-coated tape) in the well from a fixed point at the
well head. The depth of the ground water level is obtained by subtracting from the total length of
the tape suspended in the well that length which has been below water or wetted. This wetted
part can be identified, for example, by putting chalk on the lower part of the tape before
measurement. Usually approximate depth to water level is known before hand and this serves a
check on the measurement. As depth-to-water level increases and consequently the length of tape
to be used increases, this method becomes cumbersome. Depths-to-water surface of up to 50 m
can be measured without much difficulty. At these greater depths, steel tapes of narrower widths
or lightweight plastic-coated tapes can be used. Depths to water level can be measured with
accuracy of a few mm by most methods and the accuracy usually depends on the depth.

7.2.1 Automatic Recording Instruments

Many different types of automatic water level recorders are in use. A good water level recorder
should be easily installable, and capable of recording under a wide variety of climatic conditions
and operating unattended for long periods of time.
A recorder suitable for Indian conditions and in use at numerous locations is the float
type recorder. This recorder has been explained in Module 4 and the basic principle of operation
is the same for river water level as for ground water level. However, measurement of ground
water levels in a small diameter well is cumbersome since a float system is difficult to work as
the diameter of the float decreases. Miniature floats of small diameter have been developed to
follow changes in water level in a well. Although these instruments have particular value in
small-diameter wells, they can be installed in wells of any diameter greater than the working
diameter of the probe.

Maintenance of sophisticated automatic recording instruments poses problems due to

inadequate infrastructure and trained persons, particularly in remote areas. Simple repairs can be
performed on the site, but for more serious faults, the instrument has to be sent to the company or
the workshop. Many times, spare parts are to be imported which requires time and money.
Adequate protection from extremes of climatic conditions, accidental damage and vandalism
should be for these instruments for these.

7.2.2 Observation Well Network

The density of the observation well network is usually based on the data requirement for
planning and management decisions, topographic and geologic variations, and the resources
available for well construction and network operations. Drilling of observation wells is one of
the main costs in groundwater studies. The use of existing wells provides an effective low-cost
option. Therefore, in the development of an observation network, existing wells in the study area
should be carefully selected and wherever necessary, these should be supplemented with new
wells drilled and specially constructed.

7.2.3 Water Level Maps

Fluctuations in groundwater levels reflect changes in groundwater storage within aquifers. These
fluctuations can be in two main groups: long-term (e.g., those caused by seasonal changes in
natural recharge and persistent pumping, and short term (for example, those caused by the effects
of brief periods of intense pumping). Since groundwater levels generally respond slowly to
external changes, continuous records from water-level recorders are often not required.
Generally observations at large and fixed time intervals are adequate.

A useful approach to organize and present water level measurements from a network of
observation wells is to produce an accurate map of well locations and then draw contours of
water-level at each well. Two types of maps can be prepared; depth-to-water table measured in a
well from the land surface or the elevation of the water level in the wells relative to an
established datum, such as the mean sea level. Generally, these maps are produced for a single
aquifer.

Depth-to-water Maps
Another type of maps depicts the depth-to-water level in a well relative to land surface. These
maps are referred to as a depth-to water map. These types of maps provide an indication of the
depth to drill to draw water which can be useful in planning utilization of resources. A map
based on the difference in depth to water between two different times would show, for example,
the areal variation of seasonal fluctuations. Of course such maps cannot be used to establish the
direction of groundwater flow because it also depends on the variation of topographic elevation.
Water Table Maps
A map drawn by using the elevation of the water table in a well with reference to a common
datum, such as the mean sea level, is known as a the water table map. Accurate elevation of
measuring point at each observation well is required to produce such maps and this is a difficult
task. Each depth-to-water measurement collected must be subtracted from the elevation of the
measuring point relative to the datum to produce the necessary data. This type of map has a
significant benefit that it can be used to infer the gradient of groundwater flow in many cases.
The accuracy of the map is depends on the accuracy with which the elevations of measuring
point are determined. Most commonly, the elevations are determined from the Survey of India
topographic maps or by carrying out leveling from a nearby benchmark. Alternatively, a GPS
may be used but the elevation data may have considerable errors.
7.3 Well Discharge Measurements
Measurement of a pumping well’s discharge is important in ground water modeling,
management and regulation. The discharge of a pumping well will vary with rise and fall in
groundwater level. This requires repeated measurements to keep track of the rate of change of
discharge. When a pump is turned on, the water level in the well drops quickly, causing the
discharge to vary. Pumping rate becomes stable within a few hours. Water-level changes that
could affect pumping rate can also occur as a result of recharge from precipitation or changes in
pumping of nearby wells.

The simplest method to determine the discharge from a pumping well is by measuring the
time required by discharge to fill a known volume. Dividing that volume by the time yields the
pumping rate. The accuracy of the measurement depends on the accuracy of measuring time.
However, at relatively high discharge rates, a measurement of this type may require logistical
planning and creating a large enough temporary storage space. Another way is to construct a V-
notch in the channel which carries the pumped out water.

7.3.1 Aquifer Pumping Tests

Aquifer pumping tests are carried out to measure the drawdown resulting from pumping a well
and then to calculate the hydraulic parameters of the ground water zone. The magnitude and
timing of drawdown observed in a specific test is directly related to the hydraulic conductivity
and storage coefficient, respectively of the aquifer.

In an aquifer test, pumping is controlled and generally held constant for the test duration
and the ground water levels in the pumped well and a nearby observation wells are measured. A
cone of depression is formed in the aquifer due to pumping. Depending on the time required to
achieve steady state conditions, pumping is carried out for an extended period of time.

7.3.2 Measurement of Spring Discharge

Monitoring programme in ground water must also include observation on baseflow in rivers and
springs. Baseflow is an indirect measure of recharge as it represents drainage of ground water
from the aquifer storage. Because of the slow response of the ground water storage, variations
in annual recharge are not immediately seen in the base flow or in spring flow. Large ground
water abstraction results in reduced baseflow/ spring flow.

Spring discharge may vary from zero or a trickle to a large flow which varies over time.
In a monitoring programme, it is essential to prepare an inventory of major and medium springs
and measure their discharges at least once in a month. In the case of small or medium size
springs, a channel is made and a V-notch is constructed for flow measurements.

7.4 VALIDATION OF GROUND WATER DATA

After the ground water data have been entered in the database, its validation should be carried
out. Similar to the surface water data, validation of the data is done at different levels with
increasing sophisticated checks. The different levels of data validation are:
a) Field validation;
b) Data entry validation;
c) Primary validation;
d) Secondary validations; and
e) Tertiary validations.

It is a good practice to record the results of the validation in the database. For this
purpose, a special field may be created and a code, a flag, or a label is inserted after validation
task is completed and the data are considered to be correct. General concepts of data validation
have been discussed in Module 2. This section discusses some aspects specific to ground water
data.

Ground water level measurements can be checked as follows:

1. Contour maps of groundwater level data for a certain period or date are prepared to
identify deviating values. These values are shown by a high concentration of contours
around the location of the well with the erroneous value. Further check of the hydrograph
of the well is necessary to determine whether the value really is in error.
2. Multiple hydrographs for a group of wells are prepared and the visual inspection of the
graphs will show where the errors may lie. When hydrographs from wells with similar
hydro-geological conditions are prepared and compared, they would normally show
similar trends and fluctuations. If not, the deviation in behavior may be due to errors in
ground water level measurements.
3. Simple statistical tests are also performed to check errors in groundwater level
measurements:
a. Deviation from the mean or the median: values, which differ by more than three
times the standard deviation from the mean or median. The median is less affected
by values of individual measurements and better represents the centre of the data.
Therefore the median is preferred to the mean. It is preferable if the time series
contains at least 50 measurements.
b. Further, in case the data has a trend, the mean may not be the correct reference. In
case of hydrographs with a trend to identify outliers, the validation should be done
using the trend line ±3*SD.

7.4.1 Field Validation of Ground Water Data

The field validation of the data is carried out to ensure that all basic information relating to the
hydrogeological properties are correct. To carry out field validations site is visited, details of
hydrogeological features, coordinates, constructional details, local geology and geomorphology
are collected.

For observation wells, the location and the elevation of the measuring point should be
correct because the measured depth to the groundwater level depends on these. Any shift in the
measuring point should be promptly corrected into the database. The depth of the well should
also be entered in the database and is important in data validation.

7.4.2 Data Entry Validation

After the time series data, checks are entered in the database, checks are carried out to ensure
that there are no duplicate entries in the database. For example, two groundwater level data
should not have the same date and time, because it is not possible to take two measurements at
the same time.

After entry, the data entered must be carefully compared with the field records to ensure
that there are no errors. A common method to check the data entry errors is to get the data
entered by two different persons and then compare the data entered by them. The values which
are same are considered to be correct and where there are differences, the corresponding data are
checked form the original records and corrected. It is assumed that the probability of two
different persons making the same mistake is nil.

7.4.3 Primary Validation

The primary validation uses the statistical features of the data and graphical presentation
techniques for validations. A general visual inspection of the entered data may be done onscreen
or by use of print-outs. The inspection should concentrate on a comparison of the field registers.
The inspection should try to detect errors in the names and the magnitude of numbers. Examples
of print-outs to be used:
• lithological logs on a graph
• chemical data on a water quality report
• water level data on time series graphs
It is important for the visual inspection that the graphs are printed which contain all the
data relevant to the validated feature. In the case of the graphs, the scales of the axes can be
chosen carefully in order to prepare graphs with sufficient detail.

The geographical locations of the wells or other structures may be checked by plotting
the locations on a map of the area. It is important that the attributes of the map layers are based
on the same master information as the database. Otherwise, correction of the data may give
unwanted results. For example, in the case of administrative names, frequently problems arise
with spellings. Different persons use different spellings and this causes difficulties in search.
Hence, it is important to use the same spelling in all the places.
 Before validation of ground water level data, the well data should be validated including
checks on the details of wells. Some of these checks included in primary validations the well
location on a map, the elevation of the measuring point, etc.

Errors in ground water level data may be detected by carrying out simple checks which
flags:
• all groundwater levels which are below the bottom of the well,
• all groundwater levels which are above the top of the well plus 0.25 meter.

Visual inspection of the data is carried out by looking for:

• errors due to a wrong reading (mistakes in readings by 0.5 or 1.0 meter),
• errors due to a change in the height of the measuring point,
• errors due to assigning the groundwater level measurement to the wrong well or.

Visual inspection of plots of ground water levels usually identifies suspect groundwater
levels by comparison with the levels measured in a nearby well at the same time. An example is
given in Figure 7.2, where two ‘doubtful’ measurements are indicated in the well in upper
panel.
The correction of flagged doubtful measurements is not straightforward. A typical
problem is: What was the correct value? The observed value should only be corrected in case the
‘right’ value can be determined without any doubt. In all other cases where it is not possible to
alive at a ‘correct’ value, the value should be flagged as ‘doubtful’ value.
Water Level (m) 

Well No 1 
 Water Level (m) 
Well No 2 

Figure 7.2: Comparison of hydrographs

7.4.4 Secondary Validation

The secondary validation uses the specific expected properties of the data. The use of these
techniques will indicate doubtful values. The results of the validation should be judged with
knowledge of the hydrogeological conditions of the area to determine whether the selected
values are really errors or natural extremes.

7.4.5 Tertiary Validation

Tertiary validation involves advanced techniques for the analysis and validation of spatial and
temporal data. The techniques used here will include advanced statistical tools, and spatial
overlays. Tertiary validation requires knowledge on the tools used and will have to be carried out
by appropriately trained staff.
7. 5 Sampling ground water from borewells for water quality analysis
Samples are collected to determine the property of interest of the (water) body that is under
study. For this purpose, a small portion of the material is picked up such that it faithfully
represents the properties of the body being studied. Hence, it is important that a ‘representative
sample’ is collected.
Ground water samples are typically collected from piezometers which are purposefully
built observation wells, producing wells, or springs, etc. From the piezometers, samples are
picked up whereas samples from a production well or a flowing spring are collected at the
discharge point. If the wells are not in regular use, the quality of well water may not be the same
as that of water in the aquifer. It is important to pump out the stagnant water in the well so that
the sample is taken is of the water from the aquifer rather than standing water in the well. This
process is called as purging.
Samples from shallow wells can be manually collected by lowering the sampler in the
well. Care is needed to ensure that the sampler does not scrap the sides or bottom of the well and
does not collect the solid matter.
Ideally, some other data are also collected while taking a representative groundwater
sample. The water depth in the monitoring well is measured and noted. The date, time and the
name of the person carrying out the sampling are also recorded. Parameters such as the pH, EC
and temperature are observed as soon as the sampling bucket is filled with water. A sample is
also collected to note the water's physical properties (colour, turbidity and odour). After the
initial chemical measurements are recorded, the pump discharge or flow rate is recorded. This
involves emptying the bucket and recording the time taken to fill the bucket to a known volume.
By dividing the bucket volume (litres) by the time required to fill the volume (seconds) the flow
rate in litres/second (LPS) is determined.

7.5.1 Presenting Ground Water Quality Data

Piper diagrams are trilinear representations of the ions found in the water (cations, anions &
combined properties) and are used to classify water types or chemical facies. The central field
represents the anions and cations in a water sample, where they lie in terms of mixed-cation-
mixed-anion facies, and can potentially reveal information about the factors that may be
controlling the plotted water chemistry.

The anion and cation are plotted on their individual triangles and then their lines are
extended up onto the diamond diagram. Where the two lines intersect, the point for that water
sample is plotted. This point tells about the water sample in terms of its chemistry and the
controlling factors. If multiple data points from different sites form a line, it can be concluded
that all the water samples are derived from the same source. Ca2+, and Mg2+ is first plotted on the
cation triangle. The relative abundance of Cl-, SO42- and HCO3-+ CO32- is then plotted on the
anion triangle. The two data points on the cation and anion triangles are then combined into the
quadrilateral field that shows the overall chemical property of the water sample (Figure 7.3). The
Piper diaagram not on nly shows graphically
g th
he nature off a given waater sample, but also dicctates
the relationship to otther samples. For examp ple, by classiifying samplles on the Piiper diagram
m, one
can idenntify geologiic units with h chemically y similar wwater, and deefine the evvolution in w water
chemistry y along the flow
f path (Figure 7.3).

Diagrams
D succh as the Pipper diagram will bring oout deviatingg values. In a Piper diaggram,
outlying points indiicate extrem me water qu uality condiitions. However such rresults may also
indicate erroneous
e vaalues for the constituting
g componentts, e.g., the cconcentrationn of a compoonent
may havee been determmined by clo osing the ion
n balance wiith a residuall value.

Figure 7..3: Piper–diaagram showiing outliers

7.6 Reservoir
R an
nd Lake Datta
Data perrtaining to reservoirs have both time invarriant/slowly--variant andd time invaariant
componeents. The tim me invariant/ slowly varyying data inccludes the eleevation-areaa-storage cappacity
table, an
nd operating g policies. Various
V colu
umns of thee reservoir working tabble, viz., innflow,
outflow through can nals, power house and spillway, eevaporation release to m meet the vaarious
demands, hydropow wer generated d (if any) etc.
e are the time-seriess data. Depending uponn the
purpose ofo the reserv
voir, these daata may be observed
o andd stored at m
monthly, dailly, or hourlyy time
interval.

For lakes, vollume of wateer stored at different

d elevvations is im
mportant. Bathymetry surrveys
are carrieed out to maap the bed prrofile and esttimate the sttorage capaccity of a lakee. In a bathym
metry
survey, depth
d of waater at differrent location
ns can be m measured usiing an echo sounder. G Global
Positioniing System (GPS) is such s extrem
mely helpful in surveyss of catchm ments, lakes,, and
reservoirrs.

7.7 Demographi
D ic and Socioo-economic Data
D
The dataa about popu
ulation and economic
e acttivities are aalso needed for planningg water resoources
projects. Census or demographic data are periodically collected in most countries. In India, an
elaborate exercise is carried out every 10 years to gather extensive demographic data which
forms an important input in national planning, including the water resources. The census data
also form the basis to forecast the population growth for use in planning activities. Census 2011
has been completed recently and a large volume of high quality demographic data has been
collected. In this survey, some new information was also gathered and preliminary results have
also become available.

It is essential to know the human resources available, since all development proceeds
through humans, for humans and on behalf of these. The survey should determine the size of the
active and potential labor force for agriculture, together with that for the tertiary sector
contributing to agricultural development. The population features as well as future projections
are needed to estimate the present consumption pattern and future demand of agricultural
products. The level of employment and the income of the entire agricultural and non-agricultural
active population must be found. It is a settled fact that assured irrigation is key in growth of
both mechanized large farms as well as manual labor-oriented small farms.

Normally, the socioeconomic and agriculture data are compiled for two conditions: ‘with’
and ‘without’ project. It is improper to base the analysis on a comparison of conditions before
and after project construction because a decline or improvement of present conditions might
occur even in the absence of the proposed project. This factor should be recognized while
determining the impacts attributable to the project. Consistent assumptions should be applied to
future conditions with and without the project so that comparability is assured. The net
incremental benefit stream should be an accurate reflection of the project's income generating
capacity, or its net contribution to real national income.

7.8 Agriculture Data

Since agriculture is the largest user of water and may water projects in India have irrigation as
one of the purpose, agriculture data are important inputs in many studies. Data collected typically
includes cropping pattern, area under different crops, and sowing and harvesting dates for the
crops. Since these data change from year to year, it is necessary to collect this data for different
years. In terms of water use, the required data includes crop water requirements, dates of
irrigation and water supplied, and method of irrigation. For economic studies, it is necessary to
know the yield of each crop. From the point of view of water quality studies, data about the
amount of fertilizers, nutrient, and pesticides applied to the fields are required.
Agricultural and irrigation departments in India usually gather such data for
administrative units which are typically blocks (a small part of a district). Since land revenues
are collected, based on crops and agricultural production, the revenue department also gathers
some of these data. Note that the hydrologic boundaries rarely coincide with the administrative
or political boundaries and hence it is frequently necessary to interpolate the data.
7.9 Land Use Data
Till 1949-50, the land area in India was classified into five categories known as the five-fold land
utilisation classification. After 1950, a nine-fold land-use classification has been adopted,
replacing the old five-fold classification. Table 7.1 gives the nine-fold classification and its
relationship with the old five-fold classification.
The new land-use classification primarily indicates if the land-use in a particular area is
cultivated, grazed or forested. Its main purpose is to show the distribution of the existing land in
detail according to its actual use & not how a particular piece of land can be potentially utilised.
Thus, the area under culturable waste land does not represent the area which is really culturable,
as it may not be possible to bring under cultivation large part of the area, except at huge cost.
Thus the potential land-use classification is beset with several difficulties, as this classification
would depend upon the suitability of different areas for different uses, taking into account their
natural endowments, the availability of capital and other resources for the development of land
for the desired use and the likely economic returns. If the potential use of the land has to be taken
into account, a large amount of data relating to the inherent characteristics of each soil type & the
economics of putting it to a particular use would have to be specially collected through soil
surveys, land-use surveys & waste land utilisation surveys.

Table 7.1 Classification adopted for land-utilisation statistics

S.N. Old classification S.N. New classification
1. Forests 1. Forests
Area not available for
2. 2. Land put to non-agricultural uses
cultivation
Other cultivated land,
3. 3. Barren & unculturable land
excluding current fallows
4. Fallow lands 4. Permanent pastures & other grazing lands
Miscellaneous tree crops & groves, not
5. Net area sown 5.
included in the net area sown.
6. Culturable waste
7. Fallow land, other than current fallows
8. Current fallows
9. Net area sown
http://www.krishiworld.com/html/land_utilisation2.html

On the basis of the nine-fold classification, it is possible to arrive at the area according to
concepts like 'arable land' or 'potential land' available for crop. The 'arable land' would comprise
the 'net area sown' plus the 'current fallows' & 'other fallow land'. Similarly, the 'potential land'
available for cultivation would include, besides the 'arable land', the land under 'culturable
waste', 'permanent pastures & grazing land', & 'miscellaneous tree crops and groves, not included
in the net area sown'. However, such potential land cannot be ascertained without surveys of the
soil types & land-use & the economics of bringing them under cultivation.
The utility of the concepts of 'arable' and 'potential' land can be enhanced if some
indicators of arable and potential land are computed:
 Percentage Potential land exploited = Net area sown/Potential land*100
 Percentage arable land exploited = Net area sown/arable land*100
To standardize the concepts and definitions, standard definitions of the various categories
of land use have been adopted in India and these are given in Table 7.2.

Identification of areas earmarked for annual crops, orchards, pastures, forests and non-
cultivated along with the knowledge of the system of land tenure (owner occupancy, tenancy,
share cropping), size of farms, and fragmentation, enables preparation of preliminary blueprints
on the development possibilities. These details would also show the constraints on the
availability of land for cultivation. Soil properties would determine the spectrum of crops which
can be grown in the area.

It is also essential to know the present cropping pattern, crop rotations, varieties grown,
yields for each crop, and market price of each crop. These along with soil data are used to
optimize the cropping pattern and estimate irrigation water requirement for a proposed project.

Table 7.2 Standard definition of various categories of land use adopted in India

SN Category Description
1 Forest Includes all lands classed as forests under any legal
enactment dealing with forests or administered as forests.
2 Area under Non- Includes all lands occupied by buildings, roads and railways
agricultural Uses or under water, e.g. river, and canals and other lands used for
non-agriculture purpose.
3 Barren and un-cultivable Includes all barren and un-cultivable land like mountains,
land desert etc.
4 Permanent pastures and Includes all grazing lands where they are permanent pastures
other grazing lands and meadows or not. Village common grazing land is
included under this head.
5 Land under miscellaneous This includes all cultivable land, which is not included in
tree crops and groves etc. ‘Net Area Sown’ but is put to some agricultural uses. Lands
under Casuarina trees, thatching grasses, bamboo bushes, and
other groves for fuel, etc which are not included under
‘Orchards’ are classified under this category.
6 Culturable Wasteland This includes lands available for cultivation. Such lands may
 be either fallow or covered with shrubs or jungles, which are
not put to any use. Land once cultivated but not cultivated for
five years in succession should be include in this category at
the end of the five years.
7 Fallow lands other than This includes all lands, which were taken up for cultivation
current fallows but are temporarily out of cultivation for a period of not less
than one year and not more than five years.
8 Current Fallows This represents cropped area, which are kept fallow during
the current year. For example, if any seeding area is not
cropped in the same year again, it may be treated as current
fallows.
9 Net Area Sown This represents the total area sown with crops and orchards.
Area sown more than once in the same year is counted only
once.
Source: Ministry of Statistics and Programme Implementation, Govt. of India.

Agriculture land/Cultivable land/Culturable land = 5+6+7+8+9

Cultivated Land = 8+9
Reporting area of land utilization = 1 to 9

7.10 Water Use and Demand Data

Since water is a reusable commodity, water use and water abstraction are two different
quantities. Water use denotes the quantity of water that is withdrawn from water bodies for
supply to fields, cities, industries, or for environmental needs. The consumptive use of water is
that quantity which is evaporated, transpired, incorporated into crops, or consumed by humans or
animals. This water is not immediately available for other uses. For some purposes, such as
hydropower generation, navigation, and recreation, water is used where it is available (instream
use) but is not withdrawn. Off stream uses, such as municipal water supply, irrigation, and
industrial use, require that water be removed from the source. A part of this water may come
back to the river, lake or aquifer as return flow and can be subsequently used.
Water use for various activities is estimated from the sources of supply. For example, for
municipal use, the water withdrawn from surface and sub-surface sources is estimated. Same is
the procedure to estimate water use for agriculture.

The data about municipal can be obtained from municipal water works department. To
compute featuring water demand, the population is projected for some future year (say 2050).
Per capital water norms for urban and rural areas are used to compute the total water demand.
Data about industrial water demand are difficult to obtain. In absence of any better estimate, it is
assumed that the industrial water demand is the same as the municipal water demand. The
procedures to estimate of water requirements for agriculture, hydropower, etc. and their future
projections have been described in many text books.
7.11 Closure
Monitoring of ground water regime involves obtaining information on ground water levels and
quality through sampling. This chapter has also discussed some other data which are of immense
value in water resources planning and management. These include reservoir and lake data,
agriculture Data and land use data. Besides hydrology, data from social sector is also needed in
planning and management tasks. To that end, discussion includes demographic and socio-
economic data and water use and demand data.

7.12 REFERENCES
Jain, S.K., and Singh, V.P. (2003). Water Resources Systems Planning and Management.
Elsevier, Amsterdam.
WMO (1989). Management of ground water observation programmes. Operational Hydrology
Report No. 31, WMO No. 705. World Meteorological Organization, Geneva.
WMO (2008). Guide to Hydrological Practices. Volume I: Hydrology – From Measurement to
Hydrological Information. WMO No. 168. World Meteorological Organization, Geneva.
 Module 8
ACQUISITION AND MANAGEMENT OF SPATIAL DATA

8.1 Spatial Data

8.2 Point Data
8.3 Linear Data
8.3.1 The River
8.3.2 The drainage network
8.3.3 Stream Profile
8.4 Area Data
8.4.1 River Cross-section
8.5 Volumetric Data
8.6 Topographic Maps
8.6.1 Soil and Land Use Maps
8.7 Emerging Technologies for Spatial Data Acquisition and Management
8.8 Remote Sensing
8.8.1 Physical Properties of Electromagnetic Energy
8.8.2 Interaction of Electromagnetic Energy with Objects
8.8.3 Basic Components of a Remote Sensing System
8.8.4 Reflectance Characteristics of Different Objects
8.8.5 Remote Sensing Platforms
8.8.6 Digital Image Processing
8.8.7 Indian Remote Sensing Satellite Program
8.9 Geographical Information Systems
8.9.1 Advantages of GIS
8.9.2 Spatial data representations
8.9.3 Map projection
8.9.4 GIS Operations
8.9.5 Spatial Data Analysis
8.9.6 Applications
8.10 Closure
References

Keywords: Spatial Data, Point, Linear, Topography, Remote Sensing, Geographical Information
Systems

8.1 Spatial Data

Spatial data are those data that have a spatial or space dimension, for example, command area of
a project. A wide variety of spatial data are used in water resources development and
management. Maps are the most effective means of visualizing the spatial data. In water
resources, both time-varying and time-invariant data are used. The spatial data that do not vary
with time (at least on time scales of interest here) include the catchment/command boundary,
topography, soil map, stream network, and geological features. Some spatial data change with
respect to time and such data include land use, cropping pattern, etc. Many times, variation of a
hydrological variable is displayed through a map, e.g., isohyetal map.
 Spatial data pertaining to a river such as channel cross sections, longitudinal profiles and
bed characteristics are needed for many hydrological studies. These data slowly vary with time
and they can, therefore, be considered as semi-static. In alluvial areas, channel cross section and
bed characteristics may significantly change after a major flood event. River cross section area
and its properties have important bearing in stage-discharge relation. It is important that these
data are collected at the desired locations with appropriate frequency. Conventionally, such data
are stored in the form of paper maps and analyzed using manual means.

Large infrastructure is needed to capture, manage, and analyze spatial data. Spatial data
infrastructure consists of technology, hardware and software, policies and procedure/rules, and
people that are necessary to acquire, store and manage, analyze, and share the spatial data among
the users. It provides a structure of practices and relationships among data producers and users
that facilitates data sharing and use. A rigorous national data framework facilitates information
exchange and significantly reduces duplication of effort.

In GIS, maps are called spatial data. Information shown on paper maps can be input is
GIS as spatial data. Example of spatial data are stream network, well locations, villages, cities,
topographic contours, spot elevations, roads, land use, soil, geology, hydrological investigation
locations, hydrological response units etc. Spatial data are classified in to three types, namely
area, line and point. Areas are spatial data that are represented as closed figures, e.g. forests,
lakes, Thiessen polygons etc. Lines are spatial data that are represented as curves e.g. forest
boundaries, lake boundaries, contours, stream network, roads etc. Point spatial data are
represented as points on maps e.g. well locations, rain gauge stations, villages, etc. The data is
scale dependent in some instances, e.g., on small-scale maps a city will be represented as point
data, whereas on large-scale map, it will be represented as area data. A closed line data can be
converted easily to area data in GIS. For example forest boundary data can be converted to forest
land-use map. Point data cannot be converted in to area data. But reverse is true, i.e. area data
can be converted as point data.

8.2 Point Data

In the context of spatial data, a point is a unique location on a line, in an area, or in a volume. It
has no spatial extent. A point may represent the location of a rain gauge, the location of a well,
the outlet of a basin. The attribute of a point may be simple or complex. A simple attribute may
be, for example, its coordinates which uniquely identify its location in the three dimensional
space. A complex attribute may be the geological profile beneath the point.
The physiographic description of a point covers its geometric properties (form, relief,
slope, etc.) and its permanent physical properties (permeability, nature of rocks, soil structure,
land-use type, etc.). The former are limited to the local slope, while the latter comprise a whole
range of possible physical properties, expressed in scalar form for a point on a horizontal surface
or in vectorial form for a profile, for example, geological core.

8.3 Linear Data

A linear feature can be represented by a line on a map, for example a road or a canal. In
hydrology, three types of linear elements are common:
a) Boundaries;
b) Isopleths of a permanent feature, for example, contours;
c) Thalwegs.

The first two types are linked to areal aspects, which will be examined later. The thalweg is itself
to be considered not only as represented in horizontal projection and longitudinal profile, but
also by the way in which it combines with other thalwegs to form a drainage network, which has
its own physiographic characteristics. Some drainage network characteristics are linear, for
example, the bifurcation ratio, while others are areal in nature, such as the drainage density.

8.3.1 The River

Depending on the scale of the map, a river may be represented by a single curve or by two curves
representing its banks. The river is represented by two lines when either the map has small scale
or the river is very wide or both. Rivers are frequently shown by their thalweg which is a line
connecting the lowest points along the stream channel. Sometimes the line midway the two
banks is taken as the thalweg. In many cases, the banks and the lowest points are not always very
distinct and the map scale does not permit these to be marked accurately.

8.3.2 The Drainage Network

Drainage network, as the name implies, is formed by orderly joining the streams in a basin. In a
network representation, the size of the streams is not important. In hydrological studies, several
systems have been proposed to classify the streams, for example, the Horton, the Stahler, and the
Shreve system. In the Horton’s scheme which is frequently followed, any elementary stream is
said to be of order 1, any stream with a tributary of order 1 is said to be of order 2, and any
stream with a tributary of order x is said to be of order x + 1. While mapping the streams, usually
the information is picked up from the Survey of India maps at 1:5000 scale. The definition of the
smallest streams is often subjective.

The first-order streams are the channels without any tributary; they receive all the flow
from surface overland route. Two first order channels join to form a second-order channel. A
second-order channel receives flow from the first-order channels which form it and the overland
flow. Evidently, a second-order channel carries more flow than a first-order channel. A third-
order channel is formed by the junction of two second-order channels; it receives flow from the
two second-order channels that form it, from direct overland flow, and from first- or second-
order channel(s) that might join it. Thus, a stream of any order has two or more tributaries of the
next lower order and so on. This scheme of ordering of streams is referred to as Horton-Strahler
ordering scheme.

8.3.3 Stream Profile

The stream profile is the variation in elevation of the points of the stream thalweg as a function
of their distance from the origin, which is generally taken as the confluence of the stream with a
larger stream or as its mouth. On such a profile, a certain number of topographical features may
be present such as rapids, waterfalls and changes of slope that frequently mark the boundary
between two reaches with different geologic controls.
The average slope of a stream is the difference in elevation between its highest point and
its confluence or mouth divided by its total length. Slopes of the various segments of the river are
required for hydraulic models, flood routing hydropower and morphological studies.
Figure 8.1 shows examples of stream profiles of a Himalyan river and its tributary. Such
a diagram gives an over view of the variation in slope in the drainage network and is very useful
in planning studies.
Elevation (m) 

Distance (km) 

Fig. 8.1 Stream profiles of a typical Himalyan river and its tributary.
8.4 Area Data
The basin or catchment or watershed area at a point is defined as the area that receives
precipitation and other inputs and, after hydrological processes contributes to runoff at that point.
The watershed boundary or the ridge, directs any precipitation falling within its area towards the
outlet, whereas any precipitation falling outside the boundary drains to a different outlet. The
watershed is usually defined by using contour maps small scale.
The basin perimeter is measured in a GIS or with a curvometer. On a paper map, it is
determined by planimeter. The measured perimeter is a function of the scale and accuracy of the
maps or photographs, the quality of the curvometer, and the care taken in its use Many GIS have
routines to automatically delineate catchment areas by using digital elevation models.

A basin’s physical characteristics are the soil types, the land use and land cover (for
example, crops lakes, swamps, or glaciers), and the type of land use (for example, rural or
urbanized areas, lakes, or swamps). These physical features may be compiled as layers within a
GIS.

8.4.1 River Cross-section

The cross-section of a river at a given location shows the variation of its bed elevation across the
width of the river or in a direction perpendicular to the flow. Cross-section area data are input in
a variety of studies, such as river morphology, rating curve analysis, flood routing, and sediment
transport. Together with flow velocity, cross-section area is used to compute river discharge.
In India, cross-sections are typically measured at stream gauging sites, hydraulic
structures such as dam or barrages, and at bridges. For hydrologic studies (particularly flood
routing and forecasting), cross section data are needed at closely spaced locations and places
where the stream profile changes significantly. If a channel is subjected to high rates of erosion
/deposition, frequent cross-section measurements are required.

8.5 Volumetric Data

Volumetric data primarily pertain to water storage and in less number of cases to sediment
storage and groundwater storage. Surface storage related data either pertain to the storage
capacity of existing or proposed reservoirs or the volume of water stored in a reservoir at a given
time. Bathymetric methods are used to determine storage of an existing lake or reservoir.
Hydrographic surveys are carried out to determine the volume of sediment deposited in a
reservoir.

Ordinary maps do not give bathymetric data on lakes and reservoirs. Bathymetric survey
is carried out by using a boat and sounding by using echo sounders or tapes. The depths are
referenced to the national or an arbitrary datum. Volume is computed by employing the
trapezoidal, prismoidal, or any other suitable formula.
8.6 Topographic Maps
A topographic map is a detailed representation of the objects present on the surface of the earth
at a particular scale. Conventionally, topographic maps show the land features such as roads,
railway lines, power and other utilities, rivers and lakes, habited areas, etc. Elevations and
changes in it are shown by means of contour lines. Contours are the curves that connect points of
the same elevations.

Topographic maps provide the information about a terrain and thus are very useful for
water resources studies. The Survey of India has carried out extensive survey for almost the
whole India and prepared topographic maps at various scales. To identify a map of a particular
area, a map numbering system has been adopted by Survey of India. The system of identification
is as described here.

For the purpose of nomenclature or reference of the maps, an International Map Series (in
the Latitude range 4° N to 40° N and Longitude range 44° E to 124° E) at the scale of
1:1,000,000 is considered as the base. This base is divided into grids of 4° x 4° and these are
numbered from 1 (at the extreme north-west) to 136. Only land areas are covered and if any 4°
square falls completely in the sea, the area is not covered.

To refer to any Indian Topographic map, each 4° x 4° region is further divided into 16
grids, each covering 1° latitude x 1° longitude. The 1° x 1° grid in the North-West corner is
assigned the letter A and the grids are assigned consecutive letters, column-wise. Thus the letter
for the last grid on the South-East corner will be P. Due to geographic coverage, these maps are
also called degree sheets and are on a scale 1:250,000. Contours are drawn in these maps at
intervals of 100m. To refer to a particular map, the number of the base map grid and the letter
code is used, for example 39N (Fig. 8.2). Approximate area covered by each such map is 11140
sq. km. Maps in this scale are highly useful for planning studies from different sectors, viz.,
water, transportation, administrative.
Fig. 8.2 Nomenclatu
ure scheme of
o 1:250,000 topographicc maps.

The
T 1:250,000 scale maps or degree sheetss are fuurther sub-divvided in twoo ways:
1) Each
E sheet iss divided intto four partss (2 rows byy 2 columnss), each of 330' latitude x 30'
lo
ongitude (1:100,000) deesignated by y cardinal ddirections N NW, NE, SW W, and SE. Such
sh
heets are ideentified, for example
e as 39
3 P/NE (Figg. 8.3).

Fig. 8.3 Nom

menclature sccheme of 1:1
100,000 topoographic maaps.
 2) Degree
D sheets are also diivided into 16
1 sheets (4 rows by 4 ccolumns), eaach 15' latituude x
15' longitudee (scale 1:500,000) and numbered
n fr
from 1 (at tthe north-weest corner oof the
particular deg
gree sheet) to
t 16 column nwise. Thesee are identiffied, for exam
mple, as 39 B/14
(F
Fig. 8.4).

Approximate
A area covereed by each mapm is 700 ssq. km. Theese are generral purpose maps
an
nd are used by adminisstrators, plan nners, and eengineers. Inn fact, it is tthe most poopular
map
m series for all activitiees of the Gov
vt.

Fig. 8.4 Nom

menclature sccheme of 1:5
50,000 topoggraphic maps.

Each
E 1:50,00
00 scale sheeet contains four (2 row
ws by 2 coluumns) 1:25,0000 sheet (77' 1/2
latitude x 7' 1/2 longitude) which
h are numberred NW, NE E, SW, and SSE. Such sheeets are idenntified
as 53 O/114/NE. Apprroximate areea covered by y each such map is 175 sq. km. Thiss is more dettailed
map and is used for planning
p andd developmeent purposes..

Most
M topograaphic maps are fairly old these maay have to bbe updated aas required for a
particularr project. Fo
or large scalee maps, furth
her surveyingg needs to be carried outt.

A recent deveelopment prooviding topo

ographic datta is the digittal elevationn model prodduced
at 90 m resolution byb the Shutttle Radar To opography MMission (SR RTM). SRTM M consistedd of a
specially
y built radar system that was sent to
o space onbooard the Spaace Shuttle E Endeavour dduring
an 11-day mission in n February of
o 2000. SRT TM obtainedd elevation ddata on a neaar-global scaale to
generate the most complete high-resolution digital topographic database of Earth
(http://www2.jpl.nasa.gov/srtm/). The level 2 digital elevation model, currently available only
for the United States, has a horizontal accuracy of 30 m and vertical accuracy of 18 m.

CARTOSAT – 1 is the first Indian Remote Sensing Satellite capable of providing in-orbit
stereo images. The images can be used for cartographic applications. Cameras of this satellite
have a resolution of 2.5m (http://www.isro.org/satellites/earthobservationsatellites.aspx). The
Cartosat – 1 provides stereo pairs required for generating Digital Elevation Models, Ortho Image
products, and Value added products for various applications of Geographical Information System
(GIS).

8.6.1 Soil and Land Use Maps

Maps of soil types and hydrologic properties and land-use/land cover are another important
spatial data for hydrologic analysis and design.

An institute under the Indian Council of Agricultural Research (ICAR), the National
Bureau of Soil Survey and Land Use Planning (NBSS&LUP) situated at Nagpur is the premier
institute for soil related activities such as classification, mapping, uniform nomenclature, and
interpretation. Publications of NBSS&LUP include State soil maps (1:250,000 scale), District
soil resource maps (1:50,000 scale), and District land use atlas. More details can be obtained at
http://www.nbsslup.in/.

Maps of land use are prepared and maintained by Agriculture Department and revenue
authorities. These maps are extensively used in planning activities as well as in many other
sectors including water.

8.7 Emerging Technologies for Spatial Data Acquisition and Management

The acquisition and management of spatial data using the conventional means (paper maps) is
difficult, tedious and time consuming. It suffers from the following drawbacks:
 Even relatively simple tasks such as making a copy of a big size coloured map are
difficult. Distortions may be introduced in photocopying a large map.
 When the number of maps increases, storage of maps becomes difficult and requires
large infrastructure.
 Hardcopy maps tend to deteriorate with time.
 Large resources are needed to update the paper maps at desired frequency.
 Analysis of spatial data in hardcopy map form is difficult.

We will now briefly describe two proven technologies that are being increasingly used to
collect and manage spatial data. These are geographical information systems and remote sensing.
8.8 Remote
R Senssing
The termm Remote Sensing
S (RS)) denotes th he acquisitioon of data aabout an obbject withoutt any
physical contact betw ween the object and the sensing
s deviice. These daays, the termm RS techniqque is
chiefly used
u to denoote the acquiisition and analysis
a of ssatellite dataa. The main advantage oof the
RS technnology is thaat it providees a broad perspective
p oover a large area. One ccan "see" beeyond
visible electro-magn
e netic (EM) radiation baand, and daata of inacccessible areaas can be eeasily
obtained. Fig. 8.5 sh hows the EM M spectrum. Remote sennsing techniqques use the EM spectruum to
almost itts entire rangge. The interraction of EME radiationn with an objject can reveal a tremenndous
amount of
o informatio on about the object: Whaat is it? Wheere is it? Whhat are its phyysical properrties?
What arre its spatiaal relationsh hips with th he surrounddings? It is also possiible to infer the
characterristics of a to
op thin layerr of the earth
h’s surface.

Fig. 8.5
8 The elecctromagneticc spectrum.

Remote
R sensiing works on n the basic principle
p thatt depending upon the phhysical propeerties,
each object on the Earth
E surfacee reflects, emits
e and abbsorbs varyinng intensitiees of radiatiion at
different EM waveleengths. Usin ng informatiion from theese wavelenngth ranges, it is possibble to
nate between
discrimin n different types
t of grouund objects (e.g., waterr, dry soil, vvegetation, rrocks,
etc.) and
d map theirr distribution n on the grround. Inforrmation aboout the objeect concernned is
obtained by a senso or system lo ocated on a satellite, w which receives electrom magnetic radiiation
either em
mitted by the object or the radiation in
nteracted wiith the objecct.

The
T source of all sign nals collectted by mosst remote sensing insstruments iss the
electromagnetic radiiation (EM).. Sensors th hat rely on tthe Sun to generate alll the EM ennergy
needed too image terrrestrial surfaaces are callled passive ssensors. Sennsors that geenerate theirr own
energy arre called acttive sensors.
 Remote sensing is a powerful technique to explore, map and manage the earth resources.
The images collected are processed and interpreted to produce useful data for application in a
wide variety of fields. The main advantage of remote sensing is that it gives a
broad perspective or bird’s eye view of the area. Depending upon the resolution, small to
very large area can be covered in the image. Further, one can "see" the objects in
wavelengths beyond the visible bands. Using remote sensing, images of
inaccessible regions can be easily obtained.

Besides conventional, other uses of remote-sensing in hydrology include sensing of near-

surface soil moisture using airborne natural gamma or satellite based passive microwave
techniques and measurement of land surface temperature to determine evapotranspiration. Leaf
area index measurements may also be used to estimate evapotranspiration. Remote sensing of
water quality also offers considerable promise as new satellites and sensors are developed. Water
bodies that are affected by suspended sediment, algae or plant growth, dissolved organic matter,
or thermal plumes undergo changes in spectral or thermal properties that may be detected by
satellite sensors.

Airborne optical devices (Lidar) are now being used to determine topography more
accurately and at lower cost than conventional aerial photography.

8.8.1 Physical Properties of Electromagnetic Energy

The electromagnetic spectrum, as shown in Figure 8.5 ranges from the shorter wavelengths
(including gamma and X-rays) to the longer wavelengths (including microwaves and broadcast
radio waves). The (UV) portion of the spectrum has the shortest wavelengths that are practical
for remote sensing. Some Earth surface materials, primarily rocks and minerals, emit visible light
when illuminated by ultraviolet radiation. The light, which our eyes can detect, is part of the
visible spectrum which covers a wavelength range from approximately 0.4 to 0.7 m. The
longest visible wavelength is red and the shortest is violet. The infrared (IR) region covers the
wavelength range from approximately 0.7 m to 10.0 m.

8.8.2 Interaction of Electromagnetic Energy with Objects

Electromagnetic energy that reaches a target will be absorbed, transmitted, and reflected. The
proportion of each depends on the composition and texture of the target’s surface. Much of the
remote sensing is concerned with reflected energy.

Reflection occurs when radiation is neither absorbed nor transmitted. The reflection of the
energy depends on the properties of the object and surface roughness relative to the wavelength
of the incident radiation. Differences in surface properties allow to distinguish one object from
another.

Absorption, transmission, and reflection are related to one another by

 EI = EA + ET +ER (8.1)

where EI = energy incident on an object, EA = absorbed radiation, ET = transmitted energy, ER

= reflected energy.

Reflectance is simply a measurement of the percentage of incoming or incident energy that a

surface reflects

Reflectance = Reflected energy/Incident energy (8.2)

where incident energy is the amount of incoming radiant energy and reflected energy is the
amount of energy bouncing off the object.

Reflectance is a fixed characteristic of an object. Surface features can be distinguished by

comparing the reflectance of different objects at different wavelengths. Reflectance comparisons
rely on the unchanging proportion of reflected energy relative to the sum of incoming energy.
This permits the distinction of objects regardless of the amount of incident energy. Unique
objects reflect differently, while similar objects reflect differently only if there has been a
physical or chemical change.

Electromagnetic energy arrives at the sensor as a continuous stream of radiation. The

incoming radiation is sampled at regular intervals and is converted from a continuous analog to a
digital number and then assigned a value. The value given to the data is based on the hardware
on-board the satellite. The computer translates the sequence of binary numbers, given as ones
and zeros, into a set of instructions with only two possible outcomes (1 or 0, meaning “on” or
“off”). The binary scale that is chosen (i.e., 8 bit data) will depend on the level of brightness that
the radiation exhibits. The data is stored in the computer on the satellite and is sent to Earth.

8.8.3 Basic Components of a Remote Sensing System

The overall process of remote sensing can be divided into five components (Figure 8.6). These
components are: 1) an energy source; 2) the interaction of this energy with matter present in the
atmosphere; 3) subsequent interaction with the ground target; 4) energy recorded by a sensor as
data; and 5) data displayed digitally for visual and numerical interpretation.

o Energy Source or Illumination (A) – this is the basic requirement for a

remote sensing system; most frequently Sun is the energy source.
o Radiation and the Atmosphere (B) – energy comes in contact and interacts
with the atmosphere as it passes through. This interaction takes place
again when the energy travels from the target to the sensor.
o Inter
raction wi
ith the Target (C) - once t
the energy
y makes it
ts way to the
targe
et through
h the atmo osphere, it
i interaccts with tthe targett in a maanner
depen
nding on the
t propert
ties of booth the tar
rget and t
the radiatiion.
o Recor
rding of Energy by b the Seensor (D) - after the ene ergy has been
scatt
tered by, or emitt ted from the targe et, a sen nsor (remoote - nott in
conta
act with the tar rget) col llects an d records s the el lectromagn
netic
radia
ation.
o Trans
smission, Reception
n, and Pro ocessing (EE) - the energy reecorded by the
senso
or is transmitted, , often ini electro
onic formm, to a receiving and
proce
essing sta
ation wherre the datta are proocessed in
nto an imaage (hard copy
and/o
or digital
l).
o Inter
rpretation
n and Anaalysis (F) - the p processed image is s interpreeted,
visua
ally and/o
or digitall
ly, to exttract infor
rmation ab
bout the taarget
o Appli
ication (G) - the final el lement of the remo ote sensinng processs is
eved when the info
achie ormation isi extract
ted from the image ery about the
targe
et to bett
ter undersstand it, reveal somme new infformation, or assist in
solvi
ing a part
ticular pro
oblem.

8.8.4 Re
eflectance
e Characterristics of
f Different
t Objects
Reflecta
ance is the perce entage of energy reflected to the total en nergy
incident
t on a body.
b It is the ratio
r betw
ween radiation off f the surrface
(outgoin
ng) to the
e total raadiation incident
i on
n the surf oming). We see
face (inco
an objeect green because only gre een energyy is refl lected inn the vissible
spectrum
m. Pure black
b or pure
p white
e are not colours. Other col lours such
h as
magenta,
, violet, yellow, cyan
c etc. are the mi
mixing of t ic colours RBG
three basi
in diff
ferent propportions. Reflectancce charact
teristics of some oof the fea
ature
are show
wn in Figu
ure 8.7.
Fig. 8.6
6 A typic
cal remote sensing data
d collec
ction and analysis s
system.

Spectral Reflectance Curves

A surface feature’s color
c can bee characterizzed by the ppercentage oof incomingg electromaggnetic
energy (iillumination)) it reflects at
a each waveelength acrooss the electrromagnetic sspectrum. This is
its spectrral reflectancce curve or “spectral
“ signnature”; it iss an unchangging propertty of the matterial.
For exammple, an objeect such as a leaf may reeflect 3% off incoming bblue light, 100% of greenn light
and 3% of o red light. The amoun nt of light itt reflects deppends on thee amount annd wavelenggth of
incoming g illuminatio on, but the peercents are constant.
c

Itt is helpful too be familiaar with the baasic spectrall characteristtics of greenn vegetation,, soil,
and wateer as these arre of immensse use in watter sector.

Spectral Reflectancee of Green Vegetation.

V Reflectance
R of green veggetation is low in the visible
portion of
o the spectrrum owing to chloroph hyll absorptiion, high in the near IR R due to thee cell
structure of the plantt, and lowerr again in thee shortwavee IR due to w water in the cells. Withiin the
visible po
ortion of thee spectrum, there
t is a local reflectannce peak in tthe green (00.55 µm) bettween
the blue (0.45
( µm) an nd red (0.68 µm) chlorop phyll absorpption valleyss.

Spectral Reflectancee of Soil. Soiil reflectancce typically iincreases wiith wavelenggth in the visible
portion of
o the spectruum and thenn stays relatiively constannt in the neaar-IR and shhortwave IR,, with
some loccal dips due to water abssorption at 1.4
1 and 1.9 µ µm and due to clay absoorption at 1.44 and
2.2 µm.

Spectral Reflectancee of Water. Spectral

S refllectance of clear water is low in alll portions oof the
spectrum
m and decreaases as wav velength inccreases beyoond green bband. In neaar-infrared bbands
water ab
bsorbs almosst are the inncoming rad diation. Refllectance incrreases in thhe visible poortion
when maaterials are su
uspended in the water.
Fig. 8.7 Reflectance characteristics of some commonly found objects.

8.8.5 Remote Sensing Platforms

Platform is a stage to mount the camera or sensor to acquire the information about a target under
investigation. These platforms, typically satellites, freely move in their orbits around the earth,
and the entire earth or any part of the earth can be covered at specified intervals. The coverage
mainly depends on the orbit of the satellite. Enormous amount of remote sensing data is collected
through these space borne platforms, and this is one reason why remote sensing has gained
international popularity. Space borne remote sensing provides the following advantages:
 Large area coverage;
 Frequent and repetitive coverage of an area of interest;
 Quantitative measurement of ground features using radiometrically calibrated sensors;
 Relatively lower cost of data per unit area of coverage.

To acquire remote sensing data, polar orbiting or sun-synchronous satellites are used. A
satellite orbit in which the orbital plane is near polar (pass through N and S poles) and the
altitude is such that the satellite passes over all places on earth having the same latitude twice in
each orbit at the same local sun-time. Through these satellites, the entire globe is covered on
regular basis and it gives the repetitive coverage on periodic basis. Satellites such as Landsat
series, SPOT series, IRS series, may be grouped in this category.

8.8.6 Digital Image Processing

A remote sensing image consists of a number of picture elements or pixels of the same size. Each
pixel represents the spectral response of a small square shaped area on the ground and has certain
spatial and spectral properties. The spatial property defines the ground coverage of the pixel. The
size of the area depends on the resolution of the on-board camera. For example, the resolution of
LISS III camera of IRS-1C satellite is 23.5 m resulting in a pixel size of 23.5m*23.5m. Smaller the
pixel size, more details of the target are mapped. Associated with each pixel is a number known as
digital number which is the integrated radiance response of the ground covered by that pixel.

An image can be analyzed using either the visual or the digital techniques. Visual interpretation
has many limitations and digital image processing (DIP) techniques are almost universally
applied these days. DIP involves the processing of data of an image to improve the image qualities
or to enhance the features of interest with the aid of a computer. The process helps in maximising
clarity, sharpness, and details of objects of interest and leads to better information extraction. It
improves the image’s interpretability. Image processing operations are carried out to remove noise
from the data and enhance certain features based on their spectral response. DIP is a broad subject
and may involve procedures that can be simple as well as quite complex. Basically, each pixel of
an image is mathematically manipulated and the operation may involve more than one image. The
results of computations for each pixel are stored and form a new image. The new digital image
may be subject to further analysis, may be stored or printed.

8.8.7 Indian Remote Sensing Satellite Program

Remote sensing is an important part of the Indian Space Program and the Department of Space
(DOS), Government of India, is main body to exploit space technology for remote sensing
applications. Under the Indian Remote Sensing Satellite (IRS) program, the Indian Space
Research Organisation (ISRO) has launched a series of land observation satellites. The first
generation remote sensing satellites were IRS-1A and IRS-1B launched in 1988 and 1991
respectively. These were placed in near-polar, sun-synchronous orbit, with repetitive time of 22
days. The satellites had two Linear Imaging Scanning Sensors (LISS-I and LISS-II) for
providing data in four spectral bands: Visible, Infra Red (IR) and Near Infra Red (NIR). Their
ground resolutions were 72.5 m and 36.25 m respectively.

IRS-1C and IRS-1D were launched in 1995 and 1997 and were the first of the second generation,
operational, multi-sensor satellite missions with better resolution. These were placed in near-polar,
sun-synchronous orbit and had resolution of 5.2-5.8 m and 188 m, respectively. The repetitive time
of the satellites was 22 days, and on-board sensors had with Panochromatic (PAN) and LISS-III
cameras.

IRS-P3 was launched in 1996 on PSLV-D3, had two imaging sensing sensors and one non-
imaging sensors, viz. WiFs, with a resolution of 188 m and swath of 810 km.

IRS- P4, also called as Oceansat, primarily caters to oceanographic applications. The indigenous
Polar Satellite Launch Vehicle (PSLV) launched this satellite in 1999 into polar Sun-synchronous
orbit. The satellite has a high receptivity of 2 days. The payload includes an Ocean Colour Monitor
(OCM), a Multi frequency Scanning Microwave Radiometer (MSMR) and solid state memory for
recording data outside the visibility of a ground station.

IRS- P6 or RESOURCESAT-1 was launched into a polar sun synchronous orbit on October 17,
2003. The satellite carried LISS-3, AWiFS (Advanced Wide Field Sensor), and LISS-4 cameras.
LISS-3 camera had improved spatial resolution in shortwave infrared (SWIR). LISS- 4 was the
new high spatial resolution sensor. The sensor operates in three spectral bands with 5.8 metre
spatial resolution and steerable up to ± 26˚ across track to obtain stereoscopic imagery and
achieve five day revisit capability. LISS-3 operates in three spectral bands in VNIR and one
SWIR band; and an AWiFS operating in three spectral bands in VNIR and one band in SWIR
with 23.5 m spatial resolution. The payloads of this satellite have greatly aided crop/vegetation
and integrated land and water resources related applications.

IRS-P5 (CARTOSAT-1) was launched on May 05, 2005. It has an improved sensor system that
provides 2.5 m resolution with fore-aft stereo capability. This mission caters to the needs of
cartographers and terrain modelling applications. The satellite provides cadastral level information
up to 1:5000 scale and will be useful for making 2.5 m contour maps.

Researchers, practitioners and students in India are extensively using remote sensing
products to obtain spatial data required for water resources studies these days.
8.9 Geographical Information Systems
A Geographic Information System (GIS) is a computer-based system that is used in input,
storage, analysis manipulation, retrieval, and output, of spatial data. These systems consist of
computer hardware and software. GIS are increasingly being used in applications in natural
resources management. These days GIS are also being integrated with remote sensing and GPS.

GIS has origin in manual overlay operations done as early as in 1920s. In 1958, computer
based cartography initiated in USA which culminated in development of first general purpose
mapping software in 1960s. Canada GIS is also cited as first GIS and was developed around the
same time. Present day popular commercial/ open source GIS are: ArcGIS, GRASS, MapInfo,
ERDAS, IDRISI, ILWIS etc. GRASS (Geographic Resource Analysis Support System) is high-
end open source software. Arc GIS is modular high-end commercial software. ERDAS and
ILWIS packages have image processing and GIS capabilities.

8.9.1 Advantages of GIS

 After a map is available in digital form, it is easy to update or modify it.
 Conventionally, overlays are manually prepared. This process is cumbersome and error
prone. GIS allows digital overlaying a number of maps.
 Conventionally, maps are browsed to retrieve information. In GIS retrieval of
information is very fast, and much easier.
 In conventional method, hardcopy output is prepared. Updating of such maps is difficult
and the map is to be redrawn. This process is simple in GIS- the map can be easily edited
and printed.
 Annotation is clumsy in hardcopy maps. Thus, while retrieving information ambiguity
may arise. Also all features may not be annotated in paper maps. Thus, attribute
information, e.g. names for some of the features are lost in paper products. In GIS,
information is stored in tables and is linked to geographic features and thus is not limited
by availability of annotation space/ color/ symbol, etc.
 If multiple maps are prepared for same area, e.g. watershed, land use, geomorphology,
common boundaries are drawn manually and may not match in different maps. In GIS
common boundaries are digitized once and are available to all layers. Once GIS map
layers are prepared, any number of maps can be designed.
 Storing of a large number of paper maps is difficult; maps deteriorate with time. Digital
maps can be stored in a much compact way.

GIS systems are now available for standard computers in practical, low-cost formats. The
main cost factor now resides in database compilation, and training and updating of technical
staff. Data capture or input in GIS is costly but it is one-time affair. Commercially available
paper maps may be cheaper than GIS layer. Use of GIS requires investment in computer
hardware, software, and training. GIS handling requires trained manpower. GIS software should
have proper functionality as desired in an application. For example, for hydrological modeling,
DEM analysis should be available in the package.

8.9.2 Spatial data representations

Spatial data are represented mainly in two ways in GIS: raster and vector. These data
representations can be transformed from one form to the other, albeit with some information loss.
In raster, spatial data are structured as grid of cells or pixels. Their row and column numbers
addresses the cells. In many distributed hydrological models, spatial data and hydrological
computations are done in this form. This is a native representation for remotely sensed data.
Satellite data are captured/ resampled as pixels (picture elements) and information is extracted
through digital image processing. In vector model, spatial data are represented as coordinate
points. For example, point data is represented as a pair of coordinates. A straight-line is
represented as two pair of coordinates, representing end points of the line. A curved line is
represented as finite line segments. Area data are represented as line data with some additional
information e.g. centroid, adjacent areas, etc.

In raster data, points and lines are represented with finite area and finite width and thus is
not a natural representation. Lines have stepped or zagged appearance. In vector model, points
and lines have infinitesimal area and width respectively. Lines are smooth curves. Raster data
require large storage space. Vector data require small storage space. Thematic maps prepared
from remotely sensed data are available is raster form and are often processed as such. Many
hydrological models use both the representations. For example, thematic maps of catchment
variables and hydrometeorolgical measurements are prepared in raster form. Stream network is
processed in vector form etc. In raster form, value of many catchment variables is scale
dependent. For example, average slope of catchment reduces with increase in raster grid size. In
most GIS, the representations coexist. For example, it is better to capture spatial data from
conventional thematic maps, through visual interpretation of remotely sensed data in vector
form. Thematic maps from digital processing of satellite data may be obtained in raster form.

Topology: Method of representing vector data is called its topology. A line consists of two nodes
and one or more vertices. Nodes are end points of the line. Lines also have directions. Thus,
nodes are referred as ‘from node’ and ‘to node’ depending on direction of the line. Areas are
represented by ‘left area’ and ‘right area’ of each line.

Digital Elevation Model (DEM)

Topographic elevation data in GIS are called DEM. These are represented in GIS in various
manners namely contours, raster, and TIN (Triangulated Irregular Network). Contours are
conventional representations of DEM and are used in topographic maps. Contours are equal
elevation lines. Normally, equal interval contours are drawn in topographic maps to represent
topography. For example Survey of India (SOI) maps at 1:250,000, 1:50,000 and 1:25,000 scale
have contours at 100, 20 and 10 m elevation interval. DEMs are used to derive topographic
information such as slope, aspect and are also used in hydrological calculations, e.g. stream
network delineation, topographic index, delineation of catchment area, etc.

TIN
In TIN model, elevations at the vertices of triangles are used to compute elevation at interior
points of the triangles. Using elevation of the vertices of a triangle, a planner or higher order
surface can be fitted. The surface can be used to derive elevation at points inside the triangle.
TIN model requires Delaunay triangulation. In this, constituent triangles are as equilateral as
possible. Circum-circles of the triangles include no other point of the triangulation. Triangulation
is performed first by constructing Voronoi diagram (Thiessen polygons). Points included in
adjacent polygons are joined to create Delaunay triangulation. Voronoi diagram is drawn using
proximity analysis.

Interpolation
Interpolation is a technique of determining unknown value of a variable at location from known
values at other locations. Interpolation can be used for any spatial variable, e.g. topographic
elevation, pH, SAR, pollutant concentration, groundwater depth and level, population etc.
Known values can be at point, line of area locations. Point data can be spot heights, pH, pollutant
concentration etc. Line data can be topographic contours, etc. Area data can be population
density in regions, etc.

Thiessen polygons or nearest neighbor

This is popular method of interpolating rainfall values from point rainfall measurement at rain
gauges. Generally, point rainfall stations are limited in number. To determine basin wide average
rainfall, the method of interpolation is used. The Thiessen polygon diagram is prepared by
proximity analysis. For measuring average rainfall in the catchment, weights for each rain gauge
station are area of Thiessen polygon surrounding that station divided by the total catchment area.

Distance weighted averaging

In this averaging, a weight of inverse of distance function is used. Distance function is nth power
of distance. Thus, more weight is assigned to stations closer to the interpolation location.

Surface fitting: Here, n-degree polynomial surface is fit between selected known values. The
surface can be used, among the other application, to interpolate values.

Krigging: Variation of spatial variables can be partitioned in three components, namely drift or
structure, small variations and random noise. First component depicts general trend of the data.
Second component represents small variations from the general trend. These variations are
random but spatially autocorrelated. Third component depicts random values that are not
spatially autocorrelated. Kriging is best suited for interpolation of pollutant concentration,
geological and mining variables, e.g. grade of ores, etc. For these data, single smooth
mathematical equations are not suitable. The technique is based on assumption that values in
neighbourhood have generally higher correlated. Apart from the estimate of values, error
estimates are also provided in kriging technique. In presence of large random noise in data, good
semivariogram is not obtained and this results in deterioration in interpolation quality.

Georeferencing: Earth is a three dimensional surface. In maps, this three-dimensional surface is

transformed in to flat surface. For the transformation, map projections are employed. Locations
on the map are drawn using Cartesian coordinates obtained through map projections. Geographic
graticules (latitudes and longitudes) are later drawn in maps. Sometimes, Cartesian coordinate
grids are also drawn on maps. In georeferencing earth coordinates are assigned to spatial data.
Either lat/long or Cartesian coordinates can be used in georeferencing maps in GIS. Cartesian
coordinates allow measurements, e.g. area and lengths and are thus frequently used. Geographic
coordinates do not allow measurements. A map can also be referenced without using a map
projection. In such case, it is difficult to integrate GIS layers obtained from different sources.

A map which is to be geo-referenced is called the source map and the reference map is
the map which has known coordinates. Points whose reference coordinates are known and which
are clearly identifiable on both the source and reference maps are known as control points. For
coordinates of control points in two maps, coefficients of a polynomial transformation equation
are estimated.

8.9.3 Map Projection

Map projection is transfer of positions on earth to corresponding points on a flat sheet of paper.
Because of the shape of the earth, this transformation involves approximations and is not
distortion free. Distortions occur in lengths, angles, shapes and areas. Scale is a ratio of length on
map to its counterpart on the earth. Since large size of features on earth surface, a scale is needed
to draw these features on a small sheet of paper. Earth shape is assumed spherical or spheroidal.
An intermediate plotting surfaces namely cylinder, cone or plane is used in projections.
Corresponding projections are called cylindrical, conical and azimuthal respectively.

Distortions occur in projecting earth surface on to intermediate plotting surface. Ideally,

areas distances, directions, angles and shapes should be preserved. In reality, few of these
properties are preserved. Based on application, choice is made as to which propertied are to be
preserved and appropriate map projection is selected. In areal distortions area of a figure may
increase or reduce. In linear distortion length and its curvature may change. In angular distortion
an angle may increase or decrease. In shape distortion, a square may become parallelogram,
rectangle or may have curved boundaries or both. A point may be distorted in to a line. In equal
area projections, area of a figure is preserved. In the process distortions are introduced in
distances and angles or shape of figures. In conformal projection, shape is preserved. In this
process, areas figures are distorted. Projections with these contrasting properties are called equal
area and conformal projections respectively. With different orientation of intermediate plotting
surface, it is possible to obtain different projections. Azimuthal projections are called polar,
equatorial or oblique depending on point of contact of plane falling on poles, equator or at
intermediate latitude. For conic and cylindrical projection based on orientation of axis of the
plotting surfaces, the projections are classified. When the axis coincides with earth’s polar axis,
perpendicular to it and lying in equatorial plane and is oblique to it, the projections are called
regular or equatorial, transverse and oblique respectively. The plotting surfaces can also be
tangential or secant to the earth surface. Normally for projection, mathematical approach is used.

8.9.4 GIS Operations

Input of data in a GIS database is either by digitization or import. Digitization can be done on-
screen or by a digitizer to create/ edit GIS objects in vector format. Input data are, sometimes,
available in GIS image formats. These data are converted to native format of GIS.

Storage: Geographic data are stored in GIS the native format of GIS. For one spatial data, many
computer files are created which contain different information. Attribute data are stored in Data
Base Management System and are linked to geographic objects. Storage of data in the form of
layers looks very attractive from water resources data. Different types of data, such as soil and
land cover, are stored in different map layers. GIS permits analysis of single or multiple layers
and various layers can be overlaid, one on top of another. From a water resources point of view,
spatial variation of data is important, e.g., the variation of soil hydraulic properties.

8.9.5 Spatial Data Analysis

Data analysis involves operations with geographic data and their attributes to obtain derived
information, generate query, statistics etc. broad categories and operations therein are as follows.

Statistics: for example, count, length, area, perimeter, shape, centroid, etc. for geographic objects
can be derived in GIS. For continuous surfaces, average, standard deviation, maximum,
minimum, etc. are derived. Summary operation produces zonal statistics for a map. For example,
land use statistics for watershed in a basin can be generated.

Mathematical operations: Mathematical operations, e.g., addition, subtraction, multiplication,

division, exponential, logarithm, absolute, truncation, round off, negative, trigonometric
operations can be performed in GIS. For example various component maps in the universal soil
loss equation, namely R, K, L, S, C and P can be prepared as different largess and multiplied
using multiplication operation.

Logical operations: Logical operations, namely or, and, not, xor can be performed on maps.
Figure 8..8 shows log gical operatiions. For exaample, landuuse= agricullture and pH
H >= 8 will rresult
in salt aff
ffected agricu
ulture area.

Conditionnal: If-then--else conditio

onal operatio
onal can be performed oon maps. Foor example, ‘‘if 50
< return period
p <=1000 and land use=
u residen
ntial, then vuulnerability=
= high else vuulnerability=
=low’
condition
n gives floodd vulnerability map.

Figure 8.8 Lo
ogical operaations commonly used inn GIS.

Conditionnal: In this operation, all combinaations of claasses in tw

wo maps aree obtained inn the
resulted map.
m For ex xample, overrlay of soil hydrological
h soil group aand land usee/ cover mapp will
provide soil-cover
s co
omplex map..

Classification: It con
nverts valuees into interv
val. A contiinuous surfaace is input and area m map is
output fo
or the operattion. In the output area map, isolinees, i.e., line of equal vaalues, enclosse the
area. Exaamples of vaarious isolin
nes are contoours, isohyette, isotherm,, isobar etc.,, which reprresent
topographic elevatiion, rainfalll, temperatture, pressuure, respecctively. In reclassificaation;
informatiion of geogrraphic objectt is changed.. For examplle soil seriess map may bbe changed too soil
pH map.

Search/ buffer:
b The operation iss similar to distance, exxcept that at a specified distance ann area
geographhic object is created.

Neighborrhood: Inforrmation in eight

e bor, their loccations and statistics, ee.g. mean, m
neighb mode,
median, minimum,
m maximum,
m standard deviaation, coeffi cient of variiation, etc. arre extracted..
Aggregate: Cell size of raster maps can be changed in fractions of half, one fourth etc. using
functions e.g. mean, predominant, minimum and maximum.

Query: Query is done by attributes or geometry. In query by attribute, a logical expression is

written in attributes and result is obtained. For example land use=agriculture will select/ display
agriculture areas. In query by geometry, objects are selected on screen to view their attributes.
After the requisite data are stored in a GIS database, it is easy to answer complex queries like
what areas in the catchment have forest on shallow soil with 3% slope?

Output: GIS output may be obtained as paper maps or as a picture which may be shared or
printed or inserted in a report or presentation. The output maps may contain various cartographic
elements namely title, legend, graticules or grids, north arrow, scale, annotations, notes, etc. In
one output more than one GIS layer may be included apart from cartographic elements. When
design is saved, it only contains reference to the layers. Thus, if a layer is modified and designed
output map is opened at a later time, the changes are reflected in the output.

The topography of a river basin may be represented in two different ways: as a digital
elevation model or as a triangulated irregular network (TIN). The digital elevation model is a
grid of elevation values that has regular spacing while TIN is a series of points linked into
triangular surfaces that approximate the surface. The spacing of points in TIN are non-uniform,
which allows points to be located on critical terrain features, roads or river banks. The accuracy
of such digital terrain models depends on the source of the data, the point density and
distribution, and other related data used in their development. Conventional contour maps may
be prepared from a digital elevation model or TIN. Orthophotos are images of the landscape
from which features can be referenced to one another. They are digital images produced by
processing aerial photography to geodetic control elevation data to remove all sources of
distortion. The image has the properties of scale and accuracy associated with a map. Such
images can be derived using airborne or satellite sensors.

8.9.6 Applications
Several hydrological software now have a GIS interface. GIS interfaces have been developed for
hydrological models such as HEC models, SWAT, etc. ArcInfo extensions such as Spatial
Analyst, 3D analyst, are useful for hydrological tasks. Script languages are also available to write
interfaces in GIS, for example, Arc Macro Language is the script language in ArcInfo.

GIS techniques facilitate input of spatial data to hydrological model. GIS are being
incorporated in hydrological models to extract and format distributed watershed data. Use of
DEM permits complete physiographic and hydrological depiction of basins. The efficiency of
handling large volumes of data means that more comprehensive and detailed maps, isolines and
themes can be prepared. This is a significant improvement in water resources management as
map preparation is often time-consuming and expensive.

GIS are becoming very common in the field of water resources assessment and
management. Many tasks of data collection, compilation, and interpretation can be facilitated by
means of GIS. In network planning and design, the ability to map quickly and display surface
water and related stations enables a more effective integration to take place. Network maps,
showing basins or stations selected according to record quality, watershed, or operational
characteristics, can be used for both short-term and long-term planning.

Groundwater potential and quality can be studied in GIS environment. Various layers
namely slope, geology, distances to drainage channel, tanks and lineaments, depth to water table,
depth of weathered zone can be overlaid and integrated on GIS environment to obtain
groundwater potential map. Similarly, layers of water quality variables may be created to obtain
quality map.

8.10 Closure
RS and GIS are powerful tools for spatial data collection and management. Spatial information
and its attributes can be stored, analysed and output efficiently through GIS. It has many
applications in natural resources and infrastructure management. GIS also has application in
varied industries namely power, transportation etc. Whereas in many applications, it can be
independently used, it also helpful in hydrological modeling in terms of data preparation. For
such applications, varieties of interfaces, stand-alone programs, embedded programs exist. These
increase productivity and reproducibility in modeling.

Table 8.1 Map projections

Projection Type (Plotting surface)
Equal area
Albers, one standard parallel Conic, tangent
Albers, two standard parallels Conic, secant
Lambert zenithal Azimuthal
Cylindrical equal area Cylindrical

Conformal
Lambert conformal, one standard parallel Conic, tangent
Lambert conformal, two standard parallel Conic, secant
Stereographic Azimuthal
Mercator (e.g. Universal Transverse Mercator or UTM) Cylindrical

References
Engman, E.T. and R.J. Gurney (1991). Remote Sensing in Hydrology, Chapman and Hall, London.
Lillesand, T.M. and Kiefer, R.W. (1994). Remote Sensing and Image Interpretation. Wiley, New
York.
Sabins, F.F. (1987). Remote Sensing Principle and Interpretation, W.H. Freeman & Co., San
Francisco.
 Module 9
Hydrological Information System

9.1 Hydrological Information System

9.1.1 Functions of a HIS
9.2 Designing a HIS
9.2.1 Assessing the needs of users
9.2.2 Data Transmission
9.3 Hierarchy of HIS
9.4 Data Storage and Retrieval
9.4.1 Management of historical data
9.4.2 Data storage and dissemination
9.5 Institutional and Human Resources development
9.6 Use of Computers and Internet in HIS
9.7 Closure
9.8 References

Keywords: Hydrological Information System, Design, Data, Transmission, Storage, Retrieval,

Dissemination, Human Resources

The objectives of this module are:

• To define a hydrologic information system and explain the need of it.
• To explain the key features of a HIS.
• To discuss other topics related with HIS.

9.1 HYDROLOGICAL INFORMATION SYSTEM

The occurrence of water shows great variability in space and time and requires that adequate
measurement networks are established to define spatial variability and that they are maintained
over a sufficient period of time to define temporal variability of a water variable. Management of
water services for domestic, industrial, agricultural and power generation - and protection from
the vagaries of floods and droughts, requires information on storages and fluxes of water.

A Hydrological Information System (HIS) consists of the physical infrastructure,

software and human resources to collect, process, store and disseminate data on hydro-
meteorological, hydrological, geo-hydrological and related variables. The physical infrastructure
includes the data observation networks, laboratories for analysis of samples, communication
systems, and data storage and processing centres. The human resources are the trained staff who
observe, key-in, process, disseminate the data and maintain the equipment computers, etc.

The primary role of a HIS (see Fig. 9.1) is to provide reliable data sets for planning,
design and management of water resource and for research activities. The system should function
in such a manner that it provides the information to users in time and in proper form. Sometimes,
the scope of HIS is extended to provide data to users on a real-time basis for short-term
forecasting or operational purposes.

The data collected for different hydrometeorological variables is called the raw or
observed data. The raw data have to be processed to ensure the reliability of the resulting
information. Both raw and processed data sets have to be properly stored -- processed data for
dissemination and raw data to permit inspection and revalidation in response to queries from
users. Note that the users have a central role in a HIS.

DATA FROM
OTHER SECTORS
HIS

USERS STORAGE
PROCESSED
DATA
 Policy Makers and PROCESSING
Planners
 Decision Makers
 Designers
VALIDATION
 Researchers
 Other Users RECORDING

Actions/Studies /Use DATA OBSERVATION

REAL
WORLD

Fig. 9.1 Schematic diagram of a hydrological information system.

Since most of water-related development activities are controlled by the government

sector in India, the main agencies that observe hydrometeorological data are the users themselves
(HP 2003). Therefore, the network and frequency of data collection are governed by their own
needs. Other data users, such as researchers, may find that data, if available, are inadequate in
spatial and temporal coverage, are of varying/inadequate reliability and are scattered at many
locations. Importantly, the data that is once missed is not available later on. To obviate all these,
there must be a strong linkage between data collectors and the end users and a periodic review of
the HIS.

Inadequate data may result in costly construction with an unnecessary high margin of
safety or with the equally costly consequences of failure. Ideally there must be a strong linkage
between data/information producers and users, who can be from within the same organization
and external users from other agencies, consultants, research scholars, etc.

9.1.1 Functions of a HIS

The functions of a Hydrological Information System are contained in the title itself.
Hydrological: Hydrology is the science of water and is concerned with its states, storages and
fluxes in location, time and phase. Hydrometry is the branch of hydrology which is concerned
with the measurement of these states, storages and fluxes in the water cycle.
Information: is data which has been and processed to remove errors and is presented in a
meaningful way. By definition, information serves a function and is created with some purpose.
Three key features of information are: reliability, availability and presentation.
System: The HIS is a logical and structured system to collect data which are subsequently entered
into the computer, checked and stored. In a HIS, data may be associated, related and combined to
provide information in a form suitable to users.

A Hydrologic Information System (HIS) is a means to manage current and historic

hydrological and related data in an organized form. The principles of a HIS are clearly reflected
in the title itself. Hydrometry is concerned with the measurement of states, storages, and fluxes
of hydrological variables in space, time, and phase. ‘Information’ is data which has been
manipulated and processed to give them meaning and purpose. By definition, information serves
a function and is created not simply because there is something to be measured or because of
curiosity. Three key features of information are: reliability, availability, and presentation.
Presence of the word system in the title shows that a HIS is not simply a data collection or
archive. It is a logical and structured system to collect, process, store, and disseminate water
resources data.

The activities under HIS can be broadly classified in the following categories:
i. Assessing the user needs;
ii. Establishing an observational network and operating it;
iii. Data collection, validation, processing, and reporting;
iv. Management of historical data;
v. Data transmission, storage, and dissemination; and
vi. Institutional and human resource development.

9.2 Designing a HIS

The primary role of a HIS is to provide reliable data sets for a variety of purposes: planning,
design and operation of water resources development projects; formulating policies of utilization
of water resources; and for use in research activities. The system should be able to provide the
requisite information to the users in time and in a usable format. Some modern HIS also provide
data to the users in (near) real-time. To perform these roles, the activities under a HIS can be
broadly classified in the following categories:
• Assessing the needs of users,
• Establishing and running observational network in accordance the needs and applicable
standards,
• Data collection, processing, analysis and storage,
• Management of historical data,
• Data dissemination,
• Data Exchange and reporting,
• Institutional and human resource development.
A HIS should be designed so that it can carry out all these tasks in as efficient manner.

9.2.1 Assessing the needs of users

In any information system, the first question to be answered is what type of data are to be
provided to the users. Answer to this question determines the layout of the observation network
(parameters, network density, observation frequency, equipment, etc.) and the data to be stored
in the database. Of course, this question is largely answered for a HIS by the definition itself. To
assess the needs of data users, it is necessary to regularly interact with them. A successfully
tested way to formalize this is through a Hydrological Data User Group (HDUG). Potential data
users and those who are responsible to implement HIS are members of these HDUGs. These
HDUGs should meet regularly to review hydrological data needs, identify shortfalls, and make
suggestions for improvements.

In India, many Central and State Government organizations are the major users of the
hydrologic data. There are scores of other governmental, nongovernmental and private agencies
also who use this information. Usually there is a direct link between the objectives of water
resource management and the type of data needed from the Hydrological Information System.
Due to various reasons, hydrological data needs of the users change with time. Therefore, it is
very important to identify the potential data users and regularly update their data needs.
Normally, it is expected that the hydrological information service agencies will satisfy most of
the genuine data needs of the potential users. For ensuring an optimal use of the public resources
spent for maintaining a Hydrological Information Service, it is essential to have a proper balance
between the data needs of different users and mandate of various services supporting HIS.

Often, it is assumed that the information being provided is the same as required by the
potential users. User’s needs may change with time and the HIS would fulfill its obligations only
if there is a continual review of the changing needs of the users.

After finalizing the objectives of the HIS, the observational network has to be planned
and established. This aspect has been discussed in a previous module. It is important to ensure
that the observation networks of different agencies are properly integrated so that duplication of
efforts is avoided.

9.2.2 Data Transmission

WMO (1994) has given the classification of data transmission systems:

1. Manual with the observer at a station sending data to the central office.
2. Manual/semi-automatic system where the central office manually interrogates the
automatic field stations through telephone, radio, etc. and receives the data.
3. Pre-program and time system where automatic equipment initiates the transmission of
observations.
4. Automatic event indicator and the station automatically transmitting the specified change
of variable to a central location.
5. Automatic system with station transmitting and central office recording data
continuously.

The possible choices of transmission links are:

a) Telephone lines are used wherever feasible. With improvements in information technology,
high rates of data transmission are possible.
b) High frequency radio links are used when land lines are not available or topography is
difficult. The installation cost can be high.
c) A significant development of last few decades has been the use of mobile telephony or
satellites for data transmission. A satellite-based system consists of Data Collection
Platforms (DCP) that are installed at hydrometeorological stations. DCPs are (rechargeable)
battery-operated devices that collect, encode, and communicate the data of the station to a
central location through a satellite link at pre-determined time/frequency. Solar panels are
frequently employed to generate energy. This system is very useful for remote and difficult-
to-access locations.

A particular transmission system is adopted depending on a) the frequency of data

observation and the urgency of data, b) the additional benefits of having forecasts based on
telemetered data, c) robustness and reliability of the system, particularly in inclement weather,
and d) availability of finances, infrastructure and manpower to efficiently run the system.

With growth in information technology, the trend is towards automatic observation,

transmission and storage of data. The practices are undergoing rapid changes with technology.
Multi-parameter data loggers can measure, store, and transmit data observed by several
observation sub-systems. These days, the data loggers are small, rugged, and have small power
requirement. These may be battery or solar-power operated. WMO has evolved codes related to
hydrometeorological data. They have also launched an elaborate system for data observation and
transmission through the World Weather Watch (WWW) programme (www.wmo.ch/web/www).

9.3 Hierarchy of HIS

In India, the HIS may also be viewed as hierarchical system, operating at different levels of
sophistication from measurement in the field to national and/or agency level processing, storage
and data dissemination centres. A typical flow of data is as follows (HP 2003):
 At the observation stations relevant data and water quality samples are collected.
Samples are sent to Water Quality Laboratories. At regular intervals (say, monthly) the
collected data are sent to the next higher office which may be the sub-divisional/District
Data Processing Centre.
 In the Sub-divisional/District Data Processing Centres all data received from
observing stations are entered in the computer and stored in a temporary database.
Primary validation is carried out and then the data are passed on to the
Divisional/Regional Data Processing Centre. If necessary, feedback is given to those
observing the data.
 Water samples are analysed in the Water Quality Laboratories. The analysis results are
entered in the computer and subjected to primary validation. At regular intervals, the
laboratory passes the information on to the nearest Divisional or Regional Data
Processing Centre.
 In the Divisional/Regional Data Processing Centres, given their larger spatial coverage,
more advanced secondary data validation is carried out. After validation, the surface
water and groundwater data are transferred to their respective State Data Processing
Centres.
 State Data Processing Centre carries out final data validation, completion, analysis and
reporting. Here, the data are stored in temporary databases. At the end of the hydrological
year, once the data have been properly validated, the (authenticated) processed data is
transferred to the State Data Storage Centre. To improve the effectiveness of the final
validation, State Centres also uses the relevant data collected by the other organizations.
 The State Data Storage Centre stores and administers the storage of all field and
(authenticated) processed hydrological data collected in the State, and provides the data
available to authorised Hydrological Data Users. As a State archive, it also maintains an
HIS-Catalogue of all data stored in its own database and those stored in the databases of
the other states and of the Central Agencies.

In some organisations, data collected by the whole network may be kept at a central
place. It may be noted that the above is a general descriptions and there may be case specific
differences.
 It is clear from the above that data processing activities are carried out at more than one
level within each agency and this makes it essential to have adequate /communication links
between them with increasing penetrations internet, data transmission has become quite simple
and efficient. The requirement for communication is to be based on a low frequency and high
volume of communication. Information is exchanged between various agencies for data
validation as surface and groundwater networks are operated by different state and central
agencies.
9.4 Data Storage and Retrieval
The vast amounts of data which are observed by incurring huge efforts and resources should be
stored in such that they are easily obtainable and secure against losses. Also, it is much more
useful if the basic data is processed and provided to the end users in standard formats. Common
formats include ASCII files [(data are given as comma separated variables (CSV)] or as
spreadsheet files. This can save a lot of money and effort of the user agencies and they would be
encouraged to use the data to solve different water resources problems. Archival of historic data
is important in any field. When done in a logical manner, it enables the end users to make the
efficient use of data.

Vast quantities of water resources data are collected all over the world every day. In this
computer era, the archival of data may be accomplished in a very efficient and economic way.
The basic and other processed information may be stored on computer media and the same may
then be quickly made available to the users. Hard copies in the form of data year-books may also
be brought out for use by practising engineers, planners and managers. These water year-books
are frequently made available to the users on computer. Besides the processed data, it is
advisable to store the raw data as well which might be needed for research purposes and it might
be realized later that the data validation procedures had missed some aspects.

A typical setup for water resources data management is depicted in Fig. 9.2. The main
components are: a) data entry module to input data from various sources in the database, b) user
interface for data editing, display, and management, and c) applications that can retrieve data or
write to the data base.

Due to large volumes of water resources data, it is necessary that the data are stored such
that minimum space is needed. The current trend is to archive the data on optical/magnetic disks
which can hold large amount of data and do not require stringent environmental conditions for
storage. Now-a-days high capacity hard disks and pen-drives are available and the cost of
hardware has dropped drastically. Of course, the volume of data that are being generated each
day is also progressively increasing.

Data entry

Utility Application
programs 100110100111010110
programs
011011010001001001

 Data editing  Surface water

 Graphical  Ground water
displays Hydrologic  River basin
 Database Database management
management  Environment
tasks models
… 100110100111010110 …
Fig. 9.2 Conceptual representation of a hydrological database, its management and use.

Data compression is applied to reduce the requirement of storage space. WMO (1994)
has described such procedures. A number of data compression algorithms have been developed.
For example, daily rainfall is measured to an accuracy of 0.1mm. Rather than storing it as a real
number which requires 4 bytes, the values can be multiplied by 10 and stored as an integer
requiring only 2 bytes. The daily data of one month when rainfall was zero need not be stored as
30 zeros. An efficient way is to use notation ‘30*0’. The database files are not normal ASCII
files that require larger space; these are special types of files. The suitability of a particular
compression technique will depend on the characteristics of the data.

Efficient data search and retrieval module is also necessary so that the requisite data are
quickly fetched from the database. A good retrieval system should provide the user a
combination of options to select the data using criteria, such as by variable, basin, station, time
period or range of values. The user may also be given the option to choose the format of the
output, i.e. tabular, plot or ASCII data files that can be directly input to another software. If
graphs are displayed on the monitor, the user may be given option to print them or store on hard
disk for later use. Adequate security measures should also be built in the retrieval system so that
only authorized users have access to the database. Among the authorized users also, there could
be various categories. Most users are given read-only access and they cannot modify to the
database. A limited group of users are given all privileges, i.e., they can read, modify and delete
data from the database. It is useful to have a log of all users who have accessed the database and
the operations that they have performed so that the source can be identified in case of any
mishap.

9.4.1 Management of historical data

State and central agencies have maintained observational networks for many years and
voluminous records are held, the majority on manuscript or chart records, which are not readily
accessible for use and are of variable quality. Programs of historical data entry in the database
need to be established in each agency holding historical data. Of course, the first priority is to
ensure that current data are entered, validated and stored. The next priority could be for historic
data of the recent past years, and so on.

9.4.2 Data storage and dissemination

The basic objective of creating databases and storing data in an efficient manner is to encourage
the use of data for planning, design, management, and research purposes. There should be no
hurdle for genuine users in accessing the data. The first and foremost step in the dissemination
process is an up-to-date catalog of database. The catalogue of data held in various databases
should be updated periodically and automatically, if possible. Of particular importance is to
assure the quality and reliability of the data provided to users. Reports are prepared to bring out
the salient characteristics of the hydrological regime of the region for each water year. Special
reports are also prepared as and when required. These days the trend is to disseminate the reports
through internet.

Many organizations routinely publish basin-wise data-year books. A typical water year
book consists of description of the basin, its topography, soils, land use etc., major rivers, and
salient features of various water resources projects. Maps are included to illustrate all these
features. The data section contains typically precipitation, streamflow, evaporation, and ground
water data. Periodically, special reports may be published giving long-term statistics of stations
or highlighting special or unusual events, such as floods or droughts.

Of late, many organizations have begun to curtail hard copy publishing due to high costs
and handling problems. With almost all data now available in digital form in a HIS, the hard-
copy publication is not an efficient means of data dissemination. A water-year book can be
conveniently published on line as a PDF file. A browser may also be supplied to handle data
search, display and print. This trend is likely to accelerate further. An important thing to
remember is that the format and content of publications should depend on the need of users. The
contents should be so designed that the need of most data users are answered and the efforts to
handle data requests are reduced. The contents may also depend on the frequency of publication.

Finally, in this age of computers and Internet, it is appropriate that many international
organizations have established databases that can be accessed through Internet. The Global
Runoff Data Centre (GRDC) at the Federal Institute of Hydrology, Germany, has a large archive
of surface water data. The United States Geological Survey (USGS) maintains one of the largest
online hydrologic data base (www.usgs.gov).

Although an integrated HIS that contains data of all the hydrological variables is
desirable, currently in India separate databases are being maintained for surface water data,
ground water data, and hydro-meteorological data. Further, currently organizations belonging to
the Central government and State governments have separate data bases. An integrated data base
is desirable and will have many advantages:
1. An integrated data base will foster co-operation among the various Central and State
Organisations which will lead to improvement in the quality of the data,
2. It will be of great benefits to the users by providing all requisite information from a single
source,
3. It will improve the sustainability of the system and enhance the chances of a uniform
approach in the long term, and
4. It will lead to savings on the expenditure on the hardware, software and manpower.
9.5 Institutional and Human Resources development
Since HIS is a vast system, institutional and human resource development aspects are very
important. The institutions implementing of HIS must be developed in such a manner that the
system is sustainable in a long run. Trained staff is required to carry out different activities under
HIS. In India, observations are taken manually or automatically depending upon the type of
instrument at the station. Persons of various categories (e.g., Junior Engineers, Supervisors,
Technicians, Observers, Helpers, etc.) are engaged for carrying out data collection and for
regular maintenance. Very importantly, staff must have skills required to carry out the desired
tasks. Only those staff should be posted to these centers who have interest and temperament to
work in a data center. Uninterested or incapable people should never be posted in such places.
Regular trainings have to be organized to update the existing staff as well as to train the new
recruits.

9.6 Use of Computers and Internet in HIS

The rapid advance in computer technology, in hardware speed of operation and data storage
capacity as well as the capability of hydrological software has greatly simplified the management
of large quantities of hydrological data and has rendered obsolete those time consuming manual
methods which were formerly the norm. These days one can’t think of processing large volumes
of data without the use of computers. Computer based hydrological information systems offer the
following advantages:
 Very large volumes of data can be easily handled,
 Computers greatly speed up the progress from data collection to completion and storage,
 Computers permit and promote the standardization of processing, validation and
reporting procedures,
 Users can be provided with the data in the required tabular or graphical format,
 Use of computers is an incentive to the staff since it removes the tedium of manual
handling of the data.

Many countries are using the power and opportunities provided by computerized data
processing system and Internet. However, the potential has not been fully exploited in India in
the field of hydrology so far.

9.7 Closure
Two basic tools for integrated management of the environment are modeling and
environmental data. Both tools were available and valid in the past; however, the recent
requirements for integrated environmental management have also led to a significant evolution
of both modeling procedures and data management systems. Current literature provides vast
amounts of studies on modeling of different environmental processes. However, issues related
to data management systems are barely touched in a comprehensive framework. Data
requirements and data availability are mentioned merely as subtopics in most environmental
studies although it is well recognized that data constitute the basis for all environmental
management activities. Most developed countries have well-established databases which can
be accessed easily by the users. Developing countries, on the other hand, do not have extensive
data banks and these many not be easily accessible by the users. There is a need for
harmonization or standardization in development of databases so that data exchange can be
facilitated on regional and global levels.

9.8 References
Hydrology Project, “Basic Surface Water Data Processing, HIS (SW) Operation Manual Vol. 8 –
Data Processing”, Ministry of Water Resources, India, 2003.
Latkovich, V.J. and Leavesley, G.H. (1993). Automated Data Acquisition and Transmission, in
Handbook of Hydrology, edited by D.R. Maidment. McGraw-Hill Book Company, New
York.
WMO (2008). Guide to Hydrological Practices. WMO No. 168. World Meteorological
Organization, Geneva.
 Module 10
Statistical Analysis of Hydrological Data

10.1 Basic Concepts

10.1.1 Probability
10.1.2 Axioms of Probabilities
10.1.3 Properties of Random Variable
10.2 Key Statistical Measures of Data
10.2.1 Measures of Central Tendency
10.2.2 Measure of Dispersion or Variation
10.2.3 Measures of Symmetry
10.2.4 Measures of Peakedness or Flatness
10.3 Graphical Presentation of Data
10.4 Probability Distributions
10.4.1 Normal Distribution
10.4.2 Log-Normal Distribution
10.4.3 Extreme Value Type 1 (EV1) Distribution
10.4.4 Log Pearson Type - III (LP3) Distribution
10.4.5 Discrete Probability Distributions
10.4.6 Binomial Distribution
10.4.7 Poisson Distribution
10.5 Parameter Estimation Methods
10.5.1 Method of Moments
10.5.2 Method of Moments for Discrete Systems
10.5.3 Method of Maximum Likelihood (MLE)
10.5.4 Method of Least Squares
10.5.5 Method of L-Moments
10.6 Problems of Parameter Estimation
10.7 Hypothesis Testing
10.7.1 The t-distribution
10.7.2 Chi-Square Distribution
10.7.3 Tests Concerning Variances of Two Populations
REFERENCES
Questions
Keywords: Statistical Analysis, Hydrological, Axioms, Random Variable, Probability
Distributions, Normal, Log-Normal, Extreme Value, Log Pearson, Binomial, Poisson, Parameter
Estimation, Moments, Maximum Likelihood, L-Moments, Parameter Estimation, Hypothesis
Testing, t-distribution, Chi-Square

10.1 Basic Concepts

A hydrologic process is a phenomenon describing the occurrence and movement of water in the
earth phase of the hydrologic cycle. When the phenomena are observed at intervals of time, it is
termed as a discrete process and when these are observed continuously, continuous series are
generated. The observed behavior of many hydrologic variables cannot be completely explained
by applying the physical laws. Three reasons behind this are: a) incorrect or incomplete
knowledge of the underlying processes, b) inadequate data, and c) inherent randomness of the
process. In these situations we resort to statistical analysis to infer and predict the behavior of the
hydrologic variables.
Data of the concerned process are collected to apply the probability theory. All possible
observations of a process constitute its population. But we cannot completely collect the data of
population in a limited time frame. Generally, only a finite portion of the population is observed
and is called a sample. Statistical properties of the sample are determined and it is assumed that
these represent the properties of the population.
Many hydrologic data (e.g., daily rainfall amount, discharge of a river) can be
characterized by variable(s) that are unpredictable to some degree. Yet frequently there is a
degree of consistency in the factors governing the outcome which exhibits a statistical regularity.
A variable whose value at any time is not influenced by the value at earlier time(s) is known as a
random variable. A random variable can be discrete which means that it can take on only a finite
set of values, for example the number of rainy days in a year at Roorkee. It can also be
continuous (a time series) and can take on any value, for example, the water level of Ganga
River at Rishikesh or the magnitude of rainfall at a place.
The behavior of the hydrological processes may be deterministic or stochastic. For a
deterministic process, the behavior can be completely described with certainty. The governing
equation defines the process for the entire time (or space) domain. In contrast to this, a stochastic
process evolves, entirely or in major part, according to a random mechanism and hence the
future outcomes of the process are not exactly predictable. Such hydrologic variables whose
values are governed by the laws of chance are called stochastic variables.
For many hydrological problems in these cases, sample data consist of measurements
made on a single random variable and the techniques of analysis are called univariate analysis
and estimation. Univariate analysis is carried out by using the measurements of the random
variable (or the sample information) to identify the statistical properties of the population from
which the sample measurements are likely to have come. After the underlying population has
been identified, one can make probabilistic statements about the future occurrences of the
random variable; this represents univariate estimation.

Before commencing the analysis, it is always useful to plot the sample data. The
following are the main steps of statistical analysis of data:

i) Compute the basic statistical indicators of the data.

ii) Select a set of probability distribution functions.
iii) Fit the selected distributions with the sample data. Common methods of parameter
estimation are discussed subsequently.
iv) Select the best fit distribution using the goodness-of-fit tests.
v) Use the best-fit probability distribution to make inferences about the likelihood of
occurrence of the magnitudes of the random variable.

If all the values of a random variable and the corresponding probabilities are known or
found, the relation between these values and probabilities is described by a probability
distribution. Knowing this distribution, the probability of any value of the random variable can
be determined.

We now briefly discuss the basic concepts of statistical analysis.

10.1.1 Probability
Based on the daily experiences, most people have an intuitive appreciation of the concept of
probability or chance. When a coin is thrown, there are two equally likely possible outcomes:
head or tail. Let the coin be tossed n times and head occurs s times. The ratio s/n is the
probability of occurrence of heads.

Sample space is a collection of all possible random events that might arise from a random
experiment. Sample space S and two sets of events A and B are shown in Fig. 10.1. If two events
A and B do not contain any common sample points, they are said to be mutually exclusive.

If two events A and B are not mutually exclusive, the common set is called their
intersection, denoted as A∩B. The union of two events A and B gives the event which is the
collection of all sample points occurring at least once in either A or B and is denoted as AB.
Fig. 10.2 shows this concept through the Venn diagram. The complement Ac of an event A
consists of all sample points in the sample space of the experiment not included in the event A.
 A B
S

Fig. 10.1 Two mutually exclusive events A and B in sample space S.

A A

B B

A∩B
AB
Fig. 10.2 Venn diagram showing intersection and union of two events.

10.1.2 Axioms of Probabilities

Events are basically sets and the common set operators, including union, intersection, and
complement, are applicable to them. Four frequently used rules for set operations, considering
three events A, B and C are:

(1) Commutative rule: AB= BA, AB= BA

(2) Associative rule: (AB) C=A (B C), (AB) C=A (B C)
(3) Distributive rule: A(BC)=(AB)(AC), A(BC)=(AB) (AC)
(4) De Morgan’s rule: (AB)C = ACBC, (AB)C = ACBC

The notation P[A] is used to denote the probability of a random event A. Now we discuss the
axioms of probability.

Axiom 1: The probability of an event A is a number greater than or equal to zero but less than or
equal to unity:
0  P[A]  1 (10.1)

Axiom 2: The probability of an event A, whose occurrence is certain, is unity:

 P[A] = 1 (10.2)

where A is the event associated with all sample points in the sample space.

Axiom 3: The probability of an event which is the union of two events is:

P[A or B] = P[A  B] = P[A] + P[B] - P[A  B] (10.3)

where AB denotes the union of events A and B which means that either event A occurs or event
B occurs, and A  B denotes the intersection of event A and event B. Eq. (4) can be extended to
the union of n events. If A and B are two mutually exclusive (disjointed) events, the probability
of A and B, P[A B], will be zero and eq. (5) becomes:

P[A or B] = P[A] + P[B] (10.4)

Axiom 4: The probability of two (statistically) independent events occurring simultaneously or

in succession is the product of individual probabilities:

P[E1E2] = P[E1] × P[E2] (10.5)

Statistical independence implies that the occurrence of event E1 has no influence on the
occurrence of event E2.

10.1.3 Properties of Random Variable

Let X denote a random variable and x be a possible value of X. The cumulative distribution
function (CDF), FX(x) is the probability that the random variable X is less than or equal to x:

FX(x) = P(X  x) (10.6)

The probability distribution function (PDF) describes the relative likelihood that a
continuous random variable X takes on different values, and is the derivative of the CDF:

fX(x) = d {FX(x)} /dx (10.7)

 The PDF and CDF of a random variable are shown in Fig. 10.3. Note that the CDF is
denoted by capital letter F and the PDF by lower case letter f.

Let us now state some important properties of f(x) and F(x) for continuous and discrete
random variables.

1. The probability of a random variable cannot be negative

f ( x)  0,   x  
(10.8)

2. The sum of probabilities of all possible outcomes is equal to 1, i.e., the area under the
PDF is unity.


 f ( x ) dx  1

(10.9)

(a)
F(x)

F(x)
CDF

|
x X

f(x)
x
(b)
n
---
Nx

PDF

|
0 x X
Fig. 10.3 PDF and CDF of a random variable.

For discrete random variables, the following statements can be made:

 f (x )  1
i
i (10.10)

where f(xi) represents the probability of X = xi in the sample space if the sample contains finite
and f(xi) observations can be replaced by p(xi). The probability of “X assumes a value  x” is
equal to the area under PDF curve to the left of x:
x

3.
P ( X  x)  F ( X  x)  F ( x)   f ( x)dx

(10.11)

Moreover
P ( a  X  b)  P ( X  b)  P ( X  a )
b a
  f ( x)dx   f ( x)dx
 

b
  f ( x)dx, for a  b (10.12)
a

For discrete case

xi b
P ( a  x  b)   p( x )
xi  a
i (10.13)

k
P ( X  x k )   p ( xi ) (10.14)
i 1
10.2 Key Statistical Measures of Data
Four principal features which characterize a set of observations on a random variable are:
(i) the central tendency or the value around which all other values are bunched,
(ii) the spread of the sample data around mean,
(iii) the asymmetry or skewness of the spread of data, and
(iv) the peakedness of the data.

These characteristics are expressed in terms of statistical properties which are estimated
from the sample data.

10.2.1 Measures of Central Tendency

In statistics various measures of central tendency are employed. Three important measures are
the following.

(i) Arithmetic Mean: If x1, x2 … xn represent a series of observations, the mean of this series is:
1 n
x = xi
n i=1 (10.15)

Where x represents the sample mean; the mean of population is generally denoted by .

(ii) Mode: It is the value which occurs most frequently. It is the peak value of the PDF. A data
set may have more than one peak.

(iii) Median: It is the middle value of the ranked observations for a data set. The median divides
the distribution in two equal parts.

10.2.2 Measure of Dispersion or Variation

Three statistical measures of variation of data are commonly used.

(i) Variance: It represents the scatter of the data are about the mean. Variance is computed by:

1 n
s =  (xi - x )
2 2
(10.16)
n i =1

A small value of variance implies that values are bunching close to the mean.
 (ii) Standard Deviation (SD): The unbiased estimate of population standard deviation (s) is given
computed as the square root of the variance:

1 n
s= [ 
n i=1
( xi - x )2 ] 0.5 (10.17)

when n < 30, the unbiased estimate of s is found by replacing n by n-1 in the denominator. Greek
letter σ is used to denote the standard deviation of population.

(iii) Coefficient of Variation (CV) is a dimensionless parameter and is obtained by dividing the
standard deviation by the mean:

CV = s / x (10.18)
When the mean of the data is zero, Cv is not defined. This coefficient is useful to compare
different populations. Given two samples of data, the one with larger Cv will have more spread of
the values around the mean.

Example 10.1: Average annual flows (in cumec) at a river gauging site are given in the table
below. Compute the mean, variance, standard deviation, and the coefficient of variation of the
flows.
Year  1970  1971  1972  1973  1974  1975  1976  1977  1978  1979  1980 
Flow (cumec)  195.5  145.4  148.1  324.7  205.6  302.9  210.3  194.4  71.2  126.8  216.0 
Year  1981  1982  1983  1984  1985  1986  1987  1988  1989  1990  1991 
Flow (cumec)  136.4  403.9  145.3  161.6  112.5  110.0  90.2  129.8  80.5  136.3  243.3 

Solution: We have a total of 22 values. The mean of the flows can be computed as

Mean = (195.5 + 145.4 + 148.1 + …. 80.5 + 136.3 + 243.3)/22 = 176.8 cumec.

The variance can be computed by eq. (10.16)

Variance s2 = [(195.5 -176.8)2 + (145.4 - -176.8)2 + (148.1-176.8)2 + …+ (136.3-176.8)2
+ (243.3-176.8)2)]/22 = 6926.46 cumec2
SD s = (6926.46)0.5 = 83.22 cumec
CV = 83.22/176.8 = 0.47.

10.2.3 Measures of Symmetry

Usually the hydrologic data are not distributed symmetrically around the mean. If the data to the
right of the mean are more spread out than those on the left then, by convention, the asymmetry
is positive and vice versa for negative asymmetry (see figure 10.4).If the data are symmetrically
placed around the mean then the measure of symmetry would be zero.

The third moment of the data about the mean is used in indicating symmetry and is given
by:
n
1
M3 =
n
 (x
i =1
i - x )3
(10.19)

It is easy to see that this moment is zero if the data are symmetrical. Otherwise, M3 will have
certain value, a positive or negative.

Note that because the third central moment has dimensions equal to the cube of the data,
it is not useful while comparing different data sets. Being non-dimensional the coefficient of
skewness does not have this disadvantage and is preferred.

Coefficient of Skewness: A non-dimensional measure of the asymmetry of the distribution of the

data is helpful when various data are to be compared and the coefficient skewness is one such
measure. The coefficient of skewness (Cs) is given by:
n
n  (xi - x )3
Cs = i 1 (10.20)
(n - 1)( n  2 ) s 3

Symmetrical frequency distributions have very small or negligible value for skewness
coefficient Cs, while asymmetrical frequency distributions have either positive or negative
coefficients. When Cs has a small value, it indicates that the probability distribution may be
approximated by the normal distribution since Cs = 0 for this distribution. The symmetrical and
skewed distributions are shown in Fig. 10.4.

Negative skew Zero skew Positive skew

Mean Median Mode Mean = Median = Mode Mode Median Mean

f(x1)

f(x2)

f(x3)

x --
Fig. 10.4 Symmetrical and asymmetrical (+ve and –ve) skewed distributions.

10.2.4 Measures of Peakedness or Flatness

The measure used to denote the peakedness or the flatness of the frequency distribution near its
centre is known as the kurtosis coefficient. This coefficient is computed by:
n
n 2 ( x i - x )
4
Ck =
i 1 (10.21)
(n - 1)(n - 2)(n - 3) s 4

Normal distribution has the kurtosis 3. If a data set has a relatively greater concentration near the
mean than the normal distribution, the kurtosis will be greater than 3. Conversely, if the data
have a relatively smaller concentration near the mean than the normal distribution, the kurtosis
will be less than 3.

Example 10.2: Compute the coefficient of skewness and the coefficient kurtosis of the data of
example 10.1.

Solution: The coefficient of skewness can be computed by eq. (10.20)

Cs = [22/(21*20*83.223)]*[(195.5 -176.8)3 + (145.4 - -176.8)3 + …

+ (136.3-176.8)3 + (243.3-176.8)3)]
= 1.238
A positive value of CS implies that the probability distribution of the data has heavy tail
to the right.
Kurtosis can be computed by eq. (26)

Ck = [22*22/(21*20*19*83.224)]*[(195.5 -176.8)4 + (145.4 - -176.8)4 + …

+ (136.3-176.8)4 + (243.3-176.8)4)]
= 1.45
Since kurtosis is less than 3, it means that the data values are less concentrated around the
mean than the normal distribution or the peak of the distribution will be flatter compared to the
normal distribution.

10.3 Graphical Presentation of Data

Graphically presentation helps in a good insight in the behavior and variation of the data. To
graphically present the data in the form of histograms, a frequency table is prepared. For this
purpose the range of the data is divided into a number of intervals of convenient size and
frequencies of values occurring in each interval is entered alongside. The appearance of a
frequency histogram depends upon the selection of class interval. If the class intervals are very
large, the table is compact but details may be lost. If the intervals are too small, the table may be
too bulky. The following guidelines may be considered while choosing the class interval

(a) Brooks and Carruthers’ rough guide:

Number of classes  5 log (sample size) (10.22)

(b) Charlier's rule of thumb:

w = (maximum value – minimum value)/20 (10.23)

where w is the size of class interval. In general the number of classes varies between and 25.

To prepare the frequency table, steps given below can be followed:

(i) Arrange the variable (Xi) in increasing or decreasing order of magnitude.
(ii) Decide the number of class intervals (NC) and the size of the class interval X.
(iii) Divide the ordered observations Xi into NC intervals.
(iv) Determine the absolute frequency nj as the number of observations that fall in the jth class
interval, j=1, ... NC.
(v) Compute the relative frequencies of various classes as nj/n, j=1,... NC and n is the number
of observations.
(vi) Compute the cumulative relative frequencies Fj, j = 1, ... NC.

(vii) Plot the relative frequencies as well as cumulative relative frequencies with group
interval as abscissa and the relative frequencies or cumulative relative frequencies as
ordinate.

Example 10.3: The annual flow of Sabarmati River at Dharoi is plotted in Fig. 10.5 for the
period 18
868-1965. Pllot the histog
gram and thee cumulativee histogram.

Fig. 10.5 Plot of the annual flow

w of the riverr.

Solutionn: After exam

mining the daata having 98 values, thee class intervval was chossen as 100 M
MCM.
Table 10.1 shows thee mid-valuess of classes in
i which thee data has beeen divided, the frequenncy of
values in
n each class and
a the cumu ulative frequ
uencies. Theere are 17 claasses.

Table 10.1 Mid-valuees and frequ

uencies of vaarious classess of examplee data.
Mid-valu
ue of class Cuumulative
(M
MCM) Frequen
ncy frrequency
150
1 6 6
250
2 9 15
350
3 11 26
450
4 9 35
550
5 9 44
650
6 9 53
750
7 19 72
850
8 6 78
950
9 6 84
10
050 1 85
1150 5 90
12
250 2 92
 13
350 3 95
14
450 0 95
15
550 2 97
16
650 0 97
17
750 1 98

The cum
mulative histoogram of thee annual flow
w of Sabarm
mati River att Dharoi for the period 11868-
1965 (98 years) is plo
otted in Fig. 10.6.

Fig. 10.6
6 Histogram
m of the annual river flow
ws.

7 Cumulative histogram of annual flows of the rriver.

Fig. 10.7
10.4 PROBABILITY DISTRIBUTIONS
In statistics, a probability distribution gives either the probability of each value of a random
variable (when the variable is discrete), or the probability of the value falling within a particular
interval (when the variable is continuous). The probability distribution describes the range of
possible values that a random variable can attain and the probability that the value of the random
variable is within any (measurable) subset of that range.

A probability distribution gives important information about the data, how the values are
changing, whether they are bunched together or spread out, and whether they are symmetrically
disposed on the X-axis or not. Distribution also tells the relative frequency or proportion of
various X values in the population in the same way that a histogram gives information about a
sample. We now describe the distributions that are commonly used in addressing water resources
problems.

Commonly used distributions in hydrology are the Normal, Log Normal, Extreme Value
type-1 (Gumbel or EV1), Gamma, Pearson Type - III, and Log Pearson Type - III distributions.
A brief description of these distributions is given below.

10.4.1 Normal Distribution

It is also known as the Gaussian distribution. When a hydrologic variable, integrated over a large
time period is used in analysis, the variable is expected to follow a normal distribution. The
normal distribution has a symmetrical bell-shaped probability density function. The two
parameters of the normal distribution are mean  and standard deviation. Its PDF can be
expressed as
1  (x  )2  (10.24)
f(x) = exp -  -  x  
 2  2
2


the cumulative density function (CDF) of the normal distribution is:

1
x
 (x  )2 
F(x) =   2 2  du
 2 - 
exp (10.25)

The origin of the normal organization lies in the central limit theorem which states that if
a sequence of random variables xi, i = 1, 2, …, n are independently and identically distributed
n
with mean μ and standard deviation σ then the distribution of n such random variables Y= x
i 1
i
tends to the normal distribution with mean nμ and standard deviation n , as n becomes large.
This theorem holds good irrespective of the probability distribution of x.

The reduced variate of the normal distribution is defined as Z = (x - )/. The properties
of the reduced variate are mean = 0, standard deviation z = 1, and coefficient of skewness = 0.
Fig. 10.8 shows the normal distribution and the area under the curve for three values of the
reduced variate. As shown, the area under the curve within  ±  is 68.27%, within  ± 2 is
95.45 and it is 99.73 within  ± 3.

±
Area = 68.27%
f(x)

 ± 2, area =
95 45%
 ± 3, area = 99.73%
| | | | | | |
 x

Fig. 10.8 The normal distribution and the area for three values of the standard variate.

Among the probability distributions used in hydrology, the normal distribution is the
most widely. It is also employed in the analysis of variance, estimation of random errors of
hydrologic measurements, hypothesis testing, synthetic generation of random numbers, etc. A
random variable that is made up of the sum of many small independent effects is expected to
follow a normal distribution. Many hydrologic variables are not normally distributed, but
transformations can, in many cases, make them approximately normally distributed. When there
is increase in the time interval over which a hydrologic variable is measured, the variable
approximately follows a normal distribution because the number of causative effects increases.

Example 10.4: Assuming that the data of Sabarmati River follows the normal distribution, find
the parameters of the distribution and plot it.
 n: For the datta of Sabarm
Solution mati river, thee mean and S
SD are:

Mean
M of the data
d x = 6655.37 million cubic m.
Standard deviiation  = 34
46.9 million cubic m.
Coefficient
C off variation CV = 346.9/665.37 = 0.5221.
Coefficient
C off skewness Cs = 0.76 (poositively skeewed).
Kurtosis
K Ck = 3.65.

Fig. 10.9
9 shows the plot
p of the prrobability distribution off Sabarmati R
River data.

9 Probability
Fig. 10.9 y distribution
n of Sabarmaati River datta.

For the Sabarmati data, the coefficieent of skewn wness Cs is 00.76 or the ddata is posittively
skewed. This is easilly verified by
y Fig. 10.9. Further, kurrtosis Ck forr the data is 33.65 (kurtossis for
the norm
mal distributio
on is 3). Agaain, this can also be veriffied from Fiig. 10.9.

10.4.2 Lo og-Normal Distribution

D n
The log-n normal distrribution is th
he probabilitty distributioon of a randoom variable whose logarrithm
is normally distributeed. Let X bee a random variable
v withh a normal ddistribution, then Y = exxp(X)
has a log g-normal disttribution. In other wordss, if Y is logg-normally diistributed, thhen X = log((Y) is
normally y distributed
d. When a random
r variable repressents a proccess that is the resultaant of
multipliccative producct of many small
s effectss each of whhich is posittive, then it can be expressed
the sum of logarithm ms of these small effectts. The logarrithm of succh a random m variable caan be
expected d to follow a normal distrribution. Hence, if the vvariable is traansformed too the log dommain,
it is likely to follow the
t normal distribution.
d An advantagge of the logg-normal disttribution is tthat it
is often useful to represent quantities that cannot have negative values. It has proven useful to
model rainfall amounts, size distributions of aerosol particles, etc.

The PDF of the log-normal distribution is

1  (ln x   y ) 2 
f ( x)  exp   x0
x y 2  2 y2 
(10.26)

The log-normal distribution has two parameters y and y which can be estimated by
transforming all xi’s to yi’s by

yi = ln xi (10.27)

10.4.3 Extreme Value Type 1 (EV1) Distribution

Let a series of large number of (N) observations of random variable be subdivided into n
subsamples of size m each, such that N = nm. Each subseries shall have two extreme values: one
maximum and one minimum corresponding to, for example, floods and droughts. Gumbel (1958)
showed that the n largest values of subsamples asymptotically follow an extreme value type 1
(EV1) distribution. This distribution, also known as the Gumbel distribution or double negative
exponential distribution, is widely used for frequency analysis of floods, maximum rainfall, etc.
This distribution is essentially a log-normal distribution with constant skewness (approximately
1.14). Its PDF and CDF are as follows:

f(x) =  exp {-(x - ) – exp[-(x - )]} -  < x < ; - <  < ;  > 0

F(x) = exp{ -exp[ -(x - )]} (10.28)

where , and  are scale and location parameters. The estimates of parameters using the method
of moments are:

1.283
ˆ  ; ˆ  x - 0.45s (10.29)
s

It has been shown that the EVI distribution is a special case of a distribution known as the
Generalized Extreme Value (GEV) distribution. The CDF of the GEV distribution is given by
   x u  
1/ k

F ( x)  exp  1  k   (10.30)
    

Where k, u, and α are the parameters of the distribution. When k = 0, we get the EV1
distribution. For k<0, the distribution known as EV2 and it is known as EV3 distribution when k
> 0.

According to Gumbel, the probability that an event with magnitude larger than x0 occurs
is (Subramanya 2008):

Prob (X ≥ x0) = 1 – exp[-exp(-y)] (10.31)

where y is the reduced variate, given as

y = α(x - β),
β = x - 0.45 σx
α = 1.2825/ σx (10.32)

Substituting the values of a and α

y = 1.285 (x- x )/ σx + 0.577 (10.33)

The expression for the reduced variate y for return period T is

T
y = - [0.834 + 2.303 log (log )] (10.34)
T 1

Now we can compute variate x with return period T by

XT = x + kσx (10.35)

where k = (yT – 0.577)/1.2825 (10.36)

Equations (10.35) and (10.36) assume that a large data series are available to compute the
various statistics. However in practice, the record length is finite. In such cases, the following
equation may be used:

xT  x  k n n 1 (10.37)
where σn-1 is the standard deviation of the sample of size n. Frequency factor for use with sample
of size n is given as

k n   yT  y n  / s n (10.38)

where y n and sn are reduced mean and reduced standard deviations which are functions of n.
Values of these can be obtained from standard tables that are widely available (see, for example,
Subramanya 2008). Note that as n➝∞, y n ➝ 0.577 and sn ➝ 1.2825.

Example 10.5: From the flow data of a river, the mean and standard deviation were computed
and these turned out to be 660 million cubic m and 330 million cubic m, respectively. Find the
value of parameters of EV1 distribution.

Solution: The mean and standard deviation of the data are 665.37 million cubic m and 346.9
million cubic m, respectively. Therefore, the estimates by the method of moment are:

 = 1.2825/330 = 0.0039.
and  = 660 – 0.45*330 = 511.5.

Example 10.6: Annual maximum flood discharge data of a river was available for 30 years.
Mean and standard deviation were 5250 m3/s and 1650 m3/s. Compute the flood discharge with a
return period of 100 years by using the Gumbel Extreme Value 1 distribution.

Solution: From standard tables, for n = 30 years

yn = 0.5362, sn = 1.1124.
100
Hence yT = -[0.834 + 2.303 log (log )] = 4.601
99
k100 = (4.601 – 0.5362)/1.1124 = 3.654
x100 = 5250+3.654*1650 = 11279 m3/s

10.4.4 Log Pearson Type - III (LP3) Distribution

Log Pearson Type III distribution was found to give good results in numerous studies dealing
with flood peak data. This distribution is the standard distribution for flood frequency analysis in
the USA since its use for flood frequency analysis was recommended by the US Water
Resources Council.
LP3 is a three-parameter distribution and is widely used in hydrology. Its parameters are
related to mean, standard deviation, and skewness.
b 1
1  x c  x c
f ( x)    exp  
a(b)  a   a  (10.39)

where a, b, and c are scale, shape, and location parameters, respectively, and (b) is a gamma
function. If c = 0, this distribution becomes a two-parameter gamma distribution. Parameters a,
b, and c are related to mean, standard deviation, and coefficient of skewness as (method of
moment estimates)

a = /b (10.39a)
2
b = (2/Cs) (10.39b)
c =  - b (10.39c)

To determine flood for a return period T by using the LP3 distribution, the procedure
described below is followed.

First of all, the frequency factor, KT is computed by (Chow et al. 1988):

KT = z + (z2 - 1)k + (z3 - 6z)k2/3 + (z2 - 1)k3 + zk4 + k5/3 (10.40)

Where k = Cs/6. To complete z for a given return period T, exceedance probability p is obtained
as p = 1/T. Now, complete a variable w as
w  ln(1 / p 2 ) 0  p  0 .5

Now z is calculated by (Abramowitz and Stegun, 1965)

2.515517  0.802853w  0.010328 w2

z  w (10.41)
1  1.432788 w  0.189269 w2  0.001308 w3

when p > 0.5, p in eq. (10.41) is replaced by (1-p) and the negative sign is put before z computed
by eq. (10.42). Now, by following the frequency factor method, the flood for the return period T
years is computed by:
 yT  y  K T s y (10.42)

Example 10.7: For the data of Example 10.1, find the parameters of the Pearson Type III
distribution.

Solution: The estimates of parameters using the method of moments are

b = (2/Cs)2 = (2/0.76)2 = 6.93 million cubic m.

a = 346.9/6.93 = 131.78 million cubic m.
c = 665.37 – 346.9*6.93 = -247.84.

Example 10.8: Logarithms of the annual flood peak data of a river were taken and the mean was
4.146, SD was 0.403 and Cs = -0.07. Find 50 year return period flood by using the LP3
distribution.

Solution First we find the value of k50 by the following equation:

K50 = z + (z2 - 1)k + (z3 - 6z)k2/3 + (z2 - 1)k3 + zk4 + k5/3

Here k = Cs/6 = -0.07/6 = -0.0117. For T = 50, p = 1/50 = 0.02.

w  ln(1 / 0.02 2 )  2.797

From eq. (10.41)

2.515517  0.802853 * 2.797  0.010328 * 2.797 2
z  2.797   2.054
1  1.432788 * 2.797  0.189269 * 2.797 2  0.001308 * 2.797 3

Now K50 is calculated as

K50 = 2.054 + (2.0542 - 1)*(-0.0117) + (2.0543 – 6*2.054)* (-0.0117)2/3
+ (2.0542 - 1)* (-0.0117)3 + 2.054*(-0.0117)4 + (-0.0117)5/3
= 2.016

Hence, y50 = 4.146 + 2.016*0.403 = 4.959

So, the 50-year flood x50 = (10)4.959 = 90942.
10.4.5 Discrete Probability Distributions
The use of discrete probability distributions is restricted generally to those random events in
which the outcome can be described as success or failure, i.e., there are only two mutually
exclusive events in an experiment. Moreover, the successive trials are independent and the
probability of success remains constant from trial to trial. The binomial or Poisson distributions
can be used to find the probability of occurrence of an event r times in n successive years.

10.4.6 Binomial Distribution

This distribution arises in Bernoulli processes where in any trial; the event may or may not take
place. The probability of occurrence of the event is the same from one trial to another. This
distribution usually occurs while dealing with complementary events. A common example is
tossing of coins in which the probability of head appearing is the same in each trial. The
occurrence of wet and dry days over a given time interval is also a complementary event. The
probability of occurrence of the event r times in n successive years is given by:

r n- r n! r n- r
P r,n = n C r P q = p q (10.43)
r! (n - r)!

where Pr,n is the probability of a random event of a given magnitude and exceedance probability
P occurring r times in n successive years. The probability of the event not occurring at all in n
successive years is:

Po,n = qn = (1 - p)n (10.44)

The probability of an event occurring at least once in n successive years:

P1 = 1 - qn = 1 - (1 - p)n (10.45)

Example 10.9: An analysis of data on the maximum one-day rainfall depth at a station indicated
that a depth of 280 mm had a return period of 50 years. Determine the probability of a one-day
rainfall depth equal to or greater than 280 mm occurring (a) once in 20 successive years, and (b)
two times in 15 successive years.

Solution: Here, P = 1/50 = 0.02.

a) In the first case, n = 20, r = 1. Therefore, from eq. (39)
20!
P1, 20  * (0.02) * (0.98)19  0.272.
19!1!
b) In this case, n = 15, r = 2. Therefore,
15 !
P2 ,15  * ( 0 . 02 2 ) * ( 0 . 980 ) 13  0 . 0292 .
13 ! 2!

Example 10.10: What is the probability that a 5-year flood will not occur at all in a 10-year
period?

Solution: Here, p = 1/5 = 0.2, n = 10, and r = 0. Hence the probability is

10 !
P0 ,10  * 0 . 2 0 * ( 0 . 8 ) 10  0 . 1074
0! 10 !

10.4.7 Poisson Distribution

The Poisson distribution is a limiting form of the binomial distribution when p is very small and
n is very large, and np tends to a constant value . This may happen when the interval over
which the Bernoulli process is defined gets smaller and smaller and the number of trials becomes
greater and greater, keeping np constant. The Poisson distribution has only one parameter  that
denotes the expected mean frequency of occurrence of some event in a given time t. The
probability distribution of the number of events in a given time is

 x exp( )
P( X  x)  ,   0, x  0, 1, 2,... (10.46)
x!

The CDF of the Poisson distribution is

x
i exp( )
P( X  x)   (10.47)
i 0 i!

The conditions for application of Poisson distribution are: a) the number of events is
discrete, b) two events cannot coincide, c) the mean number of events per unit time is constant,
and d) events are independent. Thus, it can be applied to following situations with p relatively
small and n relatively large to determine the probability of:

(i) Droughts in a given time period,

(ii) Number of rainy days at a given location,
(iii) Probability of rare flood events, and
(iv) Probability of reservoir being empty in any one year out of a long period of record.
10.5 PARAMETER ESTIMATION METHODS
A number of methods have been developed to estimate parameters of hydrologic models. Some
commonly used methods in hydrology include: (1) method of moments; (2) method of
probability weighted moments; (4) L-moments; (5) maximum likelihood estimation; and (6) least
squares method. Each of these methods is discussed here.

10.5.1 Method of Moments

This method is very commonly employed to estimate parameters of linear hydrologic models.
This method is based on the premise that when the parameters of a probability distribution are
estimated correctly, the moments of the probability density function are equal to the
corresponding moments of a sample data. Nash (1959) developed the theorem of moments which
relates the moments of input, output and impulse response functions of linear hydrologic models.

Let X be a continuous variable and f(x) its function satisfying some necessary conditions.
The r moment of f(x) about an arbitrary point ‘a’ is denoted as M ra (f). The rth moment of the
th

function f(x) can be expressed as


M ra ( f )   ( x  a) r f ( x) dx

(10.48)

Fig. 10.10 shows the definition of various terms used in the above equation.

Consider the special case when r = 0. In this case, eq.(10.48) gives

 
M 0a   ( x  a ) 0 f ( x) dx   f ( x)dx  1 (10.49)
 

Thus, the zero-order moment is the area under the curve defined by f(x) subject to - <x < .
For probability distribution, this area is unity. If r = 1, eq. (10.48) yields

M 1a   ( x  a )1 f ( x) dx    a
 (10.50)

where  is the mean. If the moment is taken around the origin, then a = 0, and the first moment
gives the mean. When a = , the rth moment about the mean is expressed by

M r   ( x   ) r f ( x) dx

(10.51)
 f(x)

a x-a
- 0 d x 
x

Fig. 10.10 Concept of moment of a function f(x) about an arbitrary point.

For simplicity of notation, we drop the superscript if the moment is taken about the origin
0 and the familiar terminology of the moments can be written as follows:

M0 =Area
M1 = Mean
M2 = Variance,
M3 = Measurement of skewness of the function
M4 = Kurtosis,
These terms have already been defined earlier.

10.5.2 Method of Moments for Discrete Systems

For a discrete function, represented as fj, j = -,…, -1,0,1,…, , the rth moment about any
arbitrary point can be defined in an analogous manner as for continuous functions. The rth
moment about the origin, is defined as

Mr  m
m  
r
fm (10.52)

Fig. 10.11 depicts the concept of moment of a discrete function.

Example 10.11: The frequency table of annual flows of Sabarmati River is given Table 10.2.
Find the mean and variance of the data by using the method of moments.
 f(x
f -4) f(x-3) f(x-2) f(x-1) f(x
f 1) f(x2) f(x3) f(x4)

… -3 -2 -1 1 2 3 …


- 
x-2 x2 x
x-3 x3

Fig. 10.11 Concept of

o moment of
o a discrete function aboout an arbitrrary point ‘a’’.

Table 10.2 Frequenccy table of an

nnual flows of Sabarmatti River.

Disch
harge range Frequenccy Dischaarge range Frequenccy
10
00-200 6 2000-300 9
30
00-400 11 4000-500 9
50
00-600 9 6000-700 9
70
00-800 19 8000-900 6
900-1000 6 10000-1100 1
110
00-1200 5 12000-1300 2
130
00-1400 3 14000-1500 0
150
00-1600 2 16000-1700 0
170
00-1800 1

Solutionn: The sum ofo all frequenncies = (6+1 11+9+19+ ……) = 98. For the first raange of dischharge
(100-200 0), the mean value is 150
0; for the seccond range (2200-300), it is 250 and sso on.
Hence, thhe first mom
ment of the daata = (150*6 6 + 250*9 + 350*11 + … …+ 1750*1)//98
= 664.2 29 cumec.
This is th
he mean (xm) of the annuual flows.

The seco
ond moment about the mean
m gives th
he variance oof the data.
 N
( xi  x m ) 2 f ( xi )
Second M 
2  
i 1 N moment

= [ (150-664)2*6 + (250 – 664)2*9 + … + (1750-664)2*1]/98

= 120000 cumec2.
Hence, the standard deviation (s) = (120000)0.5 =346.61 cumec.
This leads to the coefficient of variation cv = s/xm =346.61/664.29 = 0.52.

10.5.3 Method of Maximum Likelihood (MLE)

The maximum likelihood (ML) estimation method is widely accepted as one of the most
powerful parameter estimation methods. Asymptotically, the ML parameter estimates are
unbiased, have minimum variance, and are normally distributed, while in some cases these
properties hold for small samples. The MLE method has been extensively used for estimating
parameters of frequency distributions as well as fitting conceptual models.

Let f(x; a1, a2,… am) be a PDF of the random variable X with parameters ai, i=1, 2, …, m,
to be estimated. For a random sample of data x1, x2, …xn, drawn from this probability density
function, the joint PDF is defined as
n
f ( xi , x 2 ,....x n ; a1 , a 2 ,....a m )   f ( xi ; a1 , a 2 ,....a m )
i 1 (10.53)

Interpreted conceptually, the probability of obtaining a given value of X, say x1, is

proportional to f(x; a1, a2,… am). Likewise, the probability of obtaining the random sample x1,
x2,… xn from the population of X is proportional to the product of the individual probability
densities or the joint PDF. This joint PDF is called the likelihood function, denoted by L.
n
L   f ( x1 ; a1 , a 2 ,...a m )
i 1 (10.54)

where the parameters ai, i=1,2,…m, are unknown.

By maximizing the likelihood that the sample under consideration is the one that would
be obtained if n random observations were selected from f(x; a1, a2, … am), the unknown
parameters are determined, and hence the name of the method. The values of parameters so
obtained are known as MLE estimators. Since the logarithm of L attains its maximum for the
same values of ai, i = 1, 2, … m, as does L, the MLE function can also be expressed as
n n
ln L  L*  ln  f ( x1 ; a1 , a2 , .... am )   ln f ( x1 ; a1 , a2 , .... am )
i 1
i 1
 (10.55)

Frequently ln[L] is maximized, for it is many times easier to find the maximum of the
logarithm of the maximum likelihood function than that of the normal L.

The procedure for estimating parameters or determining the point where the MLE
function achieves its maximum involves differentiating L or ln L partially with respect to each
parameter and equating each differential to zero. This results in as many equations as the number
of unknown parameters. For m unknown parameters, we get

L(a1 , a2 , ... am )
0
a1

L(a1 , a2 , ... am )
(10.56)
0
a2
L(a1 , a2 , ... am )
0
am

These m equations in m unknowns are then solved for the m unknown parameters.

Example 10.12: Using the method of maximum likelihood, find the parameter  of the
exponential distribution for the data of the Sabarmati River in India, given in Example 10.1.

Solution: The probability density function of the one-parameter exponential distribution is given
by
fX(x) = exp(-x) (10.57)

The likelihood function is given by

n
 n

L( )    exp(xi )   n exp    xi 
i 1  i 1  (10.58)
This can be used to form the log-likelihood function:

 n 
ln L( )  n ln( )     xi 
 i 1  (10.59)

where n is the sample size. Differentiating eq. (10.59) with respect to :

d ln L( ) n n
   xi  0
d  i 1
This yields
 n
 1
  n /   xi  
 i 1  x (10.60)

In Example 10.1, the mean of the data was found to be 664.29 cumec. Hence, the estimate of 
is:
 = 1/664.29 = 1.51×10-3 cumec-1.

10.5.4 Method of Least Squares

The method of least squares (MOLS) is one of the most frequently used parameter estimation
methods in hydrology. Natale and Todini (1974) presented constrained MOLS for linear models
in hydrology.

Let there be a function Y = f(X; a1, a2,… am), where ai, i = 1, 2, … m, are parameters to
be estimated. The method of least squares (MOLS) involves estimating parameters by
minimizing the sum of squares of all deviations between observed and computed values of Y.
Mathematically, this sum D can be expressed as
n n n
D   d12    y0 i   yc i     y0 i   f  x; a1 , a2 ,...am 
2 2

i 1 i 1 i 1 (10.61)

where y0(i) is the ith observed value of Y, yc(i) is the ith computed value of Y, and n > m is the
number of observations. The minimum of D in eq. (71) can be obtained by differentiating D
partially with respect to each parameter and equating each differential to zero, e.g.,
n
   y 0 i   f  xi ; a1 , a 2 ,...a m 
2

i 1
0 (10.62)
a1

The resulting m equations, usually called the normal equations, are then solved for estimation of
m parameters. This method is frequently used to estimate parameters of linear regression model.

10.5.5 Method of L-Moments

Greenwood et al. (1979) introduced the method of probability weighted moments (PWM) and
showed its usefulness in deriving explicit expressions for parameters of distributions whose
inverse forms X=X (F) can be explicitly defined. They derived relations between parameters and
PWMs for Generalized Lambda, Wakeby, Weibull, Gumbel, Logistic and Kappa distributions.
However, the probability-weighted moments characterize a distribution but are not meaningful
by themselves.

L-moments were developed by Hosking (1986) as functions of PWMs which provide a

descriptive summary of the location, scale, and shape of the probability distribution. These
moments are analogous to ordinary moments and are expressed as linear combinations of order
statistics, hence the name. They can also be expressed by linear combinations of probability-
weighted moments. Thus, the ordinary moments, the probability weighted moments, and L-
moments are related to each other. L-moments are known to have several important advantages
over ordinary moments. L-moments have less bias than ordinary moments because they are
linear combinations of ranked observations. As an example, variance (second moment) and
skewness (third moment) involve squaring and cubing of observations, respectively, which
compel them to give greater weight to the observations far from the mean. As a result, they result
in substantial bias and variance.

The first L-moment denoted as λ1 is the arithmetic mean:

λ1 = E[X] (10.63)

Let us consider a sample of size n and arrange the data such that X(i|n) is the ith largest
observation; clearly i = n will be the largest value. Then, for any distribution, the second L-
moment, λ2, is a description of scale based upon the expected difference between two randomly
selected observations:

λ2 = (1/2) E[X(2|1) -X(1|2)] (10.64)

To compute L-moment measures of skewness three randomly selected observations are used and
for kurtosis, we use four randomly selected observations.

λ3 = (1/3) E[X(3|3) - 2X(2|3) + X(1|3)] (10.65)

λ4 = (1/4) E[X(4|4) - 3X(3|4) + 3X(2|4) - X(1|4)] (10.66)

Sample L-moment estimates are often computed using (PWMs). The rth PWM is defined
(Loucks and Beek, 2005) as:

βr= E{X [F(X)]r } (10.67)

where F(X) is the cumulative distribution function of X. Recommended (Landwehr et al., 1979;
Hosking and Wallis, 1995) unbiased PWM estimators, br, of βr are computed as:
b0  X
1 n
b1    j  1X  j 
nn  1 j 2
1 n (10.68)
b2    j  1 j  2X  j 
nn  1n  2  j 3
1 n
b3    j  1 j  2( j  3) X  j 
nn  1n  2 (n  3) j 4

The general formula for computing estimators br of βr is given by

1 n
 j  1  n  1
br   
j  r 1  r 
X  j   
n  r 
(10.69)
1 n  j  1  n 
    X  j
r  1 j r 1  r 
 
 r  1
for r = 1, … , n -1.

L-moments are easily calculated in terms of PWMs using:

λ1 = β 0
λ2 = 2β1 - β0
λ3 = 6β2 - 6β1 + β0
λ4 = 20β3 - 30β2 + 12β1 - β0 (10.70)

As with traditional product moments, measures of the coefficient of variation, skewness and
kurtosis of a distribution can be computed with L-moments. Following L-moment ratios are
important:

L- coefficient of variation (L-CV) t2 = λ2 / λ1

L- coefficient of skewness (L-sk) t3 = λ3 / λ2
L- coefficient of kurtosis (L-ku) t4 = λ4 / λ2

Example 10.13: Table 10.3 gives annual discharge data of a river for 36 years. Compute sample
L-moments and L-moment ratios, L-CV, L-sk, and L-ku.
Table 10.3 Annual discharge data of a river for 36 years
Year  Discharge  Year  Discharge Year  Discharge Year  Discharge 
1950  400  1959  1390  1968  2291  1977  1499 
1951  1100  1960  3300  1969  1340  1978  2598 
1952  900  1961  2190  1970  3200  1979  3487 
1953  440  1962  935  1971  2200  1980  1234 
1954  3000  1963  785  1972  1014  1981  819 
1955  2500  1964  501  1973  1790  1982  1210 
1956  760  1965  1123  1974  1140  1983  1510 
1957  1250  1966  1581  1975  764  1984  1780 
1958  1340  1967  959  1976  783  1985  1398 

Solution: Equation (10.68) yields estimates of the first three Probability Weighted Moments:

b0 = 1514.19
b1 = 889.16
b2 = 655.38
b3 = 518.64 (10.71)

The sample L-moments can be calculated using the probability weighted moments to obtain:


 1 = b0 = 1514.19

 2 = 2b1 - b0 = 264.12

 3 = 6b2 - 6b1 + b0 = 111.53 (10.72)

 4 = 20b3 - 30b2 + 12b1 - b0 = -132.82

Thus, the sample estimates of the L-coefficient of variation, t2, and L-coefficient of skewness, t3,
are:

t2 = 264.12/1514.19 = 0.174
t3 = 111.53/264.12 = 0.422 (10.73)
t4 = -132.82/264.12 = -0.502
10.6 PROBLEMS OF PARAMETER ESTIMATION
The parameters of a distribution function are estimated from the available sample data. But while
doing so, errors may arise due to many reasons. The sample data may contain errors, the
assumption underlying a particular method of parameter estimation may not hold good, and there
may be truncation and round-off errors. All these may result in errors in estimates of parameter.
Each estimate of a parameter is a function of sample parameter data which are observations of a
random variable. Thus, the estimate value of the parameter itself is a random variable with
certain distribution. An estimate obtained from a given set of values can be regarded as an
observed value of the random variable. Thus, the goodness of an estimate can be judged from its
distribution.

Several questions arise in parameter estimation. How should we employ the available
data to obtain the best estimate? What is the best estimate? Are these estimates unique? A
number of statistical properties are available by which to address the above questions. These are
discussed below.

Bias
Let the estimate of parameter a be ac denoted by. Estimate ac will be called an unbiased estimate
of a if the expected value of a, denoted E (ac) = a. In general, an estimate will have a certain bias
b(a) depending on a so that

E(ac) = a + b (a) (10.74)

An unbiased estimate mean b(a) = 0. Note that an individual ac may not be equal to or
close to a even if b(a) = 0. Unbiasedness simply implies that the average of many independent
estimates of a will be equal to a.

The bias in a given quantity is usually measured in dimensionless terms and is often
referred to as standardized bias (or BIAS). Thus, BIAS is defined as

E aˆ   a
BIAS  (10.75)
a

where â is an estimate of parameter or quantile of a. In Monte Carlo experimentation, large

numbers of samples of different sizes are generated from a given population. For each sample,
then, an estimate of a is obtained. If there are, say, 1000 samples of a given size generated then
there are 1000 values of parameter a. Thus, E(a) is the average of the 1000 estimates of a for a
given sample size and is estimated as
n
Eâ    â i / n
i 1 (10.76)

where n is the number of samples generated or the number of values of the a estimate. The value
of a in eq. (10.75) is the true value of a or the value of parameter a of the population.

Efficiency
An estimate ac of a is said to be efficient if it is unbiased and its variance is at least as small as
that of any other unbiased estimate of a. If there are two estimates of a, say a1 and a2, then the
relative efficiency of a1 with respect to a2 is defined as

E (a1  a ) 2
e 1 (10.77)
E (a2  a) 2

if E (a2  a) 2  E (a1  a) 2 , then e  1. An efficient estimate has e = 1. If an efficient estimate

exists, it may be approximately obtained by use of the MLE or entropy method.

Standard Error
Another dimensionless performance measure frequently used in hydrology is the standard error
(SE), defined as

SE   aˆ  / a (10.78)

where  (.) denotes the standard deviation of a and is computed as

12
 1 n 2
 aˆ     aˆ i  E aˆ i  
 n  1 i 1 (10.79)

where the summations are over n estimates â of a. In Monte Carlo experiments, referred to as
above, for each sample size, a value of SE is obtained. Thus, this measure is similar to the
coefficient of variation.

Root Mean Square Error

The root mean square error (RMSE) is one of the most frequently employed performance
measures and is defined for parameter a estimate as
 
RMSE  E aˆ  a  
2 12
/a (10.80)

where E[. ] is the expectation of [. ]. It can be shown that RMSE is related to BIAS and SE as
12
n 1 2 
RMSE   SE  BIAS 2 
 n  (10.81)

Relative Mean Error

Another measure of error in assessing the goodness of fit of hydrologic models is the relative
mean error (RME) defined as
0.5
1  N  Q  Qc  
2

RME     0 
 (10.82)
N  i 1  Q0  
 

in which N is the sample size, Q is the observed quantity of a given probability and Qc is the
computed quantity of the same probability. Also, used sometimes is the relative absolute error
defined as

1 N
Q0  Qc
RAE 
N

i 1 Qc (10.83)
10.7 HYPOTHESIS TESTING
Many times, while analyzing water resources data, questions arise such as: does the river
discharge at the given site follow normal distribution? Is the river water quality violating the
relevant standards? Are the two given variables significantly correlated? In some cases, a
conclusion may be reached without using a statistical test but sometimes the situation could be
such that well-articulated tests are needed to arrive at a conclusion which is not subjective.
Statistical procedures known as hypothesis testing are employed in these situations.

There are two broad categories of hypothesis tests: parametric tests and non-parametric
tests. In parametric tests, the distribution of the data has to be known and is to be specified. The
examples are the t-test, and the F-tests. In the non-parametric or distribution free tests, it is not
required to specify the distribution of data or the data need not follow a specified distribution.
The Kendall’s Tau test and the Kruskal-Wallis test are the examples of these tests.

The parametric tests begin with the formulation of a hypothesis which is termed as null
hypothesis. It is denoted by H0. As the name suggests this is the hypothesis of no change or no
difference. For example, the null hypothesis could be that the flows at two stations follow the
same distribution or there is no correlation between two given variables. This null hypothesis
may be mathematically written as:

H0: 1 = 0 (10.84)

This statement indicates that the mean 1 of a data set is not different from the mean 0 of
another data.

After the null hypothesis has been chosen, an alternative hypothesis is formulated. The
alternate hypothesis is a statement of some expected departure from the null hypothesis. The
null and the alternate hypothesis are mutually exclusive and all inclusive. The alternate
hypothesis can be stated in the following forms

Ha : 1 > 0 (10.85 a)
Ha : 1  0 (10.85 b)
Ha : 1 < 0 (10.85 c)

Equations (10.85) state that the mean of the population from which the sample was drawn
is greater than, not equal to, or less than the specified population mean, respectively. Some
typical ex
xamples of null
n hypotheesis are: the distributionns of flow att two stationns are differeent or
the two variables
v und
der examinattion are relatted to each oother.

β α

Fig. 10.12 Hypothesiis testing forr the mean.

Since the rejeection of nulll hypothesiss implies thaat the speciffic statementt is not suppported
by the data,
d the altternate hypo
othesis mustt be sufficieently generaal. If the nuull hypothessis is
rejected, the truth liesl somewhhere in thee wide set of possibiliities stated in the alteernate
hypothessis. Outcomee of hypothesis testing iss stated as reejection of thhe null hypoothesis against the
failure to
o reject the null
n hypothessis.

After
A the hyppothesis has been formu ulated, the teest statistic is computed.. Test statisttic on
which the decision too accept or reject
r null hy
ypothesis is bbased is com mputed from m the sample data.
A rejectiion region iss the set of all test statiistic values for which H0 will be rejected. Thee null
hypothessis will be rejjected if and
d only if the test statistic value falls iin the rejectiion region.
The
T null hyp pothesis may y be correctt or incorrecct and it miight be acceepted or rejeected.
This prodduces four possible comb binations wh hich are indiicated in the Table 10.4.

The
T characterristics of the data that is to be examinned is quanttified by the test statisticc. The
test checks whether the t behaviorr of the test statistic is siimilar to whhat is expectted (leaving aside
the possibility of thiss to happen due
d to chancce alone) if H0 were true..

Table 10
0.4 Hypothesis testing: possible
p outccomes and thheir probabillities.

Hypothesis
H is correct Hypotheesis is incorrrect
 Hypothesis is Correct decision. This Type II error with a
accepted outcome has a probability probability 
1-
Hypothesis is Type I error with a Correct decision with
rejected probability of  probability 1-

Acceptance of a correct hypothesis or rejection of a wrong hypothesis is a right decision.

If, however, a null hypothesis which is correct is rejected, type I error is said to have been
committed. In statistics, the probability of making type I error is termed as level of significance.
This probability is specified before carrying out the test. Thus the significance level  denotes
the maximum allowable type I error. Whenever H0 is correct, its probability of rejection is less
than .

In hydrology, commonly the significance level of 0.05 (1 in 20) or 0.01 (1 in 100) are
adopted. A level of 0.05 implies that the decision of statistical test may be in error one time out
of 20. In terms of normal distribution properties, it corresponds to 5% of the area under the
curve. This concept is illustrated for a two-sided test in Fig. 10.13 in which the test statistic under
null hypothesis is normal and each shaded area near the tails contains 2.5% of the total area. This
shaded area is termed as the area of rejection or the critical region. For the inequality type,
alternative hypothesis in eq. (10.85b) the null hypothesis is rejected if the test statistic falls in the
critical region either because it is too high or too low. The choice of significance level depends
upon the risks associated with a wrong decision.

If an incorrect hypothesis is accepted, this leads to a type II error, denoted by . The

probability of making type II error is not known; this error increases as significance level 
increases. The null hypothesis is formulated with the intention that it will be rejected because this
eliminates the possibility of a type II error.

The steps in hypothesis testing can be written as:

1. Identify the parameter of interest and describe it.
2. Determine and state the null hypothesis and appropriate alternative hypothesis.
3. Selected the significance level  and state the rejection region.
4. Compute the necessary sample quantities and compute the test statistic.
5. Decide whether H0 should be accepted or cannot be rejected.
10.7.1 The t-distribution
The Student’s t-distribution is similar to the normal distribution but its shape depends upon the
size of sample. As the number of observations in the sample increases, the t-distribution
approaches the normal distribution.
An important concept in statistical tests pertains to degree of freedom. Actually, in
statistical tests, the same sample is used to estimate the parameters of the distribution as well as
perform the test. Thus, there are multiple uses of observations. This limitation is tackled by using
the concept of degrees of freedom. Degree of freedom is the number of observations in a sample
less the number of parameters being estimated. For example, if we have 20 data which are used
to estimate two parameters, the degree of freedom will be 18. The tables of t-distribution list the
value of t statistics corresponding to various levels of significance and the degrees of freedom
().

Values of t-distribution for selected degrees of freedom and significance level are given
in Table 10.5 for ready reference. The value of the t-distribution can be read from the row
corresponding to  and the column corresponding to the significance level . For example, for 
= 10 the t value is 1.81 for 5% significance level. This implies that 95% of the area of the curve
lies to the left of value 1.81. Since the t-distribution is symmetric, 5% of the area in the left tail is
to the left of t value of –1.81 for  = 10. For the case where one is interested in 95% of the area
but with 2.5% in each tail, the critical t value for  = 2.5 is 2.23 for  = 10.

Table 10.5 Values of t-distribution for selected degrees of freedom and significance level

Degrees of Significance level  (%)

freedom  10 5 2.5 1
1 3.08 6.31 12.71 31.82
2 1.89 2.92 4.30 6.96
3 1.64 2.35 3.18 4.54
4 1.53 2.13 2.78 3.75
5 1.48 2.02 2.57 3.36
8 1.40 1.86 2.31 2.90
10 1.37 1.81 2.23 2.76
15 1.34 1.75 2.13 2.60
20 1.32 1.72 2.09 2.53
30 1.31 1.70 2.04 2.46
 60 1.30 1.67 2.00 2.39
120 1.29 1.66 1.98 2.36
 1.282 1.645 1.96 2.32

We now consider some applications of t-distribution

A) Test concerning population mean when 2 is known

This test is conducted as follows:
a) Form the null hypothesis H0: 1 = 0
b) State the alternative hypothesis Ha
X  0
c) Test statistic is computed as t 
/ n
d) For rejection region corresponding to , find c such that P{| X   0 | c}  
e) Decision: depending upon the alternative hypothesis, reject H0 if t-statistic falls in the
rejection region as follows:

Alternative Hypothesis Rejection Region

Ha : 1 > 0 t ≥ t, n-1
Ha : 1 < 0 t ≤ t, n-1
Ha : 1  0 t ≥ t/2, n-1 or t ≤ -t/2, n-1

If the test is conducted when population mean 2 is not known, the value of sample standard
deviation is used in the formula in step c).

B) Test concerning difference in two means (variances are unknown)

This test is conducted as follows:
a) Set the null hypothesis H0: 1 - 0 = ∆0.
b) State the alternative hypothesis Ha
s12 s 22
c) Test statistic is computed as t  X  Y   0   where X and Y are the two
m n
series with m and n values, respectively.
d) For rejection region corresponding to , find c such that P{| X   0 | c}  
e) Decision: depending upon the alternative hypothesis, reject H0 if t-statistic falls in the
rejection region as follows:
 Alternative
A Hypothesis
H Rejectiion Region
Ha : 1 > 0 t ≥ t, n-1
Ha : 1 < 0 t ≤ t, n-1
Ha : 1  0 t ≥ t/2, n-1 orr t ≤ -t/2, n-1

Fig. 10.13 Rejection regions for z tests: a) fo

or upper-tailled tests, b) lower- tailedd tests, and cc) for
two-tailed test.

Examplee 10.14: Asssume that the annual run noff data at a site follow
ws normal diistribution annd 21
observatiions are avaailable. For these observ
vations, the mean xm iss 15.0 mm and the stanndard
deviation
n sd is 5.0 mm.
m Now, whether,
w at 5% significannce level, thhe mean annuual runoff caan be
considereed to be draw
wn from a poopulation whhose mean iss 17.0 mm?

n: The null hypothesis is H0 :  = 17

Solution 7.0 mm againnst the alternnate hypotheesis Ha :   17.0
mm.

The
T test statisstic is
t = [n (xm – )/sd]0.5 = [21
1*(15.0 – 17.0)/5.0]0.5 = -1.83.
 Since this is a two-tailed test, the null hypothesis will be rejected if the test statistic is
either too high or too low and hence value of t is needed for /2 and  = n - 1. From Table 10.5,
t/2,n-1 = t2.5,20 = 2.09. Since |t| = 1.83 which is less than 2.09, the statistic does not fall in the
region of rejection and the null hypothesis is accepted.

10.7.2 Chi-Square Distribution

Chi square distribution is also frequently used in hypothesis testing. Let there be a sample of size
n whose values are taken from a normal population having a mean  and standard deviation .
The observations can be standardized using the relation

Z = (X - )/  (10.86)

If these standardised values are squared and added they follow a new statistic:
n
Y   Z i2
i 1 (10.87)

The variable Y follows a chi-square (2) distribution with n degrees of freedom. This
distribution is a special case of the gamma distribution and has a single parameter. However, this
distribution is not symmetric and is always positive. The chi-square tests are single-tailed and
the region of rejection is near the right tail. Table 10.6 lists chi-square values for selected degrees
of freedom. For example, for 10 degrees of freedom, 5% of the area in the right tail (region of
rejection) from 2 values is from 18.31 to .

Table 10.6 Values of Chi-square distribution for selected degrees of freedom and significance
level

Degrees of Significance level  (%) Degrees of Significance level  (%)

freedom  10 5 2.5 1 freedom  10 5 2.5 1
1 2.71 3.84 5.02 6.63 2 4.60 5.99 7.38 9.21
3 6.35 7.82 9.35 11.34 4 7.78 9.49 11.14 13.28
5 9.24 11.07 12.83 15.09 8 12.02 14.07 16.01 18.48
10 15.99 18.31 20.48 23.21 15 22.31 25.00 27.49 30.58
20 28.41 31.41 34.17 37.57 30 40.26 43.77 46.98 50.89
40 51.81 55.76 59.34 63.69 50 63.17 67.50 71.42 76.15
100 118.50 124.34 129.56 135.81
 The goodness of fit test determines whether it is appropriate to use a particular
distribution for the given sample data. Visual judgment is one way in which the data are plotted
on an appropriate probability paper to check whether the match is acceptable or not. The chi-
square test is also widely used for this purpose. The test procedure consists of dividing the
sample into a number of segments or classes depending upon the data range. For each segment,
the actual number of observations and the expected according to the distribution under test are
computed. The test statistic is

k
 c2   (Oi  Ei ) 2 / Ei
(10.88)
i 1

where Oi and Ei are the observed and expected number of observations in the ith segment and k is
the total number of segments. If p parameters are estimated from data, 2c follows a chi-square
distribution with (k – p – 1) degrees of freedom. If the difference between the actual and
expected observations in the segments is large, it implies that samples were not drawn from the
assumed distribution. Therefore, the null hypothesis that the observations follow the assumed
distribution is rejected if 2c > 21-,k-p-1.

Example 10.15: In a goodness of fit test, the data were divided in 11 classes and the value of 2c
came out to be 10.44. If two parameters of the distribution were computed, test whether the
chosen distribution is appropriate for the data at a significance level of 0.1?

Solution: The degree of freedom is 11 – 2 – 1 = 8. From the table of chi-square values, 210,8 =
12.02. Since this value is greater than 10.44, the null hypothesis cannot be rejected. It is,
therefore, concluded that the chosen distribution properly describes the behavior of data.

10.7.3 Tests Concerning Variances of Two Populations

Here, a statistical test is employed to test if the variances of two populations are same or not. A
distribution known as the F distribution is employed for this purpose and the test is known as the
F-Test.
a) Null hypothesis: H0: 12 = 22
b) Test statistic is computed as: F = s12/s22
c) Decision: depending upon the alternative hypothesis, reject H0 if t-statistic falls in the
rejection region as follows:

Alternative Hypothesis Rejection Region

 Ha: 12 > 22 F ≥ F, m-1, n-1
Ha: 12 < 22 F ≤ F1-, m-1, n-1
Ha: 12  22 F ≥ F/2, m-1,n-1 or F ≤ F1-/2, m-1,n-1

10.8 Closure
After the measured data of some variable(s) becomes available, the question arises as to what are
key characteristics of the population, what useful information can be derived from the data,
which probability distribution best represents the data. Further, one may also like to statistically
test the validity of a hypothesis. This module deals with these topics.

REFERENCES
Chow, V.T., Maidment, D.R., and Mays, L.W. (1988). Applied Hydrology, McGraw-Hill Book
Company, New York.
Haan, C.T. (2002). Statistical Methods in Hydrology. Iowa State Press, Ames, U.S.A.
Jain, S.K., and Singh, V.P. (2003). Water Resources Systems Planning and Management.
Elsevier, Amsterdam.
Loucks, D.P., and Beek, E.V. (2005). Water Resources Systems Planning and Management- An
Introduction to Methods, Models, and Applications. Studies and Reports in Hydrology,
Unesco, Paris.
McCuen, R.H. (1989). Hydrologic Analysis and Design. Prentice Hall, New Jersey.
Yevjevich, V. (1972). Probability and Statistics in Hydrology. Water Resources Publications,
Fort Collins, Colorado.