13 RAINFALL – RUNOFF RELATIONSHIPS

13.1 Runoff
Runoff is the drainage of precipitation from a catchment, which flows through its drainage system. This
occurs when the infiltration capacity of an area’s soil has been exceeded.

Surface runoff is assumed to take two components: Channel precipitation and interflow. Channel
precipitation is that portion of the total catchment precipitation that flows directly through the small
natural channels on the land surface to the main drainage channels. A part of the infiltrated rainwater
moves parallel to the land surface as sub surface flow, and reappears on the surface at certain other points.
Such flows are called interflows. The other part of the infiltrated water percolates downwards to ground
water as base flow and moves laterally to emerge in depressions and rivers and joins surface flow. The
ground water contribution to flood flow is quite different from surface runoff.

13.1.1 The contribution of base flow to stream discharge

Since base flow represents the discharge of aquifers, changes occur slowly and there is a lag between
cause and effect and that can easily extend to periods of days or weeks. This will depend on
transmissibility of the aquifers bordering the stream and climate.

An influent stream shown in Fig 13.1 is one where the base flow is negative; that is the stream feeds the
ground water instead of receiving from it, for example irrigation channels. The negative contribution is
taking place at the expense of contributing aquifers on other parts of the stream, since there can be no
baseflow from a wholly influent stream. Such a stream will dry up completely in rainless periods and is
called ephemeral stream.

H igh w ater level

Low w ater level

seepage seepage
lines G round w ater lines
recharge

G .W .T G round w ater m ound

Influent stream s

Fig 13.1 Influent Stream

An effluent stream shown in Fig13.2 on the other hand is fed by the ground water and acts as a drain for
bordering aquifers.
 B a n k s to ra g e

H ig h w a te r le v e l

G .W .T
G .W .T

L o w w a te r le v e l
G ro u n d w a te r fe e d s
G ro u n d w a te r fe e d s
th e s tre a m (B a s e flo w )
th e s tre a m (B a s e flo w )

E fflu e n t s tre a m s

Fig 13.2 Effluent stream

Bank storage: This is the portion of runoff in a rising flood that is absorbed by the permeable boundaries
of a watercourse above the normal phreatic surface.

13.1.2 Rainfall Excess and Effective Rainfall

Rainfall Excess: If the initial loss and infiltration losses are subtracted from the total rainfall, the
remaining portion of the rainfall is called the rainfall excess. Surface runoff occurs only when there is
rainfall excess.

Effective rainfall: The effective rainfall is that portion of rainfall, which causes direct runoff. As direct
runoff includes both the surface runoff and the interflow, the effective rainfall is slightly greater than the
rainfall excess.

13.2 Factors affecting Runoff

The factors affecting runoff from a catchment may be broadly classified into two categories:

13.2.1 Climatic Factors

i. Form of precipitation: If the precipitation is in the form of rainfall, the runoff occurs quickly and
reaches a stream. On the other hand, if the precipitation is in the form of snow, the runoff is
delayed, as it would occur in the subsequent warm period.
ii. Intensity of precipitation: The runoff increases with the intensity of precipitation because an
intense rainfall occurs in a short duration and the evaporation and infiltration losses are relatively
small.
iii. Duration of precipitation: If rainfall occurs over a long period, the runoff is large because the
runoff is proportional to the total rainfall. Moreover, the infiltration capacity decreases with time.
iv. Rainfall distribution over the catchment: A large area extent of the rainfall over the catchment
produces large runoff. Heavy rainfall in the lower reaches of the catchment causes a rapid
 increase in the runoff. On the other hand, for heavy rainfall in the upper reaches of the catchment,
there is a slow increase in runoff.
v. Direction of storm movement: The runoff is more when the storm moves in the downstream
direction of a stream as compared to that when it moves in the upstream direction. If the storm
moves in the upstream direction, the runoff from the upper tributaries reaches the basin outlet
after the runoff from the lower tributaries.
vi. Antecedent precipitation index: The antecedent precipitation index (API) is a measure of the soil
moisture condition existing before the occurrence of the storm. The runoff is more if the API is
high and the soil quite wet.

13.2.2 Physiographic Factors

i. Type of soil: Coarse grained and loose soils have high infiltration capacity and produce less
runoff as compared to fine grained and compacted soils.
ii. Land use: Runoff depth depends on the land use. Rain falling on a thick forest or vegetated area
produces less runoff because of the interception, transpiration and infiltration losses as compared
to paved areas.
iii. Area of Basin: The rainfall intensity decreases as the area of the basin increases.
iv. Shape of catchment: The runoff from a fan shaped catchment is more than from a fern shaped
catchment. In a fan shaped catchment, tributaries tend to come together and join the main stream
near the centre of the area. In a fern shaped area, the tributaries are long and narrow. By the time
the peak discharges of the upper tributaries reaches the basin outlet, the runoff of the lower
tributaries has already left the basin.
v. Altitude of the catchment: The form of precipitation and the evaporation and transpiration
depend upon the altitude.
vi. Slope of the catchment: Infiltration losses are less in a catchment with a steep slope compared to
that in a flat slope. The surface flow and interflow are quite predominant in steep catchments,
whereas base flow is more predominant in flat areas.

13.2.3 Hydrographs
A hydrograph is a graphical representation of the relationship of stage, discharge, velocity or any other
property of flowing water with respect to time in a drainage channel or stream. It is regarded as an
integral expression of physiographic and climatic characteristics that govern the relation between the
rainfall and runoff of a particular drainage basin.

It shows the time distribution of runoff at the point of measurement, defining the complexities of the
basin characteristics by a single empirical curve.
The intensity of the rainfall, the concentration of the rainfall, the succession of storms, the abnormal
groundwater recession pattern and similar other causes determine the peak characteristics of a
hydrograph. A concentrated storm produces a single peaked skewed- distributed hydrograph. In other
cases multiple-peaked hydrographs are likely to occur. A multiple hydrograph can be separated into a
number of single peaked hydrographs, when required for analysis.

A typical single peaked hydrograph can be divided into four limbs as shown in Fig 13.3, which are:

• The approach limb, AB

• The rising limb, BC
• The crest segment, CE
• The recession limb, EG
 lag time h

D
Flow Discharge, m3/s
C
E

Rainfall

A F
B
G

Tpeak

tb

Time, h

Fig 13.3 Segments of a flow hydrograph

The approach limb is the point of rise of the hydrograph .The point D, is peak with the point C as the
point of inflection of the rising limb. The point E is the point of inflection of the recession limb. The
depletion of flow begins from point E as the water starts flowing out of storage only. The lower part of
the recession limb, FG, is the ground water recession curve showing a decreasing rate of groundwater
inflow.

Rising Limb: Before the beginning of a storm, the discharge in the stream is due to base flow, indicated by
the curve section AB. After the storm, the discharge in the stream gradually increases as the surface flow
increases. As the storm continues, more and more surface flow from the distant parts of the catchment
reaches the gauging point. The peak discharge usually occurs after some time of the occurrence of the
storm. The time period between the centroid of the hyetograph and the peak discharge is called the basin
lag, which depends upon the catchment and storm characteristics. The rising limb AB has a well-defined
point of rise, which is followed by an increasing discharge. The end of the rising limb is at the inflection
point B where the curvature of the curve changes.

Crest segment: There is another inflection point D on the other side of the peak. The segment BD of the
hydrograph between the two inflection points is called the crest segment. After reaching the peak C, the
discharge decreases. The peak discharge is required in various hydrological studies.

Recession limb: After the inflection point D, there is no inflow to the stream due to surface runoff. The
recession limb extends from point D to the point E, which represents the recommencement of the natural
base flow. Point D represents the condition of maximum storage. The shape of the recession limb depends
entirely on the characteristics of the catchment and is independent on the storm characteristics (Das,
2002).

13.3 Methods of Estimating Runoff

There are several methods for estimating runoff from a catchment and they broadly fall into two
categories. These are the peak runoff models and continuous models. Peak runoff models only estimate
the peak runoff and are required for the design of culverts, bridges, for the evaluation of flooding
potential and the design of stormwater conveyance structures such as sewer pipes. The most commonly
used method for estimating peak flow in urban hydrology is the Rational Method. Other peak runoff
models use regression equations and empirical methods (Chin, 2006). Continuous-runoff models are
used to estimate the entire runoff hydrograph from the rainfall excess, after initial abstraction, infiltration
and depression storage have been taken into account and are used in the design of reservoirs for
stormwater management systems. There are four types of continuous runoff models are used to estimate
the runoff hydrograph from the rainfall excess commonly used in engineering practice. These are i) Unit
Hydrograph models, ii) Time area models, iii) Kinematic wave models and iv) Non linear reservoir
models. They are not normally used to estimate peak runoff since their accuracy is dependent upon the
temporal resolution of the rainfall excess. The Unit Hydrograph Method is the most widely used method
for estimating the runoff hydrograph of a catchment (Viesmann and Lewis, 1996). The Rational method
and Unit Hydrograph method are discussed below.

13.3.1 The Rational Method

The rational method dates back to the middle of the 19th Century when an Irish engineer called Mulvaney
first wrote about its principles. It is based on the assumption that a constant intensity of rainfall is spread
over and area, and the effective rainfall is falling on the most remote part of the basin takes a certain
period of time, known as the time of concentration (tc), to arrive at the basin outlet. If the input rate of
excess rainfall on the basin continues for the period of time of concentration, then the part of excess rain
that fell on the most remote part of the basin will just begin its outflow at the basin outlet and with it, the
runoff will reach its ultimate and the maximum rate.

Consider a drainage basin which has rainfall of uniform intensity and of longer duration. On plotting the
relationship between the cumulative runoff rate and time, the rate of runoff shows a gradual increase from
zero to a constant value as shown in Fig. 13.4. The runoff increases with increase in flow from the remote
areas of the basin to its outlet. If the rainfall continues beyond the time of concentration, then there is no
further increase in the runoff, and it remains constant at its peak value. When the rainfall stops the
hydrograph goes into a recession period.

End of rainfall

Qp
Direct runoff, Q

Recession
period

Tc
Time

Fig 13.4 Runoff hydrograph due to uniform rainfall

The relationship for peak runoff rate is expressed as:

 Qo = CIA (13.1)
Where:
C = coefficient of runoff/rainfall
I = intensity of rainfall
A = area of the catchment (drainage basin)

i) Steady State Runoff

Consider a flat, fully, pervious rectangular catchment area, A. A depth of rain, i, falls in a time t. If there
were also an impervious wall along the edges of the catchment, and had no sewers, this rain would simply
build up over the area to a depth, i. The volume of water would be i x A. Imagine that the runoff is
flowing into a sewer inlet at point X with steady state conditions: water landing in the area and flowing
away at the same rate. The sewer will carry the volume of rain (i x A) at a steady, constant rate over the
time (t) of the rainfall.

So the flow rate Q

iA i
Q= And since the intensity of the rain, I = then
t t
Q = IA
Since catchments are not fully pervious, and there will be initial and continuing losses, the runoff
coefficient C can is introduced, to give Q = CIA

Adjusting for commonly used units gives:

Q = 2.78CIA (13.2)

Where Q = Peak runoff rate (m3/s)

I = rainfall intensity (mm/h)
A = Catchment area (ha)

ii) Catchment Area Measurement

The boundaries of the complete catchment to be drained can be defined with reasonable precision either
by field survey or use of contour maps. The catchment is divided into sub catchment areas draining
towards each pipe or group of pipes in the system. The sub areas can be measured using a planimeter if
using paper maps or automatically if using GIS based package.

iii) Runoff Coefficient

The dimensionless runoff coefficient C is the proportion of rainfall that contributes to runoff from the
surface. The coefficient accounts for the initial runoff losses (depression, storage,) continuing losses
(surface infiltration) and implicitly accounts for the hydrodynamic effects encountered as the water flows
over the catchment surface. Weighted average coefficients are needed for areas of mixed land use.
Generally the runoff coefficient encompasses all other factors that affect the surface runoff, except the
area and the intensity of the rainfall.
It is defined as:

C = Qo / AI (13.3)
iv) Time of Concentration (tc)
This is an important term in storm sewer design. It is defined as the time required for the surface runoff to
flow from the remotest part of the catchment area to the point under consideration. Each point in the
catchment has its own time of concentration. As shown in Fig 13.5, it has two components, namely the
overland flow time known as the time of entry, te, and the channel or sewer flow time, the time of flow tf.
Thus t c = t e + t f

Catchment

tf

te
Sewer or
drainage pipe

Fig 13.5 Time of concentration in a catchment

Time of entry: The time of entry will vary with catchment characteristics such as surface roughness, slope,
and length of flow path together with rainfall characteristics.

Time of flow: Time of flow is calculated from the sewer length and velocity of flow in the sewers.

For the determination of time of concentration the most widely used formula is the equation given by
Kirpich (1940). However, for small drainage basins, the lag time for the peak flow can be taken to be
equal to the time of concentration. The lag time can be determined by the Synder’s equation used for the
determination of synthetic unit hydrographs. The Kirpich formula is given as:

Tc = 0.01947 L0.77 S −0.385

(13.4)

Where:
Tc = time of concentration (min)
L= maximum length of travel of water (m)
S= slope of the drainage basin = H/L
H= difference in elevation between the most remote place in the basin and the outlet (m)

Nomographs from which Tc can be obtained are also available and T peak can also be read off. Tc can also
be obtained from:
 Tc = 0.0194( k ) 0.77 (13.5)

Where k = (L / H )
3

v) Critical Rainfall Intensity

The rainfall intensity that causes the catchment to operate at steady state needs to be known. This should
give the maximum flow from the catchment. The basis of the Rational method is an engineering worst
case. The duration of the storm must be at least the time of concentration; otherwise the maximum flow
would not be reached.

Rainfall
intensity

Flow

Time tc Time

Fig 13.6a Less than tc Fig 13.6b Greater than tc

Hydrograph response to different duration rainfall of same intensity

Fig 13.6a shows a hydrograph resulting from uniform rainfall with duration less than the time of
concentration and Fig 13.6b shows the hydrograph from the same catchment resulting from the same
rainfall but with infinite duration (Butler and Davies, 2002)

The formula for intensity of rainfall is expressed as:

I = KTra / ( (Tc + b) n )
(13.6)
Where:

I = intensity of rainfall
Tr = recurrence interval And a, b, n are constants. Note this is similar to Equation 5.2
Limitations
The Rational Formula follows the assumptions that:
i. The predicted peak discharge has the same probability of occurrence (return period) as the used
rainfall intensity (I),
ii. The runoff coefficient (C) is constant during the rain storm, and
iii. The recession time is equal to the time of rise.

Example 13.1
A catchment has an area of 6.0 km2. The average slope of the land surface is 0.007 and the
maximum travel depth of rainfall in the catchment is approximately 2.05 km. The maximum
depth of rainfall in the area with a return period of 25 years is tabulated below:

Time duration (min) 5 10 15 20 25 30 40 60

Rainfall depth (mm) 13 26 30 46 52 55 58 66

Consider that 2.2 km2 of the catchment area has cultivated sandy loam soil (C=0.25) and 3.8 km2
has light clay cultivated soil (C=0.75). Determine the peak flow rate of runoff by using the
Rational method.

Solution
The time of concentration is given by Kirpich’s equation
Tc = 0.01947 L0.77 s − 0.385
= 0.01947 (2200) (0.007 )
0.77 − 0.385
min
= 49.28 min
The maximum rainfall depth for 49.28 min duration would fall between the period 40-60 min
and is located at 9.28 min after the 40 min period at which the maximum rainfall depth is 60 mm,
as per the available data.
66 − 58
The rainfall depth during the 9.28 min period = x 9.28
20
= 3.7 mm
Therefore, at 49.28 min duration, the rainfall depth = 58 + 3.7 = 61.7 mm
The average rainfall intensity (during the period of time of concentration) =
max imum ra inf all depth
Tc
61.7
= x 60 = 75.12 mm/h
49.28
Runoff coefficient, C =
(2.2 x0.25) + (3.8 x0.75)
6.0
= 0.57
CIA 3
Peak runoff rate, Qp = m /s
3.6
1
= x 0.57 x 75.12 x 6.0
3.6
= 70.95 m3/s
13.3.2 The Unit Hydrograph
A major advance in hydrological analysis was the introduction of the unit hydrograph concept by an
American engineer called Sherman in 1932. He defined the unit hydrograph as the storm hydrograph
resulting from an isolated storm of unit duration occurring uniformly over the entire catchment area
producing unit (i.e. 1cm) depth of direct runoff. The unit duration is usually expressed in hours. The unit
duration of the storm depends upon the area of the catchment. Another key criterion of selecting the unit
duration is that the total period of surface runoff for that storm duration should not be less than that for
any storm of duration shorter than the unit duration. A unit hydrograph relates the direct runoff
hydrograph and effective rainfall.

The unit hydrograph method is based on the following assumptions:

i. There is a direct proportional relationship between the effective rainfall and surface runoff
ii. Superposition: the response to successive blocks of effective rainfall each starting at particular
times may be obtained by summing the individual runoff hydrographs starting at the
corresponding times.
iii. The effective rainfall-direct runoff relationship does not change with time.

Derivation of a Unit Hydrograph

The following data are required for deriving a unit hydrograph:
i. Recorded hydrograph of the storm runoff
ii. Recorded storm rainfall data
iii. Average depth of rainfall
iv. Mass rainfall curves of one or more recording rain gauge stations located in the watershed.

Procedure
The unit hydrograph derivation from a simple storm proceeds in the following stages:
i. The stream flow records are studied, and the storm hydrograph due to the selected storm is
obtained.
ii. The base flow is separated from the storm hydrograph to obtain the direct runoff hydrograph.
iii. The ordinates of the direct runoff hydrograph are obtained by subtracting the ordinates of the base
flow from the ordinates of the storm hydrograph.
iv. The volume of the direct runoff can be estimated from the ordinates O1, O2……On of the direct
runoff hydrograph by applying the trapezoidal rule.

⎡⎛ O0 + On ⎞ ⎤
Volume of direct runoff (m3) = tx60 x60⎢⎜ ⎟ + O1 + O2 + ..... + On −1 ⎥
⎣⎝ 2 ⎠ ⎦
Where t is the time interval (hours) between the successive ordinates of the direct runoff
hydrograph.

Because in a direct runoff hydrograph, the end ordinates are zero, the volume of direct runoff is
computed as
Volume of direct runoff= 3600xt O ∑
i. The depth of direct runoff is then computed by dividing the volume of direct runoff by the area of
the catchment.
ii. The ordinates of the direct hydrograph are divided by the depth, d, of direct runoff to compute the
ordinates of the unit hydrograph.
Uses of Unit Hydrographs
i. Once a unit hydrograph for a duration tr has been developed for a basin, the storm hydrograph for
that basin for any other storm of different intensity but the same duration can be easily developed.
ii. The unit hydrograph may be used for the development of maximum flood hydrograph from the
maximum rainfall for the design of spillways and larger hydraulic structures.
iii. The available stream flow records can be extended using the rainfall records, which are generally
available for a longer period.
iv. The unit hydrograph can be used for flood forecasting and flood warning based on the observed
rainfall in the basin.

Limitations of the Unit Hydrograph Theory

i. The unit hydrograph theory cannot be applied to catchment areas greater than 5000km2, because
for larger areas, the rainfall is not uniformly distributed and it does not remain constant.
ii. The principle of linearity of responses is not strictly correct.
iii. The direct-runoff hydrograph of a drainage basin due to a given pattern of effective rainfall does
not remain constant.
iv. The unit hydrograph cannot be applied if the major portion of the storm is in the form of snow.

Elements of a Unit Hydrograph

• Base width (T) - the period of direct surface runoff (due to a unit storm) of the unit hydrograph is
called the time base or the base width.
• Unit storm – the storm of unit duration (i.e. duration of the unit hydrograph) regardless of its
intensity is called the unit storm.
• Unit period – the time duration of the unit storm (i.e. duration of the unit hydrograph) is called
unit period.
• Lag time (tp) - the time from the centre of a unit storm to the peak discharge of the corresponding
unit hydrograph is called lag time.
• Recession time (Tr) - the duration of the direct surface runoff after the end of the excess or net
rainfall.

Diagrammatically this may be shown in Fig 13.7 as:

 Fig 13.7 Elements of a unit hydrograph

Example 13.2
The ordinate of a 5-hour hydrograph over a catchment are:

Time (hr) 0 5 10 15 20 25 30 35 40 45 50 55
m3/s 0 10 40 110 150 170 120 70 50 35 10 0

Derive the flood hydrograph in the catchment due to the storm given below.

Time from start of storm (hrs) 0 5 10 15

Accumulated rainfall (cm) 0 7.65 8.5 12.15

Φ-index in the catchment is 0.33cm/hr with a constant base flow of 40 m3/s.

Determine the effective rainfall.

Solution:
Serial no. of Time intervals in Cumulative Rainfall Infiltration during Excess
5hr interval hrs from the start rainfall (cm) during interval (cm) rain (cm)
interval
(cm)
1 0-5 7.65 7.65 1.65 6.0
2 5-10 8.50 0.85 0.85 0
3 10-15 12.15 3.65 1.65 2.0

Time Unit Hydrograph of Hydrograph of Total Total

from Hydrograph 6cm rain 2cm rain Hydrograph Hydrograph +
start Ordinates Base flow
0 0 0 0 40
5 10 60 60 100
10 40 240 0 240 280
15 110 660 20 680 720
20 150 900 80 980 1020
25 170 1020 220 1240 1280
30 120 720 300 1020 1060
35 70 420 340 760 800
40 50 300 240 540 580
45 35 210 140 350 390
50 10 60 100 160 200
55 0 0 70 70 110
The peak flow, i.e. the largest is used in the design = 1280 m3/s.

T-Hour Unit Hydrograph (TUH)

In a linear time-invariant system analysis, a unit volume-block input of excess rainfall of duration T (h) is
transformed into a T (h) unit hydrograph. The period T (h) determines the shape of the unit hydrograph. If
T (h) decreases, the intensity (1/T) increases, due to which the unit hydrograph gets more skewed.

S-Curves
An S-curve is the hydrograph of direct runoff, which occurs due to a continuous effective rainfall of 1cm
every T hours of uniform intensity of (1/T) cm per hour.

S-Curve

T-h Unit
hydrographs
Discharge (m3/s)

Time h

Fig 13.8 Superposed T-h unit hydrographs to form an S-curve

The S-Curve is obtained by adding together a series of unit hydrograph of T duration, each unit
hydrograph lagged by T hours with respect to the previous unit hydrograph as shown in Fig 13.8.

1
Rainfall Intensity, /T, cm/h

Rainfall excess for S-Curve S1

Rainfall excess for S-Curve S2

1 cm
Curve S1
T’-h
Curve S2
Discharge, m3/s

(S1-S2)
(S1-S2) T/T’

T’-h Unit hydrograph

T’
Time, h

Fig 13.9 Derivation of a T’-h unit hydrograph by the subtraction method of S-curves

S-curve is given by the sequential accumulation of ordinates of the T1UH. T1UH represents the surface
runoff hydrograph caused an effective rainfall of intensity 1/T mm/h applied indefinitely.

If a T2UH is required, the s curve is plotted at a distance T2 from the first S-Curve. S1-S2 the difference
between the two S curves displaced by T2 then represents the surface runoff from 1/T1 x T2.

∆St= the difference between the ordinates of the two S- Curves from surface runoff hydrograph produced
in the T2 hr by an effective rainfall of T2/T1 as shown in Fig 13.9.

The T2 h hydrograph will be given by ∆St/ (T2/T1) or T1/T2 x∆St.

It may be noted that the S-curve attains a constant discharge (Qe), at the end of the base period T of the
first unit hydrograph. Thus the number of unit hydrographs to be combined is (T/ tr) and not infinite as
one might expect.
⎛ A⎞
The equilibrium discharge is given by: Qe = 2.78⎜⎜ ⎟⎟ where A is the catchment area (km2) and tr is the
⎝ tr ⎠
duration of unit hydrograph (hours).

Changing the duration of a unit hydrograph

Procedure
1. Draw 1hr UH.
2. Offset abscissa by 1hr, and make S curve additions
3. Add (1) and (2) to obtain the S curve.
4. Offset the S curve by 1/2hr (by interpolation)
5. Obtain the difference (3) - (4)
6. From the relationship (S2 – S1)/(T2/T1) or ΔS T1/T2, we have T2 = ½ T1 = 1
ΔS T1/T2 = ΔSx2
Example 13.3

Given the 1hr UH derive the ½ hr UH.

Time 1hr UH S-curve S-curve S-curve offset by Difference ΔSx2

(m3/s) additions 1/2 hrs (3)-(4)= ΔS
0 0 0 0 0
1 0.66 0 0.33 0 0.33 0.66
0.66 0.33 0.33 0.66
2 0.80 0.66 1.06 0.66 0.40 0.80
1.46 1.06 0.40 0.80
3 0.99 1.46 1.95 1.46 0.49 0.98
2.45 1.95 0.5 0.98
4 0.72 2.45 2.81 2.45 0.36 0.72
3.17 2.81 0.36 0.72
5 0.49 3.17 3.41 3.17 0.24 0.48
3.66 3.41 0.25 0.50
6 0.10 3.66 3.71 3.66 0.05 0.10
3.76 3.71 0.05 0.10
3.76 3.76

Example 13.4
Given the 1hr UH, derive the 3hr UH.

Time 1hr UH S-curve S-curve S-curve offset by difference ΔST1/T2

(m3/s) additions 3 hrs (3)-(4)= ΔS
0 0 0
1 0.66 0 0.66 0.66 0.22
2 0.8 0.66 1.46 1.46 0.47
3 0.99 1.46 2.45 0 2.45 0.82
4 0.72 2.45 3.17 0.66 2.51 0.84
5 0.49 3.17 3.66 1.46 2.20 0.73
6 0.1 3.66 3.76 2.45 1.31 0.44
3.76

The unit hydrograph for effective rainfall of duration T is then plotted, and the area under the curve is
checked to see if the enclosed volume is equivalent to unit effective rainfall over the area of catchment.

When all the single-peaked storms have been analyzed and a corresponding number of unit hydrographs
obtained, it will be noted that no two are identical, though they will all have the same general shape.

One way that an average unit hydrograph may be constructed is by taking the arithmetic means of the
peak flows (Up) and the times to peak (Tp), plotting the average peak at the appropriate mean value of Tp,
and drawing the hydrograph to match the general shapes of the individual unit hydrographs. The resulting
average unit hydrograph is then checked to ensure that the enclosed volume of runoff equivalent to a unit
of effective rainfall.
For rainfall bar graphs of complex pattern, a more sophisticated rainfall separation procedure is needed.
In the figure an idealized separation is shown by a curved loss-rate line. At the beginning of a storm there
could be considerable interception of the rainfall and initial wetting of surfaces before the rainfall
becomes ‘effective’-that is, begins to form surface runoff.

The loss-rate is dependent on the state of the catchment before the storm and is difficult to assess
quantitatively. Two simplified methods of determining the effective rainfall are given. The Φ index
method assumes a constant loss rate of Φ mm from the beginning of the rainfall event: this amount
accounts for interception, evaporation loss and surface detention in pools and hollows. It could rightly be
considered, however, that there is a period of time after the commencement of the storm before any of the
hydrograph is considered lost, and there is continuing loss-rate at some level afterwards. In both methods
the rainfall separation line is positioned such that the hatched areas in the figure equates to the effective
rainfall depth.

A choice between the two methods depends on knowledge of the catchment but, as the timing of the
extent of initial loss is arbitrary, the fixing of the beginning of effective rainfall at the beginning of runoff
in the stream neglects lag time in the drainage process and is thus somewhat unrealistic. A constant loss-
rate, the Φ index, would therefore seem to be more readily applicable (Shaw, 1994).

13.3.3 Derivation of the Unit Hydrograph from Composite Storms

Very often, particularly on larger catchments, it is difficult to find, in the available records, enough single-
peaked storm events to provide a fair sample for analysis. Multi-peaked sequences of rainfalls and the
resultant hydrographs must then be used in the unit hydrograph derivation. If significant peak flows can
clearly be related to groups of higher rainfall values, it may be possible to separate the records into
distinctive individual events to be treated as single storms. However, composite effects with overlapping
storms hydrographs call for more complex treatment.

13.4 Other Runoff Estimation Techniques

This section provides an overview of more advanced methods that can be used to estimate the surface
runoff storm hydrograph in a location along the river. These methods include:
i. Hydro-dynamic methods which use the Kinematic approach.
ii. Linear models

13.4.1 Hydrodynamic Methods

The hydrodynamic method is a physically based method, which uses the equations of Barre de Saint
Venant of gradually varied flow to predict runoff hydrographs (Chow et al. 1988). The propagation of
flow is imagined to be in the form of kinematic waves which are used to describe flood movement in
rivers (Lighthill and Whitman, 1955). The Saint Venant equations then describe the propagation of these
waves. These equations are partial differential equations that allow the flow rate and the water level to be
computed in both time and space (Chow et al. 1988). They are based on solution of both the continuity
and momentum equations taking into account the geometrical and drainage characteristics of the
watershed (Ward and Trimble, 2004).

Continuity Equation – Conservation of Mass

δ H δHu δHv
+ + =n
δt δx δy (13.7)
Momentum equation – Conservation of momentum

δH δH
− ( Sox − Sfx ) = 0 − ( Soy − Sfy ) = 0
δx and δy (13.8)

Where: H(x,y,t) = water depth (m)

n(x,y,t) = net rainfall intensity (m/s)
Sox & Soy(x,y) = Surface slope (x & y direction)
Sfx & Sfy(x,y) = the friction slopes (x & y direction)
t = time (s)
u & v(x,y,t) = flow velocity (x & y direction) (m/s)

Details of the method can be found in several hydraulic and hydrological text books such as Chow et al.
(1988).

13.4.2 Linear Models

A Linear System
In a linear system, the amount of water, S stored in a hydrological system is assumed to be directly
proportional to the discharge as shown in Equation 13.7. The storage provides the attenuation of a flood
wave in Fig.13.10 and is related to the inflow and the outflow by the continuity equation in Equation
13.10 (Chow et al. 1988).

S = KQ (13.9)
dS
= I −Q
dt (13.10)

Where: S = Storage volume of the reservoir

K = A reservoir constant also called storage constant
Q = Outflow
I = Inflow

Inflow
Attenuation due to
temporary storage
Discharge

Outflow

Time
Fig 13.10 Effect of temporary storage in linear reservoirs
Several examples of linear models exist such as the unit hydrographs, the Nash cascade model and the
Muskingum model for flood routing when x = 0.

Response of linear systems

In order to determine the solution of linear systems on a continuous scale, It is important to understand
how linear systems respond to changes in input parameters such as the inflow, I. The solution of these
responses is based on two basic principles derived from methods for solving differential equations:
1. Principal of proportionality – If Q is the output of the system for input I , then c.Q is the output for
input c.I .
2. Principal of superposition - If
Q1 and Q2 are the outputs of the system for the inputs I1 and
I 2 respectively, then Q1 + Q2 will be the output for the input I1 + I 2 .

According to Chow et al. (1988), there are three common response functions:
• the impulse response function
• the step response function
• the pulse response function

(a) The Impulse Response function – U (τ )

The impulse response function is the response of a linear system to an input amount. It is normally
denoted as U (t − τ ) where t − τ is the time lag since the impulse was applied. And is shown in Fig
13.11.
INPUT/ RESPONSE

1
Impulse response
function U (t-τ)

τ Time

Fig 13.11 The impulse response function

Considering the principles of proportionality and superposition cited above, it is possible to determine the
response of a combination of discrete impulses as illustrated in Fig. 13.12
 5U (t-τ1) + 3U (t-τ2)
INPUT / RESPONSE

5 5U (t-τ1)
3U (t-τ2)
3

τ1 τ2 time

Fig 13.12 Response from a combination of discrete impulses

The response of a continuous input time function I (τ) can then be found by integrating the response of its
constituents.

t
Q(t ) = ∫ I (τ ).U (t −τ ).dτ
0 (13.11)

Equation 13.11 is also called a convolution integral.

(b) The Step Response function – S (t )

The step response function describes the response of a linear system to a unit step input in Fig. 13.13. The
unit step input is defined as:
⎧ I (t ) = 0 for t<0 ⎫
I (t ) = ⎨ ⎬
⎩ I (t ) = 1 for t>0 ⎭ (13.12)
Where i (t ) is the input rate

The output of the unit step input is then obtained from Equation 13.11 as;

⎧0 for t<0 ⎫
⎪t ⎪
Q(t ) = S (t ) = ⎨ ⎬
⎪ ∫ U (t − τ ).dτ for t>0 ⎪
⎩0 ⎭ (13.13)

The step response function therefore essentially corresponds to the integral of the impulse response
function up to that time.

t
S (t ) = ∫ U (t − τ ).dτ
0 (13.14)
 Unit step input

INPUT / RESPONSE
1

Unit step response

function

Time

Fig 13.13 The step response function

(c) The Pulse Response function –P(t)

The pulse response function describes the response of a linear system to a unit pulse input (Fig. 13.14). A
unit pulse input is an input amount occurring in a duration Δt.

Unit pulse input

1
INPUT / RESPONSE

S(t-Δt)
S(t)

Unit step response

function

t Time

Δt

Fig 13.14 The pulse response function

As shown in the Fig.13.14, the response to a pulse input is found by applying the principles of
superposition and proportionality to an input consisting of:

• A step input with intensity 1/Δt starting at t = 0

• A step input with intensity 1/Δt starting at t = t

The desired pulse response, p (t) is then determined by subtraction of these two step responses.

Δt (
p(t ) = 1 s(t ) − s(t − Δt )
(13.15)

In conclusion, the knowledge of at least one of these functions is sufficient to describe a linear system.
13.4.3 Examples of Linear Models
i) The Unit Hydrograph
As described in section 13.3.2, a unit hydrograph (UH) is defined as a hydrograph resulting from 1 cm of
excess rainfall occurring during a storm of a particular duration. Thus, it represents a unit pulse response
function of a linear hydrologic system (Chow et al, 1988). It is so far the most widely used and simplest
method for predicting a storm hydrograph resulting from a storm (Subramanya, 2001; Duggal and Soni,
2007).

The S-curve, which was also described in section 13.3.2 represents the step response function of a linear
system. It is used to convert a UH of a given duration to UH’s of other durations.

If the duration of a given UH is reduced to an instant, then this UH is called an Instantaneous Unit
Hydrograph (IUH). The Instantaneous unit hydrograph is therefore defined as a unit hydrograph which
results from application of excess rainfall occurring instantly over the watershed. Thus, it represents the
impulse response function of the system. It is used to derive the UH for conceptual overland flow models
and for analysing characteristics of the surface runoff in a river basin, independently of the duration of the
rainfall.

A more detailed analysis of the unit hydrograph is provided in several hydrologic text books such as
Chow et al. (1988).

ii) Synthetic Unit hydrographs

Unit hydrographs are generally derived from records of observed rainfall and stream flow. However, for
small watersheds where no such information is available, synthetic unit hydrographs are generally used
(Ward and Trimble, 2004; Raghunath, 1995). These unit hydrographs use the relation between the
physical geometry of the basin and the resulting hydrographs. Four basic parameters are needed to
develop a synthetic unit hydrograph as shown in Fig. 13.15:
i. Peak discharge, qp - maximum flow rate on the hydrograph
ii. Base time with, tp - is the time from the start of the hydrograph to the peak discharge
iii. Base time with, tb - is the total time duration of the hydrograph
iv. Lag time, t L - is the time from the center of mass of the effective rainfall, to the peak of the runoff
hydrograph.

Fig 13.15 Components of a synthetic unit hydrograph

 Once these parameters are known, curves with unit hydrographs can be drawn.

Several methods have been developed to construct synthetic unit hydrographs using the above mentioned
parameters. These include:

ƒ Snyder unit hydrograph: This is the best known approach for synthesizing hydrographs of un-
gauged areas. Snyder (1938) generated a number of unit hydrographs and analyzed them to
develop a set of empirical relationships connecting elements of unit hydrographs with the basin
characteristics to get significant points on representative hydrographs. This method is further
detailed in several text books.

ƒ Triangular unit hydrograph: This method developed by the U.S Department of Agriculture
(1972) SCS models the shape of a unit hydrograph as a triangular shape. In this method, a single
triangle can be developed by approximating the base time as 2.67 times the peak time tp.

The application of these methods showing their linear responses is provided in several hydrology text
books.

iii) Time Area Method

The Time-Area Method (Clarke, 1945) can be used in design as an extension to the Rational Method and
to develop the Unit Hydrograph, based on the construction of the isochrones within the river basin. The
isochrones are lines that connect the points at an equal travel time from the basin outlet. The approach is
illustrated in Fig. 13.16 A.

Isochrones
A

A4 A3 A2
A1
Outlet

3Δt
2Δt Δt
B C
Area, A

dA/dt

0 1 2 3 4 1 2 3 4
T im e, t Time, t
Fig 13.16 The Time-Area method
The isocrones make it possible to construct a graph showing the cumulative basin area as a function of the
travel time as shown in Fig. 13.16B. It is assumed that this curve represents the S-hydrograph of the river
basin. The UH is then obtained by differentiating the S-hydrograph (Fig13.16C).

Routing
The linear reservoir at the outlet is assumed to be described by S = KQ , where K is the storage time
constant. The value of K can be estimated by considering the point of inflection Pi of a surface runoff. At
this point the inflow into the channel has ceased and beyond this point the inflow is entirely due to
withdrawal from the channel storage. The continuity equation
ds
I −Q = (13.16)
dt
ds dQ
becomes −Q = =K (13.17)
dt dt
Hence K = −Qi / (dQ / dt )i (13.18)
where suffix i refers to the point of inflection and K can be estimated from a known surface runoff
hydrograph of the catchment. The constant K can also be estimated from the data on the recession limb of
a hydrograph. Knowing K of the linear reservoir, the inflows at various times are routed by the
Muskingum method. Note that since a reservoir is used x = 0 in Muskingum’s equation. The inflow rate
between an inter-isochrome area Ar km2 with a time interval Δt c (h) is
Ar x 10 4 A
I= = 2.78 r (m3/s) (13.19)
3600Δt c Δt c
The Muskingum routing equation would now be
Q2 = C 0 I 2 + C1 I 1 + C 2 Q1 (13.20)
where C 0 = (0.5Δt c ) / (K + 0.5Δt c )
C1 = (0.5Δt c ) / (K + 0.5Δt c )
C 2 = (K − 0.5Δt c ) / (K + 0.5Δt c )
i.e. C 0 = C1 . Also since the inflows are derived from the histogram I 1 = I 2 for each interval. Thus Eq.
(13.20) becomes
Q2 = 2C1 I 1 + C 2 Q1 (13.21)
Routing this equation gives the ordinates of IUH fro the catchment.

Example 13.1
A drainage basin has the following characteristics: Area = 117km2, time of concentration = 16h, storage
constant = 10h and inter-isochrone area distribution as below:
Travel time t (h) 0-2 2-4 4-6 6-8 8-10 10-12 12-14 14-16 16-18
Inter-Isochrone area
(km2) 4 8 22 24 16 19 12 8 4
Determine the IUH of this catchment.

Solution:
K = 10h , t c = 16h , Δt c = 2h
0.5 x 2
C1 = = 0.091
10 + 0.5 x 2
 10 − 0.5 x 2
C2 = = 0.818
10 + 0.5 x 2
Therefore Q2 = 0.182 I 1 + 0.818Q1
= Ordinate of IUH
At t = 0, Q1 = 0
I 1 = 2.78 Ar / 2 = 1.39 Ar m3/s

Area Ar Ordinate of IUH

Time (h) (km2) I (m3/s) 0.182 I 1 0.818Q1 (m3/s)
1 2 3 4 5 6
0 0 0 0 0 0
2 4 5.56 1.01 0 1.01
4 8 11.12 2.02 0.83 2.85
6 22 30.58 5.57 2.33 7.90
8 24 33.36 6.07 6.46 12.53
10 16 22.24 4.05 10.25 14.30
12 19 26.41 4.81 11.70 16.50
14 12 16.68 3.04 13.50 16.54
16 8 11.12 2.02 13.53 15.55
18 4 5.56 1.01 12.72 13.73
20 0 0 0 11.23 11.23
22 9.19 9.19
24 7.52 7.52
26 6.15 6.15
28 5.03 5.03
and so on and so on

iv) Linear Reservoir Models

A linear reservoir model assumes that the watershed acts as a reservoir in which the storage, S is a
function of output, Q. Linear reservoir models are generally based on two principles:

(a) The Continuity equation:

dS
= I -Q
dt
(b) Storage equation
S = KQ

Where I is the inflow rate to the reservoir.

The Single Linear Reservoir:

On combining the equations above, we get:
d ( KQ ) dQ
I −Q = =K
dt dt (13.22)
dQ
Q + K. =I
dt (13.23)
Considering the initial time, to and the initial outflow, Qo, the solution of the above equation becomes:
t to −t
1 −t t
Q = e k ∫ I .e k .dt + Qo e K
K to
(13.24)
For to = 0 and Qo = 0, then the equation above reduces to:
t
−
Q = I (1 − e k ) (13.25)

The instantaneous unit hydrograph of this model is expressed as (dQ/dt, I=1);

1 − kt
Q= e
k (13.26)
The model only requires the knowledge of the parameter K. If no data are available for the determination
of K, empirical relations can be used.

v) The Nash-Cascade Model - Reservoirs in Series:

The Nash (1957) Model (considers the watershed as identical linear reservoirs in series (Fig.13.17). Each
reservoir is considered to have the same storage constant, K. The reservoirs are connected in series and
the outflow from one reservoir becomes the input to the next reservoir. The first reservoir receives a unit
volume equal to 1 cm of effective rainfall of a catchment instantaneously. The inflow is routed through
the first reservoir to get the outflow hydrograph which becomes the inflow to the second reservoir and so
on. The outflow hydrograph from the nth reservoir is taken as the instantaneous unit hydrograph (IUH) of
the catchment.

Flow from storage

q1
Stagewise
formation of IUH
q2 after convolution

q3

qn-1
Final Final IUH
Outflow
qn

Fig 13.17 The Nash Cascade model

Mathematically, the Nash model can be derived as follows:

Considering a system at rest (at time t0 = 0, q0=0), the outflow of the first reservoir for a step input with
intensity = 1 is given by Equation 13.25

t
Q1 = ( 1 - e - K1 ) (13.27)

To calculate the outflow from the second reservoir, the outflow from the upper reservoir is considered as
the input of the second reservoir.

The general equation for the outflow of the second reservoir (if the system is at rest at t=0) is obtained
from Equation .13.27:
1 -t t t
Q2 = e K 2 ∫ 0 I 2 e K 2 dt
K2 (13.28)

Substituting Equation. 13.27 in Equation. 13.28 yields

t
1 -t t − t
e K2 ∫ 0 (1 − e 1 ) e K2 dt
K
Q2 =
K2 (13.29)
etc.

For n reservoirs with the same storage constant, the instantaneous unit hydrograph of this model may then
be expressed as:

1 t n-1 - t
u(t) = ( ) eK
K(n - 1)! K (13.30)

(n-1)! may be replaced by the Gamma function.

The parameter K2 and n are evaluated by method of moments. K2 is a constant in the Nash’s model
because of the assumption of linearity in the model.

Determination of n and K of Nash’s model

From the property of the IUH the first moment of the IUH about the origin t =0 is given by
M 1 = nK (13.31)
Also the second moment of the IUH about the origin t =0 is given by
M 2 = n(n + 1)K 2 (13.32)
Using these properties the values of n and K for a catchment can be determined adequately if the Excess
Rainfall Hyetograph (ERH) and a corresponding Direct Runoff Hyetograph (DRH) are available.If
MQ1 = first moment of the DRH about the time origin divided by the total direct runoff and
MI1 = first moment of the ERH about the time origin divided by the total effective runoff,
then, MQ1 - MI1 = nK (13.33)
Further, if
MQ2 = second moment of DRH about the time origin divided by the total direct runoff and
MI2 = second moment of ERH about the time origin divided by the total excess runoff,
then, MQ2 - MI2 = n(n + 1)K 2 + 2nKM I 1 (13.34)
Knowing MI1, MI2, MQ1 and MQ2, values of K and n for a given catchment can be calculated by Eqs.
(13.30) and (13.31).

Example 13.2
For a certain catchment, the values of the Nash model coefficients are found to be; n = 4.3 and K = 3.1.
Determine the ordinates of the IUH hydrograph of the catchment.

Solution
The ordinates of IUH by Nash model are given by
1
u (t ) = (t / k )n−1 e −t / K
K Γ(n )
Given are n = 4.3, K = 3.1 and u(t) is in cm/h.
Γ(4.3) = 4.3 Γ(3.3) = 4.3 x 3.3 Γ(2.3)
= 4.3 x 3.3 x 2.3 Γ(1.3)
From Tables, Γ(1.3) = 0.897471
Hence Γ(4.3) = 4.3 x 3.3 x 2.3 x 0.897471 = 29.291
1
u (t ) = (t / 3.1)3.3 e −t / 3.1
3.1x 29.291
u (t ) = 0.01101 (t / 3.1) e − t / 3.1
3.3

vi) The Linear Channel

Reservoir models generate an immediate response to an input. For hydrologic systems, a delay is often
observed between the input and the output. To represent this phenomenon, a linear channel model can be
used to perform a simple translation between the input and the output :

q (t ) = i (t − τ ) (13.35)
where :
i(t) = the inflow rate ( m3/s );
q(t) = the outflow rate ( m3/s );
t = the translation time ( s ).

The model is often used in combination with reservoir models, to delay the input to the reservoir or to delay
the reservoir output.

vii) The HBV Model: A Model utilising Linear Reservoirs

Introduction
The HBV-model is a conceptual hydrological rainfall-runoff model which is used to simulate the runoff from
climatic input data. The model was developed by Dr.Sten Bergström at the Swedish Meteorological and
Hydrological Institute (SMHI). Though originally developed in Scandinavia, the model has been successfully
applied in several parts of the world (Bergström, 1976, 1992). It may be used as a lumped model to simulate
flow from a catchment or it may be applied to represent surface processes at each grid-cell of a distributed
model.

The structure of the model is based on a simplified mathematical description of the natural hydrological
elements and processes. The model dynamically computes the water balance for the main storage types
(sub models or routines) in the catchment and shows how these storages change dynamically in response
to the varying precipitation, temperature and evaporation (Killingtveit et al, 1995).

The general water balance can be described as:

(13.36)

Where:
P = precipitation, E = evapotranspiration, Q = runoff, SP = snow pack, SM = soil moisture,
UZ = upper groundwater zone, LZ =lower groundwater zone and Lakes = lake volume

The input data are daily or shorter observations of precipitation, air temperature and estimates of normally
monthly potential evapotranspiration (PE) values. Air temperature data is used for calculations of snow
accumulation and melt and to adjust or calculate potential evaporation. Temperature can be left out in
snow free areas if PE is not to be adjusted or calculated.

Model Structure and Implementation Issues

The HBV structure is a sequence of sub models. These are the snow sub model, the soil moisture zone,
the dynamic part, which employs a series of linear tanks and routing as shown in Fig 13.18.that
demonstates the logic. Detail descriptions of the model are made by several authors (Killingtveit et al,
1995; Bergström, 1992) so they are not repeated here. The advantage with the model is that it can easily
be implemented in an excel workbook.
 Fig 13.18 The HBV Model Structure

As mentioned above, precipitation, temperature and evaporation data are required together with observed
daily-time series of runoff. Meteorological data needs to be pre-processed so that it can be applied over a
whole catchment. This is because the data are usually recorded at point locations and there is normally
variability in the catchment. For example there is a normally a precipitation lapse rate as one moves to
higher elevations in the catchment. Thus there is need to convert a single precipitation reading to aerial
precipitation by taking this into consideration. In other cases, there may be several stations and different
methods e.g. Thiessen polygons and spatial interpolation by GIS (inverse distance weighting, or kriging)
may be used.

The model contains a number of free parameters which must be determined by a process of calibration.
These parameters define the characteristics of the catchment that cannot easily be measured. Calibration
is done by trying different combinations of the free parameters (within acceptable limits) with the
objective of fitting the simulated to the observed runoff as best as possible. There are several objective
methods that can be used to carry out the simulation the most popular being the R-squared (see equation).
Normal values of RSQ are 0.6 to 0.9.

R2 =
∑ (Qobserved − Qobserved _ mean )2 − ∑ (Qsimulated − Qobserved )2
∑ (Q observed − Qobserved _ mean ) 2
(13.25)

The free parameters may be calibrated by trial and error. It can be hard to find an optimal set of
parameters by trial and error due to the large number of possible combinations and therefore automatic
calibration methods (optimization algorithms) are may be applied.

Other parameters, such as catchment area, elevation and percentage of area covered by open water or
lakes, are called confined parameters. These may be determined from maps, surveys or other databases.
They usually represent more easily measurable characteristics of a catchment and so they are not decided
by calibration.

It may be important during the calibration to also check the water balance for a period. For instance, how
much water in total has the model simulated for a year compared to how much was actually observed.
This can be important if the fit is not more important than determining the amount of water say for a
hydropower system. There may also be cases where one is more interested in simulation of particular
flow conditions and so a good fit for the low flows is strongly desired. These approaches are called
subjective methods for calibration and are usually based on observation. Figure 13.19 shows a typical plot
of observed and simulated flows by the HBV model for a tropical catchment.
m3/s
25 70
Observed
Simulated 60
20 Cumulative Obs
Cumulative Sim 50

15
40

30
10

20

10

0 0
01/08/1965 01/09/1965 01/10/1965 01/11/1965 01/12/1965 01/01/1966 01/02/1966 01/03/1966 01/04/1966 01/05/1966 01/06/1966 01/07/1966

Fig 13.19 Observed and HBV simulated flows for a tropical catchment

i. In this example, the low flows for this year, are fairly well simulated but the model fails to
simulate the observed peaks. This could be due to a number of reasons. Input data errors. A
 sensitivity analysis reveals that given the high evaporation rates in this catchment, small changes
in rainfall data or evaporation data lead to large changes in flow.
ii. Catchment characteristics: The assumption of saturating a soil before you get any runoff may not
work during very high intensity rains usually experienced in the part of the tropics where this
catchment is located.
iii. The model does not represent all the processes in nature to their real level of complexity.
Parametization implies up-scaling of processes that actually happen at much smaller scales in the
catchment by representing them with a single large scale parameter. The processes may have
different relationships at these different scales but the models do not take care of this.

Whereas more physically based models can improve the way models represent complex processes, they
however tend to require more data which may not be readily available, and they will have even more
parameters and require more computing resources.

The HBV model and many other similar models of low complexity have proven reliable to solve some
hydrological modelling problems and can be deployed quite quickly. They are also coupled to some of the
distributed models. The version of HBV described here is the simplest. The present model version called
HBV-96 (Lindström et al, 1997) led to slight changes in the process descriptions for snow accumulation
and melt, evapotranspiration, groundwater discharge and automatic calibration. This led to significant
improvements in model performance with average value of the efficiency criterion R2 increasing from 86
to 89%. The improvements in model performance were more due to the changes in the processing of input
data and the new calibration routine than due to the changes in the process descriptions of the model.

13.5 Ungauged Catchments

In ungauged catchments no runoff data are available for calibrating model parameter, so alternative
methods are required. There exist several methods that depend on the type of parameter, whether they are
calibration parameters associated with one particular model or physically based parameters based with
some meaning beyond the model used. Calibration parameters are preferably transposed from similar
gauged catchments in the region. Estimating calibration parameters from catchment attributes is generally
not recommended for the following reasons: i) the measureable catchment attributes may not be relevant
for catchment response, ii) there may be significant uncertainty in the calibrated parameter values, which
may cloud the underlying relationships between calibrated model parameter and catchment attributes: iii)
the structure of the model relating catchment attributes and model parameters may not be suitable.
Physically based parameters can be inferred from measurements often based on widely available
surrogates such as land use and soil type where there are significant uncertainties and comparisons with
adjusted parametes in the same region, which are extremely useful. Alternative hydrologic response data
other than runoff may be available and assist in calibrating model parameters. Significant uncertainties
remain, however (Bl schl, 2005).

13.6 Recent Studies in Uganda

The WASMOD (Water and Snow balancing MODeling system) is a conceptual lumped modeling system
for simulating streamflow from both snowmelting and rainfall and can be operated at different timescales.
The inputs are rainfall, temperature, discharge and potential evapotranspiration (Xu, 2002).
In a study (Kizza, 2008) on catchment flows into Lake Victoria, Monte Carlo simulation was applied to
the WASMOD (Water and Snow balancing MODeling system) model using uniformly sampled
parameter sets for the River Nzoia as one of the main rivers that flow into lake, for monthly data covering
the period 1970-1995.The performance of the model was assessed by comparing simulated and observed
flows using two performance criteria. Model assessment showed good performance with Nash and
Sutcliffe coefficients of 0.85 or more. For data covering the period 1973-1989, simulations bounded
measured flows 86% of the time. The model failed for the period 1990-1995 probably because of
uncertainties in the measured flow. This shows that this methodology can be used to reveal changes in
catchment properties or data errors.
The VHM is a lumped conceptual model simulating base flow, interflow and overland flow as a function
of moisture content in each of three mutually interrelated storages; Surface Storage, Root Zone Storage
and Groundwater storage. Given rainfall and evaporation data, VHM calculates runoff time series
(Willems, 2003).
In a study to calibrate ( determine the model parameters) the VHM model on Bukora and Katonga
catchments in Uganda, rainfall data, split flow series and potential evapotranspiration data were obtained
for the area and input into the model. The results show reasonable accuracy for the total flow and base
flow simulations, but poor accuracy for the simulation of interflow and overland flow. The model gives
good simulation for the cumulative flow volumes and the peak flows, while it tends to overestimate the
low flows. This result can be attributed to inaccuracies in the data measurement and collection a limited
time series of data and infilling of incomplete data sets (Ojeo, 2006).

13.7 Elements of a Unified Theory of Hydrology

Catchment hydrology is operating under a reductionist paradigm, dominated by small scale process
theories. Yet there are many examples in hydrology of highly complex behaviour, which include
nonlinearities and thresholds and paradoxes that cannot be fully explained through small scale process
theories. There are strong interactions and feedbacks between processes, which may lead to apparent
simplicities in the overall catchment response. However, the laws governing are not well understood.
Routine measurements and field experiments have provided information for observing catchment
responses and understanding the underlying process controls, but there has been little progress in
extrapolating the local knowledge and understanding from gauged catchments to ungauged catchments.
There are many theories governing the different elements of catchment hydrology but not a unified theory
of hydrology at the catchment scale.

A unified theory of hydrology has been proposed and seeks to address the following:
a) The ways catchments are organized in space and time in terms of their constituent landscape
elements, including the geomorphic and ecological processes that may have led to them;
b) The ways catchments correspond to climatic inputs and the nature of the interactions between the
heterogeneities in the climatic inputs and the landscape properties;
c) The resulting fundamental hydrological processes, their space time variabilities including the
pathways, fluxes and stores of water energy and other constituents and the interactions between
them;
d) The way that the different constituent parts of the catchments and the catchments as a whole
function, interact with, and feedback on each other; and
e) The way catchments respond to human induced changes in the climate inputs and the landscape
properties, in terms of both their form and function (eg storage of water) in the short term and
long term (Sivapalan, 2005).

Summary
The drainage of precipitation from a catchment is referred to as runoff. It occurs when the infiltration
capacity of an area’s soil has been exceeded. The factors that determine the amount of runoff and
hydrographs are discussed together with the Rational Method and the Unit Hydrograph as two important
methods of estimating runoff in a catchment. More advanced methods using linear systems analysis are
then discussed. These include the Synthetic Unit Hydrograph, the Time Area Method, Linear Reservoir
Models, the Nash Cascade, the Linear Channel and the HBV Model. Considerations for Ungauged
catchments, recent studies in Uganda and elements of a Unified Theory are also outlined.
References
1. Ayoade, J.O.,Tropical Hydrology and Water Resources, Macmillan,1998, London, UK
2. Bergström, S., Development and Application of a Conceptual Runoff Model for Scandinavian
Catchments, SMHI Reports RHO No. 4, 1976, Norrköping, Norway.
3. Bergström, S., The HBV Model - Its Structure and Applications. SMHI Reports RH, No. 7,
1992,Norrköping, Norway
4. Bl schl, G., Rainfall- Runoff Modelling of Ungauged Catchments, Encyclopeadia of
Hydrological Sciences, Part 11, John Wiley and Sons Ltd, 2005, London UK.
5. Butler,D., J.W Davies, J.W., Urban Drainage, E& FN Spon,2000, London, UK
6. Clarke,C.O., Storage and the Unit Hydrograph, ASCE, Trans.,1945,110, pp1419-1446.
7. Chin, D. A., Water Resources Engineering, Pearson Education Inc, 2006, New Jersey, USA.
8. Chow, V.T., Maidment, D.R.., Mays,L.W., Applied Hydrology, McGraw Hill,1988, New York, USA
9. Das G. Hydrology and Soil Conservation Engineering, Prentice Hall, 2002, New Dehli, India.
10. Killingtveit, Å.,Sælthun, N.R., (1995), Hydrology, Hydropower Development Book series,
Volume 7. Norwegian Institute of Technology, Divisiion of Hydraulic Engineering, 1995,
Trondheim, Norway
11. Kirpich, P.Z., Time of Concentration of Small Agricultural Watersheds, Civil Engineering,1940,
10(6).
12. Kizza, M., Rainfall Trend Analysis and Uncertainty Related Runoff Modeling within the Lake
Victoria Basin, Licentiate Thesis, Department of Earth Sciences, University of Uppsala, 2008,
Uppsala, Sweden.
13. Lighthill, F.R.S., C.B. Whitham, 1955. On Kinematic waves, Flood movement in long Rivers.
Proc. R. Soc. London, Vol. 229 (1178):281-316.
14. Lindstrom, G. A Simple Automatic Calibration Routine for the HBV Model, Swedish
Meteorological and Hydrological Institute, Nordic Hydrology,Nordic Association for Hydrology,
1997, Lyngby, Danemark (1970) (Revue)
15. Nash,J.E., The Form of the Instantaneous Unit Hydrograph, IASH, 1957, No 45,3,pp114-121.
16. Ojeo, J. Stepwise Calibration of Hydrological Models using Data Mining Techniques, Katholike
Universitie Leuven, 2006, Belgium. Unpublished M.Sc Dissertation.
17. Shaw, E.M, Hydrology in Practice, Chapman and Hall, 1994, London, UK .
18. Sivapalan, M., Pattern , Process and Function:Elements of a Unified Theory of Hydrology at the
Catchment Scale, Encyclopeadia of Hydrological Sciences, Part 11, John Wiley and Sons Ltd,
2005, London UK
19. Soil Conservation Service, National Engineering Handbook, Section 4, Hydrology, U.S.
Department of Agriculture, 1972, Washington D.C., USA.
20. Synder,F.F., Synthetic Unit Graphs, Trans American Geophysical Union, 1938, 19, pp447-454.
21. Viessman Jr, W., Lewis, G.L., Introduction to Hydrology, 4th Edition, Harper Collins College
Publishers, 1996, New York, USA.
22. Ward, A. D., Trimble, S.W., 2004. Environmental Hydrology, CRC-Lewis Press, Boca Raton,
Florida, USA.
23. Willems ,P., Stepwise Physically Based Construction and Calibration of Lumped Conceptual
Rainfall Runoff Model, International Workshop on Catchment Scale Modeling and Data
Assimilation, 2000, Wageningin, Holland.
19. Xu,C.Y., WASMOD- The Water And Snow balance MODelling System. In: V.P. Singh and D.K.
Fevert ,Ed, Mathematical Models of Small Watershed Hydrology and Applications. Water
Resources Publications, 2002, Michigan USA.
 Further Reading

1. Arora, S.O., Water Resources Hydropower and Irrigation Engineering, Standard Publishers and
Distributors, 1996, Dehli, India.

2. Garg, S.K., Hydrology and Water Resources Engineering, Khanna Publishers, 1998, Delhi, India.
Questions

1. Distinguish between an influent and an effluent stream.

2. Explain the factors which influence the amount of runoff generated from a catchment.
3. i) Define the a) Unit Hydrograph
b) T – Unit Hydrograph
c ) S – Curve
ii) Explain the limitations of the Unit Hydrograph hypothesis
4. i) Describe the ‘Rational’ method of designing surface water drainage systems. Define in full any
terms and the assumptions made in it.

ii) What is the critical rainfall intensity?

5. i) An acceptable 1-h unit hydrograph (10 mm) has been derived for a catchment. Its ordinates are
shown in Table 1. What is the approximate area of the catchment?

ii) Determine the peak flow that would result from a storm whose effective rainfall, assumed over
the whole catchement, is given in Table 2.
Table 1
________________________________________________________________
Time (h) 0 1 2 3 4 5 6 7
U(t) (m3s-1) 0 5 33 17 12 11 7 0
________________________________________________________________
Table 2
________________________________________________________________
Time (h) 0 1 2 3
Total accumulated effective rainfall (mm) 0 10 32 40
________________________________________________________________

iii) The ordinate of the 1 – h unit hydrograph of a catchment area are summarized in Table 3.

Table 3
________________________________________________________________
Time (h) 0 1 2 3 4 5 6 7
TUH ordinate
(m3s-1 mm-1) 0 30 70 95 146 264 320 386

Time (h) 8 9 10 11 12 13 14 15
TUH ordinate
(m3s-1 mm-1) 275 159 111 78 66 45 15 0
________________________________________________________________
 (a) Derive the S-curve for the catchment area.

(b) Use the S-curve to obtain the 2-h unit hydrograph; and

(c) Forecast the peak runoff that would result from a storm in which the effective rainfall
totals in two consecutive 2-h periods were 20mm and 5 mm.

6. Explain the following terms; i) the impulse response function, ii) the step pulse function iii) the pulse
response function,iv) a linear reservoir and v) a linear channel as applied to linear hydrologic systems.

7. Distinguish between rainfall excess and effective rainfall.

8. Define the terms: i) Time of concentration, ii) Runoff coefficient