Machiwal 2015

Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

Arab J Geosci

DOI 10.1007/s12517-015-2002-5

ORIGINAL PAPER

Comparing GIS-based multi-criteria decision-making


and Boolean logic modelling approaches for delineating
groundwater recharge zones
Deepesh Machiwal 1,2 & P. K. Singh 1

Received: 29 December 2014 / Accepted: 15 June 2015


# Saudi Society for Geosciences 2015

Abstract Artificial groundwater recharge is a corrective mea- the most influencing variables. These findings may be useful
sure for restoring the fast-depleting groundwater resources. for decision-makers to formulate appropriate groundwater
This study, for the first time, employed and compared both management strategies, and the approach may be well-opted
the multi-criteria decision-making (MCDM) and Boolean log- in other hard-rock regions of the world especially in econom-
ic modelling (BLM) approaches to delineate groundwater re- ically poor nations.
charge zones in a hard-rock catchment of Udaipur, India.
Thematic maps of geomorphology, soil, land use, ground ele- Keywords Groundwater recharge zones . Artificial recharge
vation, percent slope, slope length, slope steepness, drainage sites . Multi-criteria decision-making . Boolean logic
density, transmissivity and groundwater fluctuation were pre- modelling . Remote sensing . Geographical information
pared by using remote sensing and geographical information system
system. Suitable relative weights to variables and their classes
were assigned depending upon their influence on groundwater
recharge. The weights were normalized by MCDM technique Introduction
to remove subjectivity. Thematic maps were integrated to de-
lineate recharge zones. In BLM approach, decision rules were Groundwater, the largest source of freshwater in semi-arid and
developed and AND operator was used to integrate the the- arid regions worldwide, is of utmost importance for hilly ter-
matic maps to delineate recharge zones. It was found that rains where hard-rock aquifers exist. The groundwater pro-
36.07 km2 (10 % of total area) and 30.55 km2 (9 % of total ductivity potential in such hard-rock aquifers of semi-arid re-
area) were rendered as suitable zone by the MCDM and BLM gions remains constrained within shallow fractured and
approaches, respectively. Two approaches were successfully weathered formation (Machiwal and Jha 2014). In these
validated based on net recharge estimates of 50 sites, and hard-rock terrains, all kinds of domestic, agricultural and in-
results were found in agreement and comparable to each other dustrial water demands are largely met by groundwater sup-
in 90 % area. Moreover, 29 favourable artificial recharge sites plies because of the effects of mainly climate change and
were identified in suitable recharge zone based on drainage population growth among other factors, which cause severe
map. Finally, the recharge sites were grouped into four clusters stress to available surface water supplies in these areas
by using multivariate statistical analyses, which characterized (Edmunds et al. 2003).
In India, it is reported that the shallow hard-rock aquifers
represent around 73 % (2.386 million km2) of the country’s
* Deepesh Machiwal total land area (www.portal.gsi.gov.in). In such hard-rock for-
dmachiwal@rediffmail.com mations especially those situated in semi-arid and arid regions,
overexploitation of the vital groundwater resources along with
1
SWE Department, College of Technology and Engineering, MPUAT, other hazardous climatic factors such as scanty rainfall, severe
Udaipur 313 001, Rajasthan, India drought occurrences and high air temperatures has caused
2
Present address: ICAR - Central Arid Zone Research Institute, lowering of groundwater levels during the past few years
Regional Research Station, Bhuj 370 105, Gujarat, India (CGWB 2011). Accordingly, the groundwater levels are
Arab J Geosci

found to be fast depleting in the hard-rock catchment of Ahar MCDM technique, and based on BLM approach, (ii) to com-
River, Udaipur, India (Machiwal et al. 2012), which put pare and validate the recharge zones by two approaches; (iii)
groundwater resources of the area at a great risk. The declining to identify favourable artificial recharge sites and (iv) to char-
groundwater levels can only be restored by adopting correc- acterize the dominating variables influencing the recharge by
tive measures of artificial groundwater recharge. Presently, employing HCA and PCA techniques. This study, for the first
there are no systematic guidelines to locate suitable zones or time, employed and compared both the MCDM and BLM
sites for implementing artificial groundwater recharge tech- approaches for delineating the groundwater recharge zones
niques. Also, no systematic artificial recharge studies have and also employed multivariate statistical techniques to char-
ever been conducted in the area. acterize the most influential variables of the groundwater
The modern technologies like remote sensing (RS) and recharge.
geographical information system (GIS) have proved to be
the most successful in groundwater potential zoning
(Krishnamurthy et al. 1996; Lachassagne et al. 2001; Study area description
Solomon and Quiel 2006; Madrucci et al. 2008; Jha et al.
2010; Machiwal et al. 2011a, 2015; Deepika et al. 2013; Location and surface water resources
Awawdeh et al. 2014; Çelik 2014; Nampak et al. 2014).
However, integration of the RS and GIS techniques is com- The Ahar River catchment (study area), located between 73°
paratively new in artificial recharge studies (Saraf et al. 2004; 36′ 51″ and 73° 49′ 46″ E longitude and 24° 28′ 49″ and 24°
Jasrotia et al. 2007; Yeh et al. 2009; Chowdhury et al. 2010; 42′ 56″ N latitude, is situated in Udaipur district of Rajasthan,
Singh et al. 2013; Kaliraj et al. 2014; Mahmoud et al. 2014). India (Fig. 1). Climate of the study area is semi-arid and is
Extensive literature search clearly shows that, except for few characterized by sub-tropical and sub-humid region. The area,
of the studies conducted in Iran (Ghayoumian et al. 2005; encompassing about 348 km2, partly covers Girwa and
Dashtpagerdi et al. 2013; Rahimi et al. 2014), Jordan Badgaon blocks of Udaipur and is surrounded by almost a
(Hammouri et al. 2014), Lebanon (Shaban et al. 2006) and girdle of hills. The average annual rainfall in the study catch-
Saudi Arabia (Mahmoud 2014; Mahmoud et al. 2014), all the ment is 60.9 cm, about 90 % of which is received from the
studies have been conducted in India only. Boolean logic southwest monsoon. Surface water resources of the area are
modelling (BLM) is another useful approach for developing mostly available in the rivers and lakes. Ahar is the major river
the simplest and best-known logic-based GIS models draining the entire catchment from the northwest towards
(Robinov 1989). The BLM involves logical combination of southeast. The lakes with significant water storage capacities
different binary maps resulting from the application of condi- that exist in the area are Pichhola, Fatehsagar and Udaisagar;
tional operators (Bonham-Carter 1994). The application of the all lakes are artificial and their storage capacity is filled up by
BLM approach in artificial recharge studies is very few (e.g. the runoff generated in their catchments. All the rivers flow
Ghayoumian et al. 2007; Dashtpagerdi et al. 2013). Further, seasonally in the area, and all the lakes frequently dry up
many researchers in the recent past have found that multi- during the scanty rainfall years. Hence, there is dearth of sur-
criteria decision-making (MCDM) is a vital tool for water face water throughout a year.
resources management, which adds structure, transparency,
auditability and rigour to the judgements (e.g. Flug et al. Geology, hydrogeology and groundwater conditions
2000; Joubert et al. 2003; Hajkowicz and Higgins 2008).
One of the widely used MCDM techniques in the field of Geology of the area comprises of Aravalli and post-
water resources is Saaty’s Analytic Hierarchy Process Aravalli systems including gneiss, schist and phyllite-
(AHP), which have sought only few applications in ground- schist rock formations and their combinations (Machiwal
water studies during recent years (Chowdhury et al. 2009; Jha et al. 2011a). Gneiss, grey to dark-coloured and medium
et al. 2010; Machiwal et al. 2011a). The application of the to coarse-grained rocks, comprises porphyritic gneissic
AHP-MCDM approach in artificial recharge studies is highly complex associated with aplite, amphibolite, schist and
limited (Chowdhury et al. 2010; Mahmoud et al. 2014; augen gneiss. Schist rocks are compact, hard and fine to
Rahimi et al. 2014). Moreover, none of the past artificial re- medium grained and are characterized by alternating
charge studies utilized multivariate statistical techniques such bands of light and dark colour ferromagnesiam minerals.
as hierarchical cluster analysis (HCA) or principal component Phyllite-schist litho units represent argillaceous sediments
analysis (PCA), which is very useful for groundwater studies and grades from shale, slate, phyllite to mica schist, which
(Machiwal et al. 2011a). are soft and friable. Depth to groundwater levels remains
The objectives of this study were (i) to delineate ground- shallow within the upper weathered hard-rock strata/
water recharge zones in Ahar catchment of Udaipur, aquifer c haracteri zed by unconfined conditions
Rajasthan, India, using integrated RS, GIS and AHP- (Machiwal et al. 2011b). The average groundwater depth
Arab J Geosci

Udaipur District

Study Area

Ahar River Catchment

Fig. 1 Location map of study area showing location of recharge estimation sites

varies from 2 to 23 m below ground surface (bgs) during Methodology


pre-monsoon season, which reaches up to average depth
of 2–14 m bgs in post-monsoon season. Among the Database preparation
existing groundwater extraction means, dug wells account
for 68.52 %, handpumps for 29.35 %, tubewells for In this study, spatial database of ten hydrologic/
1.62 % and stepwells for 0.51 % (Singh 2002). hydrogeologic parameters was prepared in GIS; data
Arab J Geosci

Data were recorded by means of TLC (temperature level conductivity)


description is summarized in Table 1. Boundary of the

Shuttle Radar Topographic Mission (SRTM) data downloaded from


study area was demarcated in GIS by using the

National Bureau of Soil Survey & Land Use Planning, Regional


Integrated Land and Water Information System (ILWIS

Pumping tests were conducted and test data were analysed


2001) software. A coordinate system was generated
based on the Universal Transverse Mercator as projec-

Indian Institute of Remote Sensing, Dehradun, India


tion system with Ellipsoid as Everest India 1956 and

Research Station, Udaipur (Jain et al. 2005)


Datum as Indian (India Nepal).
False colour composite (Fig. 1), generated from the IRS-P6

Survey of India Office, Dehradun, India


satellite image, was utilized along with toposheets to extract
geomorphology map. Soil map was prepared by digitizing and

Metre made by Solinst, Canada


classifying the collected hardcopy map, and land use map was
prepared by adopting supervised classification of the satellite
image (Machiwal et al. 2010).
The ground elevation map was computed by using
digital elevation model (DEM) with no ‘sinks’. The per-

(USGS 2004)
cent slope map was generated by applying two differen-
tial gradient filters (shown in Fig. 2) on the DEM in x-

Source
and y-directions and then using the following equation
(ILWIS 2001):

(May 2006–July 2009)


qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
D2x þD2y
Slopeð%Þ ¼ 100  ð1Þ

February 2000

2006–2008
February 8, 2004

36-month period
Pixel size ðDEM Þ

Period
Where, Dx and Dy = DEM with x- and y-gradient filters,
respectively, and Pixel size (DEM) = pixel size of the DEM.
The slope length was estimated using the following rela-

1:250,000
1:50,000
1:50,000

Monthly
tionship (McCool et al. 1987):
Scale
Details of data collected and used for delineating recharge zones in this study

 m


λ
L¼ ð2Þ
Linear Imaging Self-Scanning (LISS) – III

22:13
45H9, 45H10, 45H11, 45H14 and 45H15
IRS-P6 (Indian Remote Sensing Satellite)

Where, L = slope length (unit less); λ = field slope length Transmissivity and specific yield
Pixel resolution of 3 arc-second

(m) and m = dimensionless exponent being 0.5, 0.4 and 0.3 for
slopes >5, 4 and <3 %, respectively.
(or 90-m pixel size)

The slope steepness was derived for slope length longer


than 4 m using the following equations (McCool et al. 1987).
data at 18 sites
50 sites’ data
Specification

Soil texture

S ¼ 10:8sinθ þ 0:03; slope < 9% ð3Þ

S ¼ 16:8sinθ−0:05; slope≥ 9% ð4Þ


Digital elevation model

Groundwater levels

Aquifer parameters
Geodetic toposheet
Satellite imagery

Where, S = slope steepness (unit less) and θ = slope angle


Type of data

(degree).
Soil map

Drainage network map was prepared by simulating and


digitizing drainage lines from the DEM and then checked
from the satellite imagery and toposheets. Thereafter, sub-
Table 1

catchments were delineated by adopting the methodology of


S. no.

creating stream network available in the ILWIS software. The


1
2

6
Arab J Geosci

a dfdx filter b dfdy filter ranging from 0 to 9 for ten thematic variables and their fea-
tures/classes. The maximum weights were assigned to the
1 -8 0 8 -1 × 1/12 -1 themes/features of highest potential for groundwater recharge
and the minimum weights to the lowest potential recharge
8
themes/features. This approach has been strongly recom-
0 × 1/12 mended for studies on groundwater potential zoning using
RS and GIS (Saaty 1980). The weights were then finalized
-8 based on experts’ opinion, personal knowledge and experi-
1 ence about the study area. Furthermore, the weights were nor-
malized by AHP-MCDM technique (Saaty 1980), which uses
Fig. 2 Differential filters a dfdx and b dfdy to determine gradients in x-
eigenvector method for removing subjectivity of the weights.
and y-directions, respectively Moreover, consistency of the normalized weights was
checked by computing consistency ratio for the thematic var-
iables and their classes. If value of the consistency ratio re-
drainage density was calculated by using the following ex- mains less than or equal to 10 %, the assigned weights are
pression (Haan 2002): considered as consistent (Saaty 1980). Otherwise, it is sug-
gested that the assigned weights should have been revised to
X
n
avoid inconsistency. The consistency ratio (Cratio) was com-
ðLi Þ
i¼1 puted as (Saaty 1980):
Dd ¼ ð5Þ
A C index
−2
C ratio ¼ ð6Þ
Where, Dd = drainage density (km km ), Li = length (km) C random
of ith drainage line present within A area, A = area of the sub-
Where, Crandom = random consistency index, whose values
catchment (km2) and n = total number of streams present
were obtained from the standard table provided in Saaty
within the sub-catchment.
(1980), and Cindex = consistency index.
Spatially distributed groundwater level maps for pre- and
The consistency index was calculated by using following
post-monsoon seasons of 2006–2009 period were developed
expression (Saaty 1980):
by employing the kriging technique. Experimental semi-
variograms of the groundwater levels based on 50 sites’ data λmax −n
(Fig. 1) were fitted with four theoretical semi-variograms, i.e. C index ¼ ð7Þ
n−1
spherical, circular, exponential and Gaussian. Nugget, sill and
range parameters were chosen by curve-matching technique. Where, λmax = principal eigenvalue computed by eigenvec-
Exponential semi-variogram was selected as the best fit for tor method and n = number of thematic variables.
developing the groundwater level maps. The groundwater
fluctuation for dry season (November in particular year to
May next year) was computed by subtracting the groundwater Delineation of artificial recharge zones by MCDM
levels of post-monsoon season of a year from that of pre- approach
monsoon season of the next year and the average was com-
puted. Specific yield and transmissivity were estimated from The groundwater recharge zones were delineated by adopting
analysis of pumping tests’ data recorded at 18 sites (shown in weighted linear combination method (Malczewski 1999),
Fig. 3j), and spatially distributed maps were interpolated by where groundwater recharge index (GWRIMCDM) was com-
inverse distance weighting method. Depending upon the puted by integrating weights of the thematic variables normal-
availability of the data, groundwater fluctuation method was ized by the AHP-MCDM approach multiplied by respective
employed to estimate the mean annual net recharge. normalized weights of different classes as shown by following
expression:
Weight assignment and normalization using MCDM n X
X m  
approach GWRIMCDM ¼ wi  w j ð8Þ
i¼1 j¼1

The MCDM approach considers the relative importance of


one hydrologic/hydrogeologic variable over another in artifi- Where, wi = normalized weight of ith thematic variable,
cial groundwater recharge. Few international and local experts wj = normalized weight of jth class of the variable, n = total
of hydrogeology were asked to know their opinions for number of variables and m = total number of classes of a
selecting the appropriate relative weights on Saaty’s scale variable.
Arab J Geosci

Evaluating performance of MCDM approach Fig. 3 Maps showing spatial distribution of a geomorphology, b soil„
texture, c land use, d ground elevation, e percent slope, f slope length, g
slope steepness, h drainage density, i transmissivity and j groundwater
The developed groundwater recharge zone map was verified fluctuation in the study area
from the point estimates of the net recharge. The GWRIMCDM
estimates were obtained from the recharge zone map for 13 of
the total 50 sites falling in suitable artificial recharge zone. The where first-, second- or third-order streams intersected with
groundwater recharge zone map was verified by both graphi- each other in the suitable recharge zone. In this way, feasible
cally plotting net recharge versus GWRIMCDM on scatter plot sites for artificial recharge were selected.
and fitting linear regression model and by statistically com-
puting coefficient of determination (R2) for the fitted model. Characterizing recharge sites according to most
influencing variables
Development of groundwater recharge index by Boolean
logic modelling The identified artificial recharge sites were grouped into dif-
ferent clusters such that the thematic variables at sites within a
In the BLM approach, evidential layers from ten thematic cluster are similar to each other with respect to their properties
maps were prepared based on decisions rules to support the but different from those in other classes. The clusters of the
hypothesis of suitability for artificial recharge. Each pixel of sites were formed by the HCA, which is a proven unsuper-
evidential layers was assigned one (or zero) value, depending vised pattern recognition technique that uncovers intrinsic
on whether it satisfies (or not) the decision rule. The decision structure or underlying behaviour of a dataset without making
rules, which are a set of logical conditions, were derived based a priori assumption about data, in order to classify the objects
on the knowledge gained from textbooks (e.g. Freeze and of the system into clusters based on their similarities (Otto
Cherry 1979; Karanth 1987), information/suggestions avail- 1998). Such a clustering of recharge sites helps understanding
able in literature (e.g. Ravi Shankar and Mohan 2005; the relative influence of the different variables on the ground-
Chowdhury et al. 2010) and personal experience learnt during water recharge. The most influencing variables in different
field visits. The BLM approach consists of AND/OR opera- clusters of recharge sites were then characterized by the
tors, which works among two or more variables/datasets. PCA, which is used to reduce multi-dimensional datasets to
According to set theory, the AND operator results in the log- lower dimensions for analysis (Dillon and Goldstein 1984).
ical intersection of two variables, whereas the OR operator
calculates the logical union of them. In this study, the AND
operator was employed to find suitable recharge zone Results and discussion
extracting those pixels which satisfies all the decision rules.
Thematic maps of variables
Verification of the delineated groundwater recharge zones
Thematic maps of ten hydrologic/hydrogeologic variables
The groundwater recharge zones of the BLM approach were prepared in this study using RS and GIS are discussed ahead.
verified by superimposing point map of the mean annual net
recharge values estimated at 50 sites. The point values of the Geomorphology
net recharge falling into suitable and unsuitable recharge
zones were compared. The accuracy of the groundwater re- The study area was classified into six geomorphologic classes
charge zones was further validated by comparing them with (Fig. 3a) namely deep buried pediment, inselburg, residual
suitable and unsuitable zones delineated by the MCDM hill, shallow buried pediment, structural hill and waterbody.
approach. A pediment is a gently inclined slope of transportation and/or
erosion that truncates rock and connects eroding slopes or
Locating suitable sites for artificial groundwater recharge scarps to the areas of sediment deposition at lower levels
(Oberlander 1989). If the pediments are covered by alluvial
The literature review reveals that researchers have considered or weathered materials, they are termed ‘buried pediments’.
different hydrological and/or geological factors in selecting The buried pediments are considered as the most favourable
suitable artificial recharge sites (Ramaswamy and location for artificial recharge due to covered alluvium and
Anbazhagan 1997; Saraf and Choudhury 1998; Ravi weathered substances underneath. The deep buried pediments
Shankar and Mohan 2005; Agarwal et al. 2013; Rolland and were mainly present nearby rivers covering 59.69 km2 in
Rangarajan 2013). In this study, drainage map showing stream southeast, northwest and western portions (Fig. 3a). An iso-
orders was superimposed on the groundwater recharge zone lated hill of massive type abruptly rising above surrounding
map of the BLM approach. The recharge sites were chosen plains is called inselburg. It forms runoff zones and barriers for
Arab J Geosci

a b

c d

e f

g h

i j
Arab J Geosci

groundwater movement. It has less significant recharge poten- restricting water entries into the ground, whereas waterbodies
tial and prospects than that for deep buried pediment. already acted as natural recharge source.
Inselburgs were present in a small portion (1.01 km2) at iso-
lated places, which provides barrier for groundwater move- Ground elevation
ment forming adequate runoff zones. Residual hills also
encompassed small area of 6.2 km2 and existed in scattered The ground elevation map was classified into five classes
patches. The largest part of the area, i.e. 56 % (187.58 km2), (Fig. 3d). Surface near boundaries had relatively higher eleva-
was covered by the shallow buried pediment (Table 2). tion (>700 m MSL) whereas the ground looked like a depres-
Structural hills having the least opportunity for artificial re- sion (ground elevation <550 m MSL) in the southeast portion.
charge were dominating in 80.16 km2 mainly near the catch- Ground surface in the major portion (154.37 km2, 44 %) was
ment boundaries. The surface waterbodies with adequate stor- elevated up to 600–700 m MSL (Table 2). The higher ground
age of the surface water resources for few months in year may elevations near boundaries induced large quantities of the run-
have opportunities for ample quantities to get recharged off, which had lesser chances to get infiltrated down the sur-
through beds of the waterbodies. The surface waterbodies, face. However, the entire generated runoff finally reached to
spreading over 13.67 km2 with large storage capacities, may less elevated southeast portion.
be considered as good source of natural recharge.
Percent slope
Soil
Percent slope map was classified into six classes (Fig. 3e). It is
apparent that the percent slope was very steep (more than
The soil map classified into six textural classes is
15 % in 78.15 km2 or 22 % area, Table 2) near boundaries
shown in Fig. 3b. The soil in a major portion was
and around residual and structural hills. These sharp slopes
represented by fine loam (194.33 km2, 56 % area) and
generated large runoff amounts, which quickly flowed out of
coarse loamy sand (75.69 km2, 22 % area) (Table 2).
the area, and hence, there was lesser scope for the surface
Three soil textural classes, coarse loamy sand followed
water to get recharged. The major portion (164.21 km2,
by coarse to fine loam and fine loam, contained rela-
47 % area, Table 2) belonged to the almost flat lands to little
tively large number and size of macro-pores and thus
slopes (<4 %) where rainwater had the better chances to reach
permitted more opportunity for rainwater to get infiltrat-
up to the aquifer.
ed compared to that by other classes. This may ulti-
mately lead to favourable conditions for the artificial
Slope length
recharge. All soil classes of skeletal and rocky outcrops,
encompassing 69.48 km2 (20 % area) may have less
Three classes of slope length, i.e. 0.736, 0.783 and 0.832 were
possibility that surface water get enters into subsurface
obtained (Fig. 3f) with respect to slope steepness exceeding 5,
formation and hence may present comparatively poor
4 and <3 %, respectively. The dominating slope length value
recharge zone.
was 0.736, covering 153.04 km2 (44 % area, Table 2) nearby
boundaries where the steep slopes existed. Thus, it is clear that
Land use lesser the slope length, the lesser were the chances for the
surface water to get recharged. The most favourable slope
Five types of land use classes were identified (Fig. 3c). The length for artificial recharge was 0.832, which was present
major land use was rangeland spreading over 170.36 km2 in 119.18 km2 (34 % area) where surface waterbodies were
(49 % area) and that existed in-between forest and agricultural located.
lands. Forest lands were restricted in 80 km2 (23 % area)
mostly near boundaries where structural hills were present. Slope steepness
Agricultural lands, largely manipulated through tillage,
ploughing and irrigation, covered 68.79 km2 (20 % area) Slope steepness map was classified into four classes as
mainly close to the river courses. Thus, these had good scope shown in Fig. 3g. The slope steepness was relatively
for artificial recharge. The built-up lands (23.04 km2, 7 % low <0.50) in 50 % area (175.18 km2), mostly in the
area) and waterbodies encompassed 6.15 km2 (2 %), and these central portion. This portion had the most favourable
were of relatively less significance for the artificial recharge. conditions for groundwater recharge. Relatively high
The built-up lands presented almost impermeable surfaces values of the slope steepness (>4) could be seen in
Arab J Geosci

Table 2 Areal extent, assigned and normalized weights for different classes of ten thematic variables

Theme Feature Recharge Area Assigned Normalized Consistency


condition (km2) weight weight ratio (%)

Geomorphology Deep buried pediment Very good 59.69 7 0.36 1.41


Inselburg Good 1.01 3.5 0.18
Residual hill Moderate 6.20 2 0.10
Shallow buried pediment Poor 187.58 5 0.26
Structural hill Poor 80.16 1 0.05
Surface waterbody Very poor 13.67 1 0.05
Soil Coarse loamy sand Very good 75.69 8 0.30 1.05
Coarse to fine loam Good 8.84 7 0.26
Fine loam Moderate 194.33 6 0.22
Fine loamy rock outcrop Poor 19.20 3 0.11
Loamy skeletal rock outcrop Poor 9.25 2 0.07
Skeletal fine loam Very poor 41.03 1 0.04
Land use/land cover Agricultural land Very poor 68.79 7.5 0.37 0.31
Built-up land Poor 23.04 1 0.05
Forest land Moderate 80.00 5 0.24
Rangeland Good 170.36 6 0.29
Waterbody Very good 6.15 1 0.05
Ground elevation (m MSL) <550 Very good 15.78 7 0.31 0.35
550–575 Good 59.59 6.5 0.29
575–600 Moderate 82.16 5 0.22
600–700 Poor 154.37 3 0.13
>700 Very poor 36.43 1 0.04
Percent slope 0–1 Very good 24.53 7 0.27 0.54
1–4 Very good 139.68 6 0.23
4–6 Good 51.35 5 0.19
6–15 Moderate 54.63 4 0.15
15–30 Poor 38.73 3 0.12
>30 Very poor 39.41 1 0.04
Slope length 0.736 Poor 153.04 2 0.17 0.83
0.783 Moderate 76.12 4 0.33
0.832 Good 119.18 6 0.50
Slope steepness 0–0.25 Good 56.90 6 0.40 1.54
0.25–0.50 Moderate 118.29 4.5 0.30
0.5–4.0 Poor 124.45 3 0.20
>4.0 Very poor 48.61 1.5 0.10
Drainage density (km km−2) 0–0.8 Good 71.04 5 0.50 1.86
0.8–1.2 Moderate 149.66 3.5 0.35
>1.2 Poor 127.64 1.5 0.15
Groundwater fluctuation (m) 0–5 Very Poor 69.84 1.5 0.09 1.94
5–8 Poor 171.25 3 0.19
8–11 Moderate 86.86 4.5 0.28
11–23 Good 20.39 7 0.44
Transmissivity (m2 day−1) 70–150 Poor 1.20 2 0.08 2.43
150–300 Moderate 206.33 3 0.12
300–450 Moderate 71.41 5 0.20
450–600 Good 55.52 7.5 0.29
600–2250 Very good 13.87 8 0.31
Arab J Geosci

48.61 km2 (14 % area, Table 2), the portion which was fluctuation may be quick replenishment of the depleted
close to boundaries where conditions for groundwater re- groundwater from river water recharge. Thus, this area was
charge were unfavourable. the most suitable for artificial recharge.

Drainage density Normalized weights for thematic variables


and their classes
Drainage density was classified into three classes (Fig. 3h). It
was relatively high (>1.2 km km−2) in 127.64 km2 (37 % The assigned and normalized weights of ten variables along
area), which indicated easy runoff removal and thus lesser with their consistency ratios are shown in Table 3 where nor-
scope for the recharge. However, relatively moderate (0.8– malization is also explained. A similar procedure was adopted
1.2 km km−2 in 149.66 km2, 43 % area, Table 2) and low for normalizing the assigned weights of different classes of the
density (<0.8 km km−2 in 71.04 km2, 20 % area) values sug- variables, and their weights and consistency ratios are present-
gested favourable conditions for the recharge. ed in Table 2. All the final normalized weights were found to
be consistent based on the consistency ratio whose value was
Transmissivity obtained less than 10 % in all the cases. Thus, the weights
selected for the thematic variables and their classes were free
Spatially distributed transmissivity values were classified into from subjectivity.
five classes ranging from 70 to more than 600 m2 day−1
(Fig. 3i). The shallow aquifer contained relatively large trans- Groundwater recharge zones by MCDM approach
missivity values (>450 m2 day−1) in 69.4 km2 (20 % area,
Table 2) in the northern and northwest portions, whereas the Groundwater recharge zone map resulting from the MCDM,
transmissivity was relatively low (150–300 m2 day−1) in and classified into suitable and unsuitable classes, is shown in
206.33 km2 (59 % area) in the southern, southwest and south- Fig. 4a. Classification threshold value for delineating the suit-
east portions. able recharge zone was chosen as 0.28 by few trials such that
the suitable and unsuitable zones encompass almost equal
Groundwater fluctuation areas by the MCDM and BLM approaches. It is apparent from
Fig. 4a that the suitable and unsuitable zones of the artificial
The kriged mean groundwater fluctuation map classified into recharge covered 36.07 km2 (10 %) and 312.13 km2 (90 %),
four classes is shown in Fig. 3j. It was depicted that the vast respectively. Figure 4a reveals that the suitable recharge zone
groundwater fluctuation (8–23 m) occurred in 107.25 km2 was mainly located in the north-central and southeast portions.
(31 %) area nearby boundaries where steeply sloping structur- Proximity of the suitable zone may be explained by the pres-
al hills with low slope length were present. On the other side, ence of river courses and highly transmissive aquifer in the
groundwater in the southeast and central portions (69.84 km2, northern portion, whereas in the southeast portion, it was
20 % area, Table 2) fluctuated with relatively less magnitude favoured by existence of the lesser ground elevation, little
(0–5 m). These areas mainly coincide with the lower reaches percent slope, less slope length, agricultural land and deep
of the Ahar River, and thus, possible causes for the less buried pediment.

Table 3 Assigned relative weights to different thematic layers on Saaty’s scale

Theme Assigned weight Normalization process on themes Normalized Consistency


weight ratio (%)
GM SO LU GE PS SL SS DD TR GF

GM 7 7/7 7/5 7/5.5 7/3.5 7/4.5 7/4 7/4 7/6 7/6.5 7/3.5 0.14 0.02
SO 5 5/7 5/5 5/5.5 5/3.5 5/4.5 5/4 5/4 5/6 5/6.5 5/3.5 0.10
LU 5.5 5.5/7 5.5/5 5.5/5.5 5.5/3.5 5.5/4.5 5.5/4 5.5/4 5.5/6 5.5/6.5 5.5/3.5 0.11
GE 3.5 3.5/7 3.5/5 3.5/5.5 3.5/3.5 3.5/4.5 3.5/4 3.5/4 3.5/6 3.5/6.5 3.5/3.5 0.07
PS 4.5 4.5/7 4.5/5 4.5/5.5 4.5/3.5 4.5/4.5 4.5/4 4.5/4 4.5/6 4.5/6.5 4.5/3.5 0.09
SL 4 4/7 4/5 4/5.5 4/3.5 4/4.5 4/4 4/4 4/6 4/6.5 4/3.5 0.08
SS 4 4/7 4/5 4/5.5 4/3.5 4/4.5 4/4 4/4 4/6 4/6.5 4/3.5 0.08
DD 6 6/7 6/5 6/5.5 6/3.5 6/4.5 6/4 6/4 6/6 6/6.5 6/3.5 0.12
TR 6.5 6.5/7 6.5/5 6.5/5.5 6.5/3.5 6.5/4.5 6.5/4 6.5/4 6.5/6 6.5/6.5 6.5/3.5 0.13
GF 3.5 3.5/7 3.5/5 3.5/5.5 3.5/3.5 3.5/4.5 3.5/4 3.5/4 3.5/6 3.5/6.5 3.5/3.5 0.07
Arab J Geosci

Fig. 4 a Groundwater recharge zones generated by multi-criteria decision-making (MCDM) and b scatter plot of groundwater recharge index
(GWRIMCDM) versus net recharge, fitted linear model and R2 value for the sites located in suitable zones

Accuracy of MCDM approach respectively) could not be suitable for recharge as runoff from
the highest elevation may easily escape out from the area.
It was observed that 13 of the total 50 net recharge sites were Whereas at the lowest elevations, Udaisagar Lake already
situated in the suitable recharge zone delineated by the acted as recharge structure. The percent slope ranging from
MCDM approach. It is evident from the scatter plot drawn
between GWRIMCDM and the mean annual net recharge
values of the 13 sites (shown in Fig. 4b) that 10 data points Table 4 Boolean logic criteria to delineate suitable/unsuitable
groundwater recharge potential zones
were very close to the fitted linear model especially at the
extremes values. Only three middle points deviated from the Thematic variable Criteria or decision rules for suitable/unsuitable
straight line, which may not be considered significant. recharge zones
Overall, the scatter plot was fairly fitted by the linear regres-
Suitable class or range Unsuitable class or range
sion model. This indicates that the suitable recharge zones
were satisfactory. The R2 value for the fitted model was 0.70 Geomorphology Deep buried pediment Inselburg
(Fig. 4b), which further confirmed the accuracy of the re- Shallow buried Residual hill
charge zone map. Also, the mean annual net recharge for the pediment Structural hill
sites located in the suitable and unsuitable zones was 43.82 Surface waterbody
and 15.11 cm, respectively, which further favoured the delin- Soil Coarse loamy sand Fine loamy rock outcrop
eated zones. Coarse to fine loam Loamy skeletal rock
Fine loam outcrop skeletal fine
loam
Decision rules for favourable recharge zones Land use/land cover Agricultural land Built-up land
Rangeland Forest land
The developed decision rules are listed in Table 4, and justi- Waterbody
fication for the choice of each parameter and logic for choos-
Ground elevation 550–700 m MSL <550 m MSL
ing the favourable class for the artificial recharge is discussed
>700 m MSL
ahead. The deep and shallow buried pediments were the most
Percent slope 1–6 % <1 %
favourable for the artificial recharge in comparison to others,
>6 %
and hence, these were considered for suitable recharge zone.
Slope length >0.782 <0.782
Another criterion was soil texture where three most favourable
Slope steepness 0–1 >1
soil types, i.e. coarse loamy sand, coarse to fine loam and fine
Drainage density 0–1.2 km km−2 >1.2 km km−2
loam, were decided for the suitable recharge zone. From land
Transmissivity >250 m2 day−1 70–250 m2 day−1
use map, either agricultural land or rangeland could be chosen
Groundwater ≥5 m <5 m
as favourable for implementing artificial recharge structures. fluctuation
The highest and lowest elevations (>700 and <550 m MSL,
Arab J Geosci

1 to 6 % might be good for artificial recharge as the slope Validation of BLM approach
lesser than 1 % mainly existed inside the lakes. Also, slope
>6 % was mostly available near boundaries, residual and Point map of the mean annual net recharge estimates at 50
structural hills, where topography was highly steep leading sites was superimposed on the recharge zones as depicted in
to unfavourable conditions. The L > 0.782 and S ≤ 1 were Fig. 5b. It is seen that the net recharge varied from 0 to
considered as positively related with the suitable recharge 121.8 cm. Furthermore, it is apparent that the net recharge
zones. The low and moderate Dd values (≤1.2 km km−2), was mostly large for the sites located in the north-central por-
where runoff had opportunity to stay at the ground surface tion compared to the other sites located elsewhere. In the
and to get recharged, were chosen for the suitable recharge north-central portion, mainly suitable recharge zones existed.
zone. For the favourable recharge condition, area with trans- Also, the net recharge at sites located within the suitable re-
missivity of more than 250 m2 day−1 was considered as suit- charge zones was moderately large in the eastern and south-
able. The aquifer with high transmissivity transmitted the west portions. Thus, the groundwater recharge zone map was
recharged water quantities at relatively high transmission reliable.
rates. As the groundwater fluctuation was considered for the A comparison of the MCDM and BLM approaches re-
dry season, more fluctuation (>5 m) of the groundwater might vealed that the suitable and unsuitable zones for artificial re-
provide more opportunity for the surface water to get percolate charge delineated by two approaches (Figs. 4a and 5a) were
down and fill up the void space within the fractures, fissures or somewhat similar in the north-central, northwest, southwest
faults present in the shallow weathered aquifer formations. and eastern portions. However, in the southern and southeast
portions, the suitable zone delineated by the MCDM was con-
siderably larger than that delineated using the BLM. The suit-
Groundwater recharge zones by BLM approach able recharge zones by the BLM were more sharp and definite
compared to that by the MCDM. Considering the MCDM
The groundwater recharge zone map developed from the approach more reliable, spatial distribution of the correctly
BLM approach is shown in Fig. 5a. The suitable recharge and incorrectly classified suitable/unsuitable zone by the
zone covered 30.55 km2 (9 % area) and were mainly concen- BLM is shown in Fig. 6. The correctly classified areas under
trated in the northwest, southwest, eastern and upper central suitable and unsuitable recharge zones were 13.63 km2 (3.9 %
portions. A large portion, i.e. 317.65 km2 (91 % area), was area) and 298.49 km2 (85.7 % area), respectively, which cov-
delineated as unsuitable for artificial recharge. The area under ered 89.6 % of total area. Thus, the BLM approach-based
suitable and unsuitable zones based on the BLM approach recharge zone map was accurate in 90 % area. Furthermore,
was almost similar to that demarcated by the MCDM ap- it was observed from Fig. 6 that the incorrectly classified area
proach. It is apparent from Fig. 5a that almost entire portion under suitable and unsuitable zones was 19.15 km2 (5.5 %)
in the southern, southeast and near the boundaries was found and 16.90 km2 (4.9 %), respectively. This discrepancy may be
unsuitable for artificial recharge. attributed to features of the BLM where a unit area was

Fig. 5 a Groundwater recharge zones delineated by Boolean logic analysis approach and b point map of net recharge estimates superimposed on
recharge potential zones based on Boolean logic analysis
Arab J Geosci

Fig. 6 Correctly and incorrectly


delineated suitable/unsuitable
groundwater recharge potential
zones by the Boolean logic
modelling (BLM) approach

assigned to a single class of evidence layers and inability to which assigned only binary values (0 or 1) and AND operator
assign partial suitability to each appropriate suitability class function as multiplication resulting either 0 (unsuitable) or 1
(Ahamed et al. 2000). Also, the decision rules were stiff, (suitable).

Fig. 7 Location of the favourable


artificial recharge sites along with
stream network map
superimposed on delineated
recharge potential map
Arab J Geosci

farm reservoir is a small-scale structure that can be built where


first- and second-order stream intersected on agricultural
fields. The check dams and percolation dams are moderate-
scale structures, which could suitably be constructed where
second- and third-order streams intersected each other on
rangelands and forest lands. The exact location and techno-
economic feasibility of the proposed recharge structures may
be found out by detailed site-specific field investigations.

Characterization of dominant variables influencing


recharge

Results of HCA rendered four statistically significant (Dlink /


Dmax × 100 < 75 %) clusters of 29 recharge sites at Euclidean
Fig. 8 Clusters of the identified artificial recharge sites using the Ward’s distance of 20 (Fig. 8). Geographical distribution of 8, 11, 1
method and 9 sites of clusters I, II, III and IV, respectively, on the
transmissivity map is shown in Fig. 9, which depicted that
four clusters almost fell in distinct transmissivity classes.
Suitable artificial recharge sites The site 4 of cluster III fell at the highest transmissivity of
600–2250 m2 day−1, whereas sites of cluster II existed in
The groundwater recharge zone map generated by the BLM low transmissivity (150–300 m2 day−1) class. On the other
approach and drainage network map helped locating side, sites of cluster I associated with moderate transmissivity
favourable artificial recharge sites where first-, second- and (300–450 m2 day−1) and sites of cluster IV came in high trans-
third-order streams intersected with either second- or third- missivity class (450–600 m2 day−1). Thus, it is clear that the
order streams. In such a manner, 29 favourable artificial re- transmissivity might have played a major role in the ground-
charge sites were selected as shown in Fig. 7. The artificial water recharge.
recharge structures recommended for the identified sites were The first two PCs, explaining 73.68, 57.56 and 67.02 %
on-farm reservoir, check dam and percolation dam. The on- variation of system defined by ten variables for clusters I, II

Fig. 9 Spatial distribution of four


clusters of artificial recharge sites
Arab J Geosci

Fig. 10 Principal component a b


(PC) loadings of first two 1.0
1.0 Cluster I Cluster II
significant PCs for three clusters

Principal Component 2 : 27.28%


Principal Component 2 : 31.63%
(GM geomorphology, SO soil, LU SL
land use, GE ground elevation, PS GF
0.5 DD 0.5 SL SO
TR
percent slope, SL Slope length, SS GE
slope steepness, DD drainage
density, TR transmissivity, GF
groundwater fluctuation, GWRI 0.0 GE 0.0
GM
LU
groundwater recharge index) GWRI
SS PS
PS SS
GWRI
-0.5 -0.5 DD
TR GM
LU
GFSO
-1.0 -1.0
-1.0 -0.5 0.0 0.5 1.0 -1.0 -0.5 0.0 0.5 1.0
Principal Component 1 : 42.05% Principal Component 1 : 30.28%

c
1.0 Cluster IV
GWRI TR
Principal Component 2 : 21.12%
GF
0.5
GM

SL
0.0 SS
PS
LU
SO

-0.5 DD
GE

-1.0
-1.0 -0.5 0.0 0.5 1.0
Principal Component 1: 45.98%

and IV, respectively, were plotted on the unit circle of PC variables collectively described the topography, which might
loadings (Fig. 10). The loadings were classified into strong have strong influence on the artificial recharge in cluster II.
(>0.75), moderate (0.50–0.75) and weak (<0.50) significance Likewise, Fig. 10c depicted moderate loadings of transmissiv-
as suggested by Liu et al. (2003). It is worth mentioning that ity (0.70), ground elevation (−0.65), drainage density (−0.55)
GWRIMCDM indicating favourability of artificial recharge and groundwater fluctuation (0.54) on recharge factor. Thus, it
sites had the strong significance (PC loading >0.75) on the is apparent that artificial recharge in cluster IV was largely
first PC in clusters I and II and on the second PC in cluster controlled by aquifer characteristics (transmissivity and
IV, which were termed as ‘recharge factor’. Other variables groundwater fluctuation) in addition to topography.
influencing artificial recharge should have strong to moderate On comparing the above characteristics of clusters I, II and
loadings on the recharge factor and should be close to IV, it is suggested that cluster IV was the most favourable for
GWRIMCDM in Fig. 10. It is discernible from Fig. 10a that the artificial recharge as sites in this cluster had significant
slope steepness (0.85), percent slope (0.81), ground elevation influence of both surface and subsurface variables.
(0.79) and drainage density (0.79) had strong positive load-
ings on the recharge factor, whereas slope length (−0.63),
geomorphology (−0.54) and land use (−0.54) had moderate Conclusions
negative loadings. Thus, it was evident that variables related
to topography were the most influential for artificial recharge In hard-rock catchment of Ahar River located in Udaipur dis-
in cluster I. Similarly, it was apparent from Fig. 10b that the trict, Rajasthan, India, fast-depleting groundwater resources
recharge factor had strong positive loadings of percent slope are at a great risk, which needs to be restored by adopting
(0.84) and moderate loadings of ground elevation (0.74), artificial recharge techniques. This study, for the first time,
slope length (−0.71), and slope steepness (0.66). All these involved and compared both the MCDM and BLM
Arab J Geosci

approaches for delineating groundwater recharge zones and to References


identify favourable artificial recharge sites. A total of ten the-
matic maps were prepared in GIS using satellite image data, Agarwal R, Garg PK, Garg RD (2013) Remote sensing and GIS based
toposheets and conventional maps. The groundwater recharge approach for identification of artificial recharge sites. Water Resour
Manag 27:2671–2689
zone maps were validated from net recharge estimates of 50
Ahamed TRN, Rao KG, Murthy JSR (2000) GIS-based fuzzy member-
sites by graphical and statistical indicators. Then, favourable ship model for crop-land suitability analysis. Agric Syst
artificial recharge sites were identified by superimposing 63(2):75–95
drainage network map on the suitable recharge zone map Awawdeh M, Obeidat M, Al-Mohammad M, Al-Qudah K, Jaradat R
where first-, second- and third-order streams intersected with (2014) Integrated GIS and remote sensing for mapping groundwater
potentiality in the Tulul Al Ashaqif, Northeast Jordan. Arab J Geosci
second- and third-order streams. Finally, the identified sites 7(6):2377–2392
were grouped into four clusters to characterize the most Bonham-Carter GF (1994) Geographic information systems for geosci-
influencing factors by hierarchical cluster and principal com- entists—modelling with GIS, first edn. Love Printing Service Ltd.,
ponent analyses. Pergamon 398 pp
Results indicated that the suitable recharge zone delin- Çelik R (2014) Mapping of groundwater potential zones in the Diyarbakir
city center using GIS. Arab J Geosci. doi:10.1007/s12517-014-
eated by the MCDM and BLM covered 36.07 (10.36 %) 1485-9
and 30.55 km2 (8.77 %), respectively. In suitable re- CGWB (2011) Dynamic ground water resources of India (as on 31
charge zones, the groundwater recharge index showed a March 2009). Central ground water board (CGWB). Ministry of
fair linear relationship with net recharge estimates. Thus, Water Resources, Government of India, p. 243
Chowdhury A, Jha MK, Chowdary VM, Mal BC (2009) Integrated re-
the recharge zone map generated by the MCDM ap-
mote sensing and GIS-based approach for assessing groundwater
proach was reliable, which was further confirmed from potential in West Medinipur district, West Bengal, India. Int J
the significant R2 value of 0.70. A comparison of results Remote Sens 30(1):231–250
of both the MCDM and BLM approaches revealed that Chowdhury A, Jha MK, Chowdary VM (2010) Delineation of ground-
the suitable and unsuitable recharge zones in 312.13 km2 water recharge zones and identification of artificial recharge sites in
West Medinipur district, West Bengal, using RS, GIS and MCDM
area (approximately 90 % of total area) were in agree- techniques. Environ Earth Sci 59:1209–1222
ment and comparable to each other. These findings ver- Dashtpagerdi MM, Nohegar A, Vagharfard H, Honarbakhsh A,
ified the delineated recharge zones. Moreover, 29 artifi- Mahmoodinejad V, Noroozi A, Ghonchehpoor D (2013)
cial recharge sites were identified in suitable recharge Application of spatial analysis techniques to select the most suitable
areas for flood spreading. Water Resour Manag 27:3071–3084
zone where on-farm reservoir, percolation pond and
Deepika B, Kumar A, Jayappa KS (2013) Integration of hydrological
check dams were recommended. Of the four clusters, factors and demarcation of groundwater prospect zones: insights
sites of cluster IV were found as the most favourable from remote sensing and GIS techniques. Environ Earth Sci. doi:
due to influence of both surface and subsurface variables 10.1007/s12665-013-2218-1
on the artificial recharge. These findings may be useful Dillon R, Goldstein M (1984) Multivariate analyses: methods and appli-
cations. Wiley, New York
for decision-makers and researchers in planning and Edmunds WM, Shand P, Hart P, Ward RS (2003) The natural (baseline)
management of the groundwater resources. quality of groundwater: a UK pilot study. Sci Total Environ
Overall, it is suggested that the methodology of uti- 310(1–3):25–35
lizing RS, GIS, MCDM and BLM approaches, and mul- Flug M, Seitz HLH, Scott JF (2000) Multicriteria decision analysis ap-
plied to Glen Canyon Dam. J Water Resour Plan Manag ASCE
tivariate statistical techniques adopted in this study,
126(5):270–276
saves huge amount of time, labour and cost in exploring Freeze RA, Cherry JA (1979) Groundwater. Prentice-Hall, Inc.,
suitable artificial recharge zones and identifying the Englewood Cliffs
most favourable location for recharge sites in the study Ghayoumian J, Ghermezcheshme B, Feiznia S, Noroozi AA (2005)
area. The proposed approach for delineating groundwa- Integrating GIS and DSS for identification of suitable areas for arti-
ficial recharge, case study, Meimeh Basin, Isfahan, Iran. Environ
ter recharge zones may be well-opted in other hard-rock Geol 47(4):493–500
regions of the world. The methodology is most suitable Ghayoumian J, Mohseni Saravi M, Feiznia S (2007) Application of GIS
for developing nations especially those economically techniques to determine areas most suitable for artificial groundwa-
poor regions where adequate information about hydro- ter recharge in a coastal aquifer in southern Iran. J Asian Earth Sci
30:364–374
geology and data of groundwater resources are lacking.
Haan CT (2002) Statistical methods in hydrology. Iowa State Press, Iowa
496 pp
Hajkowicz S, Higgins A (2008) A comparison of multiple criteria analy-
Acknowledgments The groundwater level and pumping test data col- sis techniques for water resource management. Eur J Oper Res 184:
lected from the All India Coordinated Research Project on Groundwater 255–265
Utilization, College of Technology and Engineering, MPUAT, Udaipur, Hammouri, N., Al-Amoush, H., Al-Raggad, M. and Harahsheh, S.
Rajasthan, India, is gratefully acknowledged. The authors are grateful to (2014). Groundwater recharge zones mapping using GIS: a case
three anonymous reviewers for providing useful suggestions, which im- study in southern part of Jordan valley, Jordan. Arab J Geosci,
proved quality of earlier version of this article. 7(7): 2815-2829.
Arab J Geosci

ILWIS (2001) Integrated land and water information system, 3.2 Mahmoud SH (2014) Delineation of potential sites for groundwater re-
Academic, User’s Guide. International Institute for Aerospace charge using a GIS-based decision support system. Environ Earth
Survey and Earth Sciences (ITC), The Netherlands, pp. 428–456 Sci 72(9):3429–3442
Jain BL, Singh RS, Shyampura RL, Gajbhiye KS (2005) Land use plan- Mahmoud SH, Alazba AA, Amin MT (2014) Identification of potential
ning of Udaipur district—soil resource and agro-ecological assess- sites for groundwater recharge using a GIS-based decision support
ment. NBSS Publication No. 113. National Bureau of Soil Survey & system in Jazan region—Saudi Arabia. Water Resour Manag 28:
Land Use Planning, Nagpur 69 p 3319–3340
Jasrotia AS, Kumar R, Saraf AK (2007) Delineation of groundwater Malczewski J (1999) GIS and multicriteria decision analysis. Wiley, New
recharge sites using integrated remote sensing and GIS in Jammu York 392 pp
district, India. Int J Remote Sens 28(22):5019–5036 McCool DK, Brown LC, Foster GR, Mutchler CK, Meyer LD (1987)
Jha MK, Chowdary VM, Chowdhury A (2010) Groundwater assessment Revised slope steepness factor for the universal soil loss equation.
in Salboni Block, West Bengal (India) using remote sensing, geo- Trans ASAE 30:1387–1396
graphical information system and multi-criteria decision analysis Nampak H, Pradhan B, Manap MA (2014) Application of GIS based data
techniques. Hydrogeol J 18(7):1713–1728 driven evidential belief function model to predict groundwater po-
Joubert A, Stewart TJ, Eberhard R (2003) Evaluation of water tential zonation. J Hydrol 513:283–300
supply augmentation and water demand management options Oberlander TM (1989) Slope and pediment systems. In: Thomas DSG
for the City of Cape Town. J Multi-Criteria Decis Anal (ed) Arid zone geomorphology. Belhaven, London, pp. 58–59
12(1):17–25 Otto M (1998) Multivariate methods. In: Kellner R, Mermet JM, Otto M
Kaliraj S, Chandrasekar N, Magesh MS (2014) Identification of potential and Widmer HM (eds) Analytical chemistry. Wiley-VCH,
groundwater recharge zones in Vaigai upper basin, Tamil Nadu, Weinheim, 916 pp.
using GIS-based analytical hierarchical process (AHP) technique. Rahimi S, Roodposhti MS, Abbaspour RA (2014) Using combined AHP-
Arab J Geosci 7(4):1385–1401 genetic algorithm in artificial groundwater recharge site selection of
Karanth KR (1987) Ground water assessment: development and manage- Gareh Bygone Plain, Iran. Environ Earth Sci 72:1979–1992
ment. Tata McGraw-Hill Publishing Company Limited, New Delhi, Ramaswamy SM, Anbazhagan S (1997) Criteria and techniques of de-
p. 720 tecting site-specific mechanisms for artificial-recharge—a case
study from Ayyar basin, India. J Geol Soc India 50:449–456
Krishnamurthy J, Venkatesa Kumar N, Jayaraman V, Manuvel M (1996)
Ravi Shankar MN, Mohan G (2005) A GIS based hydrogeomorphic
An approach to demarcate groundwater potential zones through re-
approach for identification of site-specific artificial-recharge tech-
mote sensing and a geographical information system. Int J Remote
niques in the Deccan Volcanic Province. J Earth Syst Sci 114(5):
Sens 17(10):1867–1884
505–514
Lachassagne P, Wyns R, Bérard P, Bruel T, Chéry L, Coutand T, Desprats
Robinov CJ (1989) Principles of logic and the use of digital geo-
J-F, Strat PL (2001) Exploitation of high-yields in hard-rock aqui-
graphic information systems. In: Ripple WJ (ed) Fundamentals
fers: downscaling methodology combining GIS and multicriteria
of geographic information systems compendium. American
analysis to delineate field prospecting zones. Ground Water 39(4):
Society for Photogrammetry and Remote Sensing, Bethesda,
568–581
pp. 61–80
Liu C-W, Lin K-H, Kuo Y-M (2003) Application of factor analysis in the
Rolland A, Rangarajan R (2013) Runoff estimation and potential re-
assessment of groundwater quality in a Blackfoot disease area in
charge site delineation using analytic hierarchy process. Geocarto
Taiwan. Sci Total Environ 313:77–89
Int 28(2):159–170
Machiwal D, Jha MK (2014) Characterizing rainfall-groundwater dy- Saaty TL (1980) The analytic hierarchy process: planning, priority set-
namics in a hard-rock aquifer system using time series, geographic ting, resource allocation. McGraw-Hill, New York 287 pp
information system and geostatistical modelling. Hydrol Process 28: Saraf AK, Choudhury PR (1998) Integrated remote sensing and GIS for
2824–2843 groundwater exploration and identification of artificial recharge
Machiwal D, Srivastava SK, Jain S (2010) Estimation of sediment yield sites. Int J Remote Sens 19(10):1825–1841
and selection of suitable sites for soil conservation measures in Ahar Saraf AK, Choudhury PR, Roy B, Sarma B, Vijay S, Choudhury
river basin of Udaipur, Rajasthan using RS and GIS techniques. J S (2004) GIS based surface hydrological modelling in iden-
Indian Soc Remote Sens 38(4):696–707 tification of groundwater recharge zones. Int J Remote Sens
Machiwal D, Jha MK, Mal BC (2011a) Assessment of groundwa- 25(24):5759–5770
ter potential in a semi-arid region of India using remote Shaban A, Khawlie M, Abdallah C (2006) Use of remote sensing and GIS
sensing, GIS and MCDM techniques. Water Resour Manag to determine recharge potential zones: the case of Occidental
25(3):1359–1386 Lebanon. Hydrogeol J 14:433–443
Machiwal D, Nimawat JV, Samar KK (2011b) Evaluating efficacy of Singh S (2002) Water management in rural and urban areas. Agrotech
groundwater level monitoring network by graphical and multivariate Publishing Academy, Udaipur 192 p
statistical techniques. J Agricult Eng ISAE 48(3):36–43 Singh A, Panda SN, Kumar KS, Sharma CS (2013) Artificial groundwa-
Machiwal D, Mishra A, Jha MK, Sharma A, Sisodia SS (2012) Modeling ter recharge zones mapping using remote sensing and GIS: a case
short-term spatial and temporal variability of groundwater level study in Indian Punjab. Environ Manag 52(1):61–71
using geostatistics and GIS. Nat Resour Res 21(1):117–136 Solomon S, Quiel F (2006) Groundwater study using remote sensing and
Machiwal D, Rangi N, Sharma A (2015) Integrated knowledge- geographic information systems (GIS) in the central highlands of
and data-driven approaches for groundwater potential zoning Eritrea. Hydrogeol J 14:729–741
using GIS and multi-criteria decision making techniques on USGS (2004) Shuttle Radar Topography Mission, 1 Arc Second scene
hard-rock terrain of Ahar catchment, Rajasthan, India. SRTM_u03_n008e004, Unfilled Unfinished 2.0, Global Land
Environ Earth Sci 73:1871–1892 Cover Facility, University of Maryland, College Park, Maryland,
Madrucci V, Taioli F, de Araújo CC (2008) Groundwater favorability map February 2000
using GIS multicriteria data analysis on crystalline terrain, São Paulo Yeh H-F, Lee C-H, Hsu KC, Chang PS (2009) GIS for the assessment of
State, Brazil. J Hydrol 357:153–173 the groundwater recharge potential zone. Environ Geol 58:185–195

You might also like