1 s2.0 S258975782030007X Main

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

HydroResearch 3 (2020) 1–14

Contents lists available at ScienceDirect

HydroResearch

journal homepage: http://www.keaipublishing.com/en/journals/hydroresearch/

GIS-based multi-criteria analysis for identification of potential


groundwater recharge zones - a case study from Ponnaniyaru watershed,
Tamil Nadu, India
Devanantham Abijith a, Subbarayan Saravanan a,⁎, Leelambar Singh a, Jesudasan Jacinth Jennifer a,
Thiyagarajan Saranya a, K.S.S. Parthasarathy b
a
Department of Civil Engineering, National Institute of Technology, Tiruchirappalli, India
b
Department of Applied Mechanics and Hydraulics, National Institute of Technology Karnataka, Surathkal, India

a r t i c l e i n f o a b s t r a c t

Article history: Groundwater is one of the most vital natural resources; spatially varying in quality and quantity. Increased urban-
Received 8 November 2019 isation and population creates tremendous pressure on the quality and quantity of the groundwater resources. In
Received in revised form 4 February 2020 this study, Ponnaniyaru watershed of Cauvery basin was considered for this research. Geographical information
Accepted 7 February 2020
system (GIS) and remote sensing (RS) plays a vital role in preparing various thematic layers for targeting the
Available online 21 February 2020
groundwater potential zones (GWPZ). This study adopts the Analytical Hierarchy Process (AHP) and Multi influ-
Keywords:
ence factor (MIF), multi-criteria decision-making approaches to determine the weights for the influencing fac-
AHP tors. Weighted linear overlay analysis was carried out to determine the GWPZ. Further, the resultant GWPZ
MIF map has been reclassified into five different classes, namely Very good, Good, Moderate, Poor and Very poor.
GIS The results were validated with observed well-yield data, and the predictive precision for AHP and MIF was
Groundwater potential zone found to be 75%, and 71% respectively.
Ponnaniyaru © 2020 The Authors. Production and hosting by Elsevier B.V. on behalf of KeAi Communications Co., Ltd. This is an
open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction Therefore, identifying the potential zone of groundwater becomes an in-


tegral part of water resource management and planning (Kulkarni et al.,
Climate change has a significant effect on earth systems which is ev- 2018). India is one of the world's biggest groundwater consumer and
ident through severe water scarcity in recent decades. Due to the rapid the consumption rate is estimated to be 230 cubic km per year
increase in population, industrialization and irrigation, the demand for (Fienen and Arshad, 2016). Groundwater contributes 62% for irrigation,
water increased, which severely influenced groundwater stress. There- 85% for agricultural water supply and 45% for urban water supply
fore the issue is considered with primary importance to conserve and (CGWB, 2013). N60% of the irrigated cropland and 85% of the drinking
maintain the available water resources (Chezgi et al., 2016; water supply depend on the groundwater sources. In India, over 90%
Venkatesan et al., 2019). Water Security is expected to be the biggest of the rural population and almost 30% of the metropolitan population
challenge in the 21st century (Snyder, 2019). The city of Cape Town rely only on groundwater to meet their drinking and household de-
has implemented day zero's concept for focusing everyone's attention mands resulting in overexploitation in some parts of India
on controlling water consumption as tightly as possible, and literally, (Parthasarathy and Deka, 2019; Siddha and Sahu, 2018).
day zero is when most of the town's taps are turned off. India experiences sufficiently good annual rainfall on an average of
In Asia, rainwater is the prime source of groundwater, which reach 1100 mm, however, the uneven spatial distribution of rainfall results
down through percolation to the aquifer (Ray and Shaw, 2019). In the in deprivation of the normal rainfall in some regions even during mon-
past decade, demand for water increased due to human development soon which causes water scarcity (Garg and Hassan, 2007). The water
activities and climate change (Shah, 2005; Taweesin et al., 2018). crisis in India is often linked to a lack of government's management
plan, unscientific utilization of water resources, industrial and human
⁎ Corresponding author at: Department of Civil Engineering, National Institute of
intervention. Furthermore, water scarcity is anticipated to stagnate in
Technology, Tiruchirappalli, Tamil Nadu, India. India as the general population is predicted to rise to 1.6 billion by
E-mail address: ssaravanan@nitt.edu (S. Saravanan). 2050 (Snyder, 2019). Therefore, groundwater potential mapping is

https://doi.org/10.1016/j.hydres.2020.02.002
2589-7578/© 2020 The Authors. Production and hosting by Elsevier B.V. on behalf of KeAi Communications Co., Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
2 D. Abijith et al. / HydroResearch 3 (2020) 1–14

crucial to understand the status of the subsurface water sources and (WOE) method uses the Bayesian probability model's log-linear form
that could be beneficial for sustainable groundwater management. to assess through relative statistical importance of evidence (Nampak
Integrating RS and GIS has become a useful technique for various ap- et al., 2014; Lee et al., 2012 and Mogaji et al., 2016). Entropy weighted
plications in water resource and groundwater mapping. The presence of linear aggregate (LWA) is a weighted average-based technique in
groundwater could be assessed by the interaction of hydrological, geo- which the continuous criteria are normalized to a common numeric
logical and atmospherical factors (Ghosh et al., 2016; Msaddek et al., scale and combined using a weighted average. The assigned weights ei-
2019; Saravanan, 2012). Multi-criteria decision making (MCDM), prob- ther subjectively or objectively are used along with standardized criteria
abilistic models (PM), bivariate statistics (BS), multivariate statistics as input for the LWA (Drobne and Lisec, 2009).
(MS), and data-mining/machine learning (DM/ML) were few methods In recent years, the data-mining/machine learning (DT/ML) tech-
adopted for more accurate groundwater potential zone mapping in var- nique has been widely adopted in groundwater studies (Lee et al.,
ious regions worldwide. From the previous studies, it is evident that 2019). These hybrid statistical methods includes artificial neural net-
MCDM is an effective tool for decision making for the problem involving work (ANN), boosted regression tree (BRT) (Schapire, 2003) Random
too many influencing factors (Chenini et al., 2010). In the recent past, forest (RF) (Breiman, 2001), decision tree (DT) (Duan et al., 2016), hy-
numerous studies were carried out in adopting a semi-quantitative ap- brid genetic algorithm (HGA), self-adaptive hybrid genetic algorithm
proach such as AHP (Sandoval et al., 2018) and (Arulbalaji et al., 2019) (SAHGA), fisher's linear discriminant function (FLDA) (Espinoza et al.,
frequency ratio model (FR). BS model provides an easy and fast 2005), least-square support vector machine (LSSVM) (Tang et al.,
geospatial assessment for calculating the probabilistic relationship be- 2019), extreme learning machine (ELM) (Alizamir et al., 2018).
tween the dependent and independent variables (Guru et al., 2017; The approaches mentioned above-made identification of Ground-
Naghibi et al., 2015; Razandi et al., 2015). Logistic regression model water potential zone (GWPZ) simple, reliable, cost-effective with each
(LR) is opted to identify the suitable model to define the relationship be- model having its own merits and demerits. AHP method significantly
tween a dependent variable and conditioning factors to produce the co- reduces the mathematical complexity of decision-making, where a
efficient for each variable (Umar et al., 2014). The ratio of each score is assigned based on expert's opinion, it is a pairwise comparison
conditioning factor can be estimated by using the coefficients derived where score one represents equal influence and score nine represents
from LR. Evidential belief function model (EBF) according to Dempster, the extreme influence of parameters. Whereas MIF is also based on ex-
generalizes the lower and upper probabilities in Bayesian (Dempster, pert opinions, but the score is assigned based on the major and minor
1967). Degrees of belief (Bel) and plausibility (Pls) are the lower and influence between parameters influencing the GWPZ. These two ap-
upper probabilities. The feature of the EBF is the degree of uncertainty proaches have managed to gain more popularity, as it is relatively sim-
(Unc) and the degree of doubt (Dis) which improve the precision of ple and accurate. It also helps in identifying the most sensitive area of
forecasts of groundwater. (Carranza and Hale, 2002; Althuwaynee GWPZ and assessing the error that leads to greater certainty.
et al., 2014; Pourghasemi, 2016). The EBF is a renowned spatial predic- (Anbarasu et al., 2019; Pradhan, 2009; Shimpi et al., 2019). In this
tion tool, a robust framework that helps formalize and address problem- paper, attempts were made to delineate the groundwater potential
atic management problems (Al-Abadi, 2017). The weights-of-evidence zone for Ponnaniyaru watershed using AHP and MIF methods. Several

Fig. 1. Study area map of Ponnaniyaru watershed.


D. Abijith et al. / HydroResearch 3 (2020) 1–14 3

Table 1 Cauvery. This watershed is a major attraction and attention for several
Data sources used for mapping groundwater potential mapping. geologists for its structural configuration and presence of anorthosite
Parameters Data source Source location (Maheswaran et al., 2019; Muralitharan and Palanivel, 2015). This wa-
Geology Geological Survey of India https://www.gsi.gov.
tershed is bestowed with hard rock terrain and has an unconfined aqui-
in fer system and N110 villages are present in the watershed depending on
Geomorphology Landsat 8 OLI https://earthexplorer. the groundwater for domestic and agricultural activities. The watershed
06-MAR-19 usgs.gov gets the majority of its rainfall from the North-East monsoon and the av-
Landuse and Landsat 8 OLI https://earthexplorer.
erage rainfall is estimated to be 816 mm. The mean minimum and a
Landcover 06-MAR-19 usgs.gov
Soil National Bureau of Soil Survey and Land https://www. maximum temperature range from 23 °C to 38.56 °C. (CGWB, 2008)
Use Planning (NBSS and LUP) nbsslup.in The maximum elevation of the study area has been observed to be
Lineament Shuttle Radar Topography Mission https://earthexplorer. 1016 m in the North West part and the eastern part of the watershed
density (SRTM) usgs.gov is dominated by plain region. The slope ranges from 0° to 60° degree to-
(km/km2)
Drainage Shuttle Radar Topography Mission https://earthexplorer.
wards East-North-East. The predominant geology of the watershed is
density (SRTM) usgs.gov found to be gneiss and followed by granite, syenite, Conglomerate,
(km/km )2
Sandstone and Quartz Vein which are seen as patches. Geomorphologi-
Slope (°) Shuttle Radar Topography Mission https://earthexplorer. cal units like pediplain has covered most of the area followed by struc-
(SRTM) usgs.gov
tural hills, Denudational hills, and bazada. Majority of soil types are
Rainfall(mm) Regional Meteorological Centre India http://imdchennai.
Meteorological Department gov.in/ categorized under well-drained and moderately drained class. The lat-
est report of the Oxford Economics research suggested that
Tiruchirappalli district is one of the fastest-growing cities and the with
studies for investigating the subsurface geology were carried out in the tremendous increase in growth rate over the past two decades (Holt,
selected region but no attempts were made for mapping the groundwa- 2018). Landuse and Landcover of the study area are found to be com-
ter potential zone for this watershed. The robustness of these tech- prised of gullied land, barren rocks, forest, land with scrub, plantation,
niques has been validated using ROC and AUC from the observed well river, settlements and land without scrub and cropland. Major crops
yield data. grown in the watershed are Paddy, Millets, Pulses, Banana, and
Sugarcane.
2. Description of the study area
3. Methodology and Data used
Ponnaniyaru watershed (Fig. 1) of Tiruchirappalli district, Tamil
Nadu geographically located in 10°27′00” N to 10°50′0” N latitude and The primary and secondary data sources are described in Table 1. The
78°08′0″ E to 78°40′00″ E longitude having an area of 819.2 sq.km. thematic maps required for this study were availed from various
The Ponnaniyaru River is 60 km long and confluences at the River sources and were digitized and pre-processed using GIS tools. The

Fig. 2. Methodology flow chart of the study.


4 D. Abijith et al. / HydroResearch 3 (2020) 1–14

Table 2
AHP weights factors for groundwater potential mapping.

Parameters a b c d e f g h Normalized AHP weights

Geomorphology(a) 1 2 2 4 3 3 2 3 0.24
Lineament density (b) 0.5 1 2 2 3 2 3 4 0.19
Landuse and Landcover (c) 0.5 0.5 1 3 4 4 3 3 0.18
Geology(d) 0.25 0.5 0.33 1 2 3 2 3 0.14
Slope (e) 0.33 0.3 0.25 0.5 1 3 4 2 0.1
Soil(f) 0.33 0.5 0.25 0.33 0.33 1 2 2 0.07
Drainage Density(g) 0.5 0.3 0.3 0.5 0.25 0.5 1 3 0.07
Rainfall(h) 0.33 0.25 0.3 0.33 0.5 0.5 0.33 1 0.04

Table 3
visual and digital interpretation of satellite imagery Landsat-8 OLI
MIF weight for groundwater potential mapping. (Path 143 Row 53) and cross-verified with published district resources
map. From the National Bureau of Soil Survey and Land Use Planning
Parameters Major Minor Relative weights
(NBSS and LUP), the soil data was obtained and digitized. Landsat-8
GEOLOGY 1+1+1+1 0.5 4.5 OLI (Path 143 Row 53) satellite data were used to generate Land-use
LULC 1+1 0.5 + 0.5 + 0.5 + 0.5 4
and Land-cover map and using Shuttle Radar Topography Mission
GM 1+1 0.5 + 0.5 + 0.5 3.5
DD 1+1 0.5 + 0.5 3 (SRTM) Version 3.0 Global 1 arc sec data slope map was prepared.
SLOPE 1. + 1 0.5 2.5 False colour composite in spatial analysis tool in ArcGIS 10.4 was used
LD 1+1 0.5 2.5 to extract the lineaments and their density have been calculated. Layers
RAIN 1 0.5 + 0.5 2 extracted through digitization process were transformed to raster data
SOIL 1 1
using the feature to raster tool in ArcGIS. Rainfall data were obtained
from the Indian Meteorological Department (IMD) and spatial distribu-
tion was generated by using Inverse Distance Weightage (IDW) tool in
methodology adopted is shown in Fig. 2. The eight thematic layers ArcGIS 10.4 All the layers including delineated watershed were
opted in this study includes geomorphology, lineament density, land- resampled using nearest neighbourhood technique and converted into
use/land cover, geology, slope, soil, drainage density, and rainfall that 30 m * 30 m cell size.
are found to be the influencing factors essential in assessing the ground- Using AHP and MIF, with respect to the scale based on the priority,
water potentiality. All the maps were in Universal Transverse Mercator the weights were assigned to each factor. Then the layers were inte-
(UTM) coordinate system and WGS84 spatial reference (WGS84-UTM- grated using GIS to find the GWPZ and classified as very high, high,
Zone44N). From the published Survey of India (SOI) toposheet (58 J, moderate, low and very low zones. The resultant GWPZ map was vali-
1:250,000) the watershed was delineated, and the drainage pattern dated with 56 observed well-yields of various present wells, observa-
was digitized using ArcGIS 10.4 software. The geology map was pre- tion wells, boreholes and tube wells obtained from Central
pared by digitizing the district resource map of the Geological Survey Groundwater Board (CGWB). Verification of the groundwater potential
of India (GSI) and geomorphology map has been prepared from the map using the groundwater yield data shows how reliable and efficient

Fig. 3. Multi influencing factors for groundwater potential mapping.


D. Abijith et al. / HydroResearch 3 (2020) 1–14 5

was the adopted technique compared to the groundwater potential Table 4


mapping using ROC and AUC. AHP and MIF rank and weight for calculating groundwater potential.

Parameters Rank MIF weight AHP weight


3.1. Analytic hierarchy process (AHP) Rain
786.93–810.69 10 8.5 0.04
Multi-criteria decision-making technique is a widely used technique 810.69–841.05 8
for groundwater management, and AHP is one of them which is pro- 841.05–878.01 6
878.01–917.61 4
posed by (Saaty, 1980); According to (Şener et al., 2011), it defined as
917.61–955.24 2
extensive technology that integrates practical and subjective opinions
of experts to complete decision-making by assessing several factors. In Landuse and Landcover
Plantation 9 17 0.18
this method, each factor assigned rank on a scale of one (equal signifi-
Settlements 3
cance) to nine (extreme significance) by assessing each element based Water Bodies 10
on the pairwise comparison. This analysis evaluates the importance of Land Without Scrub 5
the parameter that influences groundwater potential (Razandi et al., Barren Rocky 2
2015). The prepared layers are assigned weights derived by this tech- Land With Scrub 7
Forest 6
nique and were subjected to weighted overlay analysis. In pair-wise
Guilled Land 4
comparison matrix, row follows the reciprocal value of each parameter Crop Land 8
and its importance with the second parameter. The generic pair-wise
Geology
comparison matrix is made by normalizing pair comparison matrix.
Gneiss 3 19 0.14
After that weights are estimated by arithmetic mean. The consistency Granite 2
index (CI) is estimated by the corresponding expressions (Eqs. (1) and Syenite 2
(2)) (Saaty, 1980): Conglomerate, sandstone and shale 8
Shales with bands of limestone 7
CI Quartz vein 6
CR ¼ ð1Þ
RI Lineament density
0–346.119043 2 10.6 0.19
CI is consistency index, CR is consistency ratio, and RI is the random 346.11–610.79 4
610.79–868.69 6
index,
868.69–1160.51 8
1160.51 - 1730.59 10
λ max−n
CI ¼ ð2Þ Drainage density
n−1
0–1029.103581 2 12.7 0.07
1029.10–1976.96 4
where n is the number of conditioning factors, and λmax is the principle
1976.96–2897.73 6
eigenvalue of the matrix. RI value is directly taken from the (Saaty, 2897.73–3953.92 8
1980) research paper, which depends upon the number of components. 3953.92–6905.82 10
The weights determined using the matrix is valid only when CR ≯0.1.
Slope
Pairwise comparison matrix is shown in Table 2. 0.01–2.86 10 10.6 0.1
2.86–8.09 8
3.2. Multi-influencing factors (MIF) 8.09–16.41 6
16.41–25.20 4
25.20–60.63 2
The weights of the various influence parameters are determined
using MIF techniques. The MIF is shown to be feasible to understand Soil
the factors influencing the potentiality of groundwater. The MIF is a Well Drained 9 6.3 0.07
Somewhat Excessively Drained 8
quick and effective method wherein the weights for all the input vari-
Moderately well Drained 7
ables can be determined effectively with greater accuracy. The multi- Imperfectly Drained 5
influencing groundwater variables are described in Table 3. A weight
Geomorphology
of 1.0 and 0.5, is allocated to the effects of each major and minor variable
Pediplain 9 14.8 0.24
respectively. The combined weight of both major and minor is consid- Structural Hills 2
ered for calculating the comparative rates. Each connection is weighted Denudational Hills 3
by its effectiveness (Fig. 3) (Singh et al., 2019; Thapa et al., 2017). The Bazada 7
estimated weight for each influencing factor was equally distributed
and applied to each subclass as a rank for all seven factors (Eq. (3)).
 
ðE þ Em Þ between 1 and 10. The weights are calculated in such a way that impor-
P e  100 ð3Þ tance of influence between each parameter.
ðE e þ E m Þ

where 4.1. Geomorphology

Ee −Major Influencing factor; Em −Minor Influencing factor Geomorphology (Fig. 4) impacts the occurrence significantly, water
percolation and recharge into the earth's subsurface (Etikala et al.,
2019).Various landforms using image interpretation techniques which
4. Results and discussion helps in identifying the prospects of groundwater. The Study area has
structural hills, Denudational hills, pediplain, and bazada. Structural
Table 4 denotes the weights and ranks of each parameter in AHP and hills in the northern part where it has more runoff and poor in ground-
MIF. Those weights are multiplied to each sub-feature inside the param- water potential. Denudational hills have a gentle slope with medium
eter. For each parameter, the weights of AHP and MIF is feed according vegetation cover and infiltration can happen as it has some fissures.
to subclass influences the ranks are assigned, and the rank ranges Pediplains are widely found to occur in the research region, having
6 D. Abijith et al. / HydroResearch 3 (2020) 1–14

Fig. 4. Geomorphology of the study area.

gentle steepness with wealthy vegetation and rich in groundwater po- 4.2. Lineament
tential. Bazada zones are deposition along the stream channel which is
very rich in potential. Pediplain is assigned high rank followed by The fractured zones are called lineaments which formed due to de-
bazada, Denudational hills and least rank is assigned to structural hills. formations which have considerable permeability that is potential for
The derived AHP weight is 0.24 and MIF weight is 14.8 which has two groundwater (Sander, 2007). Lineament map (Fig. 5) is prepared from
major and three minor influences. satellite data visual interpretation and then line density is calculated

Fig. 5. Lineament density of the study area.


D. Abijith et al. / HydroResearch 3 (2020) 1–14 7

Fig. 6. Landuse and Landcover of the study area.

(Eq. (4)). Density Scale of high to low directly implies the groundwater and least rank to 0–0.346 km/km2. The derived AHP weight is 0.19
potential and lineaments are aligned in E-SW, NNE-SSW, NW-SE and E- and MIF weight is 10.6 which has two major and one minor influence
W in the study area. Groundwater potential is high near intersection
areas of lineaments. Higher lineament concentrations are found in the
X
n
Li
study area's northeast and south-central section. Lineament density is Ld ¼ ð4Þ
classified into five classes, High rank is assigned to 1.16–1.73 km/km2 i¼1
A

Fig. 7. Ground truth field points for classifying LULC.


8 D. Abijith et al. / HydroResearch 3 (2020) 1–14

where which regulates the infiltration of groundwater, therefore, ranks were


assigned accordingly. Conglomerate, Sandstone and shale are good
 
aquifer and aquitard
2 as it is highly weathered and fractured. The high
Ld−lineament density; Li −Total length of lineaments ðKmÞ; and A area of the grid km
ranks were given to Conglomerate, sandstone and shale and lower
rank for Granite and Syenite. The derived AHP weight is 0.14 and MIF
weight is 19 which has four major influencing factors and one minor
4.3. Landuse and Landcover influence.

Land-use and land cover influence groundwater potential by affect- 4.5. Slope
ing the evapotranspiration, soil erosion and runoff (Saravanan et al.,
2019; Wang et al., 2005). LULC (Fig. 6) was interpreted using Landsat- Slope (Fig. 9) controls the runoff which in turn controls the infiltra-
8 Satellite data Supervised classification technique was adopted and tion of rainwater. In the flat and sloping area, infiltration capacity is ex-
Level 1 classification of Anderson was used to classify. (Fig. 7) presents pected to be higher (Subba Rao et al., 2006) due to more inundation
field photos during the field visit to gather information about the time. High quantities of runoff and less infiltration are linked to regions
ground truth for the Landuse classification. The study area has Gullied with steep elevation angles (L. Singh and Katpatal, 2018). The hilly re-
land, Barren Rocks, cropland, forest, land with scrub, plantation, river, gion of the research area shows high slope and the overall area has a
settlements and land without scrub. Settlements have a diminishing im- gentle slope and the prospects for groundwater is good. The slope is
pact on water percolation and Vegetation cover influences groundwater graded into five classes and allocated according to the slope degree
recharge by water absorption. Barren rocks and settlement areas are and slope b 1° is considered a plain area with a lower slope due to the
poor in groundwater potential were as cropland, plantations are very low runoff that is usually a very good recharge zone. A high rank was
good for groundwater potential. Waterbody was given high rank assigned to lower slope from 1° to 2° and the least rank is assigned to
followed by cropland, plantation, forest and least rank is assigned for 25°-60°.The derived AHP weight is 0.1 and MIF weight is 10.6 which
settlement. The derived AHP weight is 0.18 and MIF weight is 17 has two major and one minor influencing.
which has two major and four minor influencing factors.
4.6. Soil
4.4. Geology
Soils (Fig. 10) emerged from rocks fragments as a result of the
Geology is the most significant element for groundwater potential weathering process. Soil characteristics are important prospects in de-
because the infiltration depends primarily on the permeability of spe- lineating groundwater potential as it controls the water holding capac-
cific types of rock (Aggarwal et al., 2019; Mogaji et al., 2016). The geol- ity and ranks is allocated to kind of soil depending on the ability of
ogy map was prepared from a district resource map procured from the composition and water retention (Stephenson, 1979). Soil data pre-
Geological Survey of India (GSI).Gneiss, Granite, Syenite, Conglomerate, pared by the Indian Council of Agricultural Research – NBSS & LUP is
Sandstone and Shale, Shales with bands of limestone and Quartz Vein exploited in this study. Soil is classified according to the drainage classes
are the geological forms found in the area of study is shown in (Fig. 8) which implies the duration of wetness and imperfectly drained,

Fig. 8. Geology of the study area.


D. Abijith et al. / HydroResearch 3 (2020) 1–14 9

Fig. 9. Slope of the study area.

moderately well-drained, well-drained, and somewhat excessively 4.7. Drainage density


drained are found in the study area. Mostly the area is covered by
well-drained soil and those are cropland regions. High rank is given to Drainage network helps in evaluation of recharge levels. The drain-
Well Drained and least rank assigned to imperfectly drained soils. The age density is calculated as the total stream length divided by the total
derived AHP weight is 0.07 and MIF weight is 6.3, which has one basin area (Eq. (5)). The high-density regions have high percolation
major influences. level which means low runoff (Nasir et al., 2018). In low elevated

Fig. 10. Soil of the study area.


10 D. Abijith et al. / HydroResearch 3 (2020) 1–14

regions, the number of streams is more, and drainage density (Fig. 11) 5. Groundwater potential zone mapping
can directly indicate the groundwater potential. Drainage density
reclassified into five classes, Higher rank is assigned to 3.95–6.90 km/ In this study, AHP and MIF are used, and thematic layers were
km2 and least given to 0–1.02 km/km2.The derived AHP weight is 0.07 assigned weights for detection of GWPZ. All the layers were converted
and MIF weight is 12.7, which has two major, two minor influence. into raster format with 30 m*30 m cell size and multiplied with derived
weights and rank using ArcGIS10.4. The final output produced from
weighted overlay analysis the groundwater potential is evaluated by in-
n
∑i¼1 ðDi Þ tegrating all layers in ArcGIS raster calculator and the results were cate-
Dd ¼ ð5Þ
A gorized into five classes using natural breaks (Jenks optimization
method), namely very low, low, moderate, high and very high to pro-
duce the final GWPZ map (Eq. (6)).
where, Dd - drainage density, Di -the sum of the length of all streams,
A is the area in km2. X
GPZ ¼ ½GLW GLR þ GMW GM R þ SLW SLR þ LU W LU R þ LDW LDR þ SOW SOR þ RF W RF
ð6Þ
4.8. Rainfall

Rainfall (Fig. 12) is the only primary source of groundwater recharge Where, W and R - represents weight and rank respectively of each
and the major source of water in the hydrological cycle and the most parameter.
prevalent influencing factor in a region's groundwater. Rainfall is the The low potential region is found in the northern and western region
primary cause of groundwater recharge where water is infiltrated by of the study area, which is covered by hilly terrain. High potential is
fractures and soil into the subsurface it is also essential to have other found near in the central parts of the study area which has croplands
rainfall features such as duration and intensity to calculate runoff (Yu were the infiltration is high. The most prominent groundwater pros-
and Lin, 2015). Rainfall data were obtained from Regional Meteorologi- pects identified are pediplain in geomorphology, expect gneiss in geol-
cal Centre, Chennai and data are interpolated spatially using the inverse ogy, cropland in Landuse and Landcover, well-drained soil are found
distance weighting method as this approach is more appropriate for across the study area. High groundwater potential is seen in the central
data-sparing regions. IDW assigns weights to adjacent observed values region of the watershed and the slope 2–8% is gentle and proportionally
based on the distance to the interpolation location, and the interpolated it has high drainage density. The north-west part of the watershed com-
value is the weighted average of the observations, the average annual prises of structural hills and their Lineament density is low and those
rainfall of the study area is 956 mm. Rainfall is reclassified into five clas- areas are under the category low and very low. As rain is the only source
ses, the entire area is in fair condition, whereas there is sufficient rainfall of this watershed rainfall plays a critical role, low rainfall is identified
to the north-eastern regions. High rainfall is given a higher rank of across the watershed; therefore, south-east regions are moderate level
917.61–955.24 mm and least rank is assigned to 790- 810 mm.The de- for groundwater potential. The possible areas of good and poor ground-
rived AHP weight is 0.04 and MIF weight is 8.5 which has one major, water are spread primarily over steep slopes and ridges. Thus finding
two minor influence. the groundwater potential helps in accurate zones for exploration for

Fig. 11. Drainage density of the study area.


D. Abijith et al. / HydroResearch 3 (2020) 1–14 11

Fig. 12. Rainfall of the study area.

agricultural, industrialization and urban uses. AHP and MIF estimated plot analysis (Fig. 15) indicate that the Area under the curve (AUC) was
results are in Table 5. The results state that the percentage of the region 75% for AHP and 71% for MIF in the groundwater potential map. The ap-
under very low and very high is 9.9%, 10.54% and 20.10%, 17.68% for AHP proach adopted in this research can, therefore, assumed fairly accurate
(Fig. 13) and MIF (Fig. 14) respectively. Very high GWPZ zones are iden- and consistent in predicting groundwater potential.
tified in the central region of the study area along with the river courses.
Geographically it covers an area of 164.5 km2 identified using AHP 7. Conclusion
method and 144.6 km2 for MIF method. This variation may be attributed
to weight averaging in AHP and MIF methods. Also, geomorphology of The current study addresses the identification of groundwater po-
these region contains Pediplain with shallow weathered rocks, which tential zonation in Ponnaniyaru watershed. The main objective of this
has more ability to infiltration. The very low potential zones are identi- analysis is to explore the applicability of AHP, MIF, GIS and RS tech-
fied in the western part of the regions where steep slope and hard rocks niques to map groundwater potential. The crucial parameters were con-
terrains are presented. Also, the presence of lineaments and drainage sidered including geomorphology, drainage density, soil, slope,
formations are less, which controls the GWP and geographically it com- lineament density, geology, land use/land cover and rainfall. High linea-
prises 81 km2 for the AHP method and 86.2 km2 for the MIF methods. ment, drainage density regions were also associated with good ground-
water potential. The hard-rock terrain in the northern part of the study
6. Validation of results area was led to the generally low groundwater potential due to
hydrogeological conditions of the subsurface. Verification of the
The Receiver Operating Characteristic curve (ROC) is a mathematical groundwater potential map using field data indicates that this approach
technique developed to explain the efficiency of probabilistic determin- of prediction is effective and reliable. The results indicate for AHP and
istic detection and prediction systems (Bui et al., 2019; Chepchumba MIF 9.9%, 10.54% (very low), 12.9%, 11.53% (Low), 27.3%, 28.76% (mod-
et al., 2019; Manap et al., 2014; Rajaveni et al., 2017). In the validation, erate), 29.5%, 31.46% (high) and 20.10%, 17.68% (very high). Finally, the
56 observed wells are taken whose post-monsoon water level is ROC curve was ready to test the precision of the AHP and MIF model.
analysed with resultant groundwater potential maps obtained from Validation of results showed that the AHP has a relatively strong predic-
AHP and MIF. The water level fluctuation ranges from 10 to 20 m3/h. tive precision of 75% and MIF has 71%. Based on the assignment of
The region below the ROC curve is between 0 and 1. A larger area weights and ranks the output results can also vary. The MIF approach
under the ROC curve refers to the higher efficiency of spatial modelling has produced acceptable results but this approach has a significant im-
models such as groundwater mapping potential. The findings of the ROC pact on the accuracy with 22% lower accuracy comparing AHP.
This research would encourage the effective production and protec-
tion of groundwater supplies in the field of research. This groundwater
Table 5 potential map can be utilized for groundwater resources management
Estimated area of groundwater potential from AHP and MIF. in Ponnaniyaru watershed and can help policymakers to formulate bet-
Class Very Low Low Moderate Good Very Good ter planning. The assessment of groundwater potential would support
decision-makers in groundwater management and the selection of ap-
(Km2)
propriate locations for drilling wells based on demand. Furthermore,
AHP 81.04 106.24 224.06 242.08 164.47 augmentation of available water resources can be done in order to
MIF 86.22 94.38 257.39 257.39 144.64
maintain the sustainability of the aquifer management.
12 D. Abijith et al. / HydroResearch 3 (2020) 1–14

Fig. 13. Groundwater potential map using AHP.

CRediT authorship contribution statement Acknowledgement

Devanantham Abijith: Writing - original draft, Formal analysis. The author is thankful to NITT/MHRD for financially support ex-
Subbarayan Saravanan: Conceptualization, Methodology, Supervision. tended to the Ph.D. scholars. This research was also possible with the
Leelambar Singh: Software, Data curation. Jesudasan Jacinth Jennifer: use of publicly available datasets, including SRTM DEM; Landsat 8 pro-
Visualization, Investigation. Thiyagarajan Saranya: Software. K.S.S. vided by the United States Geological Survey (USGS), from their Web
Parthasarathy: Writing - review & editing. site at https://earthexplorer.usgs.gov/. Special thanks for Rainfall data

Fig. 14. Groundwater potential map using MIF.


D. Abijith et al. / HydroResearch 3 (2020) 1–14 13

Etikala, B., Golla, V., Li, P., Renati, S., 2019. Deciphering groundwater potential zones using
MIF technique and GIS: a study from Tirupati Area, Chittoor District, Andhra Pradesh,
India. HydroResearch 1, 1–7.
Fienen, Michael N., Arshad, Muhammad, 2016. The International Scale of the Groundwa-
ter Issue. Concepts, Approaches and Challenges, Springer International Publishing, In
Integrated Groundwater Management, pp. 21–48.
Garg, N.K., Hassan, Q., 2007. Alarming scarcity of water in India. Curr. Sci. 93, 932–941.
https://www.jstor.org/stable/24099258.
Ghosh, P.K., Bandyopadhyay, S., Jana, N.C., 2016. Mapping of groundwater potential zones
in hard rock terrain using geoinformatics: a case of Kumari watershed in Western
part of West Bengal. Modeling Earth Systems and Environment 2 (1), 1–12.
Guru, B., Seshan, K., Bera, S., 2017. Frequency ratio model for groundwater potential map-
ping and its sustainable management in Cold Desert, India. Journal of King Saud Uni-
versity - Science 29 (3), 333–347.
Holt Richard. 2018. "Which Cities Will Be Leading the Global Economy in 2035? Over-
view: Cities to Lead National Growth Rates". https://workplaceinsight.net/wp-con-
tent/uploads/2018/12/Global-Cities-Dec-2018.pdf. (Accessed 02 November 2019).
Kulkarni, H., Aslekar, U., & Patil, S. 2018. "Groundwater management in India: status, chal-
lenges and a framework for responses". In Groundwater of South Asia (pp. 615-642).
Springer, Singapore.
Lee, S., Kim, Y.S., Hyun Joo, O., 2012. Application of a weights-of-evidence method and GIS
to regional groundwater productivity potential mapping. J. Environ. Manag. 96 (1),
91–105.
Lee, S., Hyun, Y., Lee, M.-J., 2019. Groundwater potential mapping using data mining
models of big data analysis in Goyang-Si, South Korea. Sustainability 11 (6), 1–21.
Fig. 15. ROC of groundwater potential map using AHP and MIF of Ponnaniyaru watershed. Maheswaran, S.U., et al., 2019. Lithology and structural mapping of Kadavur Basin, Tamil
Nadu, India, using IRS P6 LISS III satellite data. Journal of the Indian Society of Remote
Sensing 47 (8), 1275–1286.
provided by Regional Meteorological Centre of India Meteorological De- Manap, M.A., et al., 2014. Application of probabilistic-based frequency ratio model in
partment (IMD) from their Web site at http://imdchennai.gov.in/ and groundwater potential mapping using remote sensing data and GIS. Arab. J. Geosci.
7 (2), 711–724.
Geology data provided by Geological Survey of India from their Web site Mogaji, K.A., Omosuyi, G.O., Adelusi, A.O., Lim, H.S., 2016. Application of GIS-based eviden-
at https://www.gsi.gov.in. tial belief function model to regional groundwater recharge potential zones mapping
in hardrock geologic terrain. Environmental Processes 3 (1), 93–123.
Msaddek, M.H., et al., 2019. Groundwater potentiality assessment in an arid zone using a
References
statistical approach and multi-criteria evaluation, southwestern Tunisia. Geological
Quarterly 63 (1), 3–15.
Aggarwal, M., Subbarayan, S., Jacinth Jennifer, J., Abijith, D., 2019. Delineation of ground-
Muralitharan, J., Palanivel, K., 2015. Groundwater targeting using remote sensing, geo-
water potential zones for hard rock region in Karnataka using AHP and GIS. In
graphical information system and analytical hierarchy process method in hard rock
315–317.
aquifer system, Karur District, Tamil Nadu, India. Earth Sci. Inf. 8 (4), 827–842.
Al-Abadi, A.M., 2017. The application of Dempster–Shafer theory of evidence for assessing
Naghibi, S.A., Pourghasemi, H.R., Pourtaghi, Z.S., Rezaei, A., 2015. Groundwater Qanat po-
groundwater vulnerability at Galal Badra Basin, Wasit Governorate, East of Iraq. Appl
tential mapping using frequency ratio and Shannon’s entropy models in the Moghan
Water Sci 7 (4), 1725–1740.
watershed, Iran. Earth Sci. Inf. 8 (1), 171–186.
Alizamir, M., Kisi, O., Zounemat-Kermani, M., 2018. Modelling long-term groundwater
Nampak, H., Pradhan, B., Manap, M.A., 2014. Application of GIS based data driven eviden-
fluctuations by extreme learning machine using hydro-climatic data. Hydrol. Sci. J.
tial belief function model to predict groundwater potential zonation. J. Hydrol. 513,
63 (1), 63–73.
283–300.
Althuwaynee, O.F., Pradhan, B., Park, H.J., Lee, J.H., 2014. A novel ensemble bivariate statis-
Nasir, M.J., Khan, S., Zahid, H., Khan, A., 2018. Delineation of groundwater potential zones
tical evidential belief function with knowledge-based analytical hierarchy process
using GIS and multi influence factor (MIF) techniques: a study of district swat,
and multivariate statistical logistic regression for landslide susceptibility mapping.
Khyber Pakhtunkhwa, Pakistan. Environ. Earth Sci. 77 (10), 1–11.
Catena 114, 21–36.
Parthasarathy, K.S.S., Deka, P.C., 2019. Remote sensing and GIS application in assessment
Anbarasu, S., Brindha, K., Elango, L., 2019. Multi-influencing factor method for delineation
of coastal vulnerability and shoreline changes: a review. ISH Journal of Hydraulic En-
of groundwater potential zones using remote sensing and GIS techniques in the
gineering 0 (0), 1–13.
western part of Perambalur District, Southern India. Earth Sci. Inf. 1–16.
Pourghasemi, H.R., 2016. GIS-based forest fire susceptibility mapping in Iran: a compari-
Arulbalaji, P., Padmalal, D., Sreelash, K., 2019. GIS and AHP techniques based delineation
son between evidential belief function and binary logistic regression models. Scand.
of groundwater potential zones: a case study from southern Western Ghats, India.
J. For. Res. 31 (1), 80–98.
Sci. Rep. 9 (1), 1–17.
Pradhan, B., 2009. Groundwater potential zonation for basaltic watersheds using satellite
Breiman, L., 2001. Random forests. Mach. Learn. 45 (1), 5–32.
remote sensing data and GIS techniques. Central European Journal of Geosciences 1
Bui, D.T., et al., 2019. A hybrid computational intelligence approach to groundwater
(1), 120–129.
spring potential mapping. Water (Switzerland) 11 (10), 1–30.
Rajaveni, S.P., Brindha, K., Elango, L., 2017. Geological and geomorphological controls on
Carranza, E.J.M., Hale, M., 2002. Mineral imaging with Landsat thermatic mapper data for
groundwater occurrence in a hard rock region. Appl Water Sci 7 (3), 1377–1389.
hydrothermal alteration mapping in heavily vegetated Terrane. Int. J. Remote Sens.
Ray, B., Shaw, R., 2019. Water insecurity in Asian cities. Urban Drought. Springer,
23 (22), 4827–4852.
Singapore, pp. 17–32. https://doi.org/10.1007/978-981-10-8947-3_2.
Central Ground Water Board (CGWB) 2008. "District groundwater brochure,
Razandi, Y., Pourghasemi, H.R., Neisani, N.S., Rahmati, O., 2015. Application of analytical
Tiruchirappalli district, Tamil Nadu, Technical report series". http://cgwb.gov.in/Dis-
hierarchy process, frequency ratio, and certainty factor models for groundwater po-
trict_Profile/TamilNadu/Trichy.pdf. (Accessed 01 November 2019).
tential mapping using GIS. Earth Sci. Inf. 8 (4), 867–883.
CGWB 2013 "Ground water pollution by industrial clusters. Bhu-Jal news, 28 (1–4), Min-
Saaty, Thomas L., 1980. The Analytic Hierarchy Process : Planning, Priority Setting, Re-
istry of Water Resources, river development and ganga rejuvenation, government of
source Allocation. McGraw-Hill International Book Co.
India". (Accessed 28 October 2019).
Sander, P., 2007. Lineaments in groundwater exploration: a review of applications and
Chenini, I., Mammou, A.B., El May, M., 2010. Groundwater recharge zone mapping using
limitations. Hydrogeol. J. 15 (1), 71–74.
GIS-based multi-criteria analysis: a case study in Central Tunisia (Maknassy Basin).
Sandoval, E., et al., 2018. Groundwater recharge assessment in a rural, arid, mid-mountain
Water Resour. Manag. 24 (5), 921–939.
basin in north-central Chile. Hydrol. Sci. J. 63 (13–14), 1873–1889.
Chepchumba, Mary Christine, James M. Raude, and Joseph K. Sang. 2019. “Geospatial De-
Saravanan, S., 2012. Identification of artificial recharge sites in a hard rock terrain using
lineation and Mapping of Groundwater Potential in Embu County, Kenya.” Acque
remote sensing and GIS. International Journal of Earth Sciences and Engineering 5
Sotterranee - Italian Journal of Groundwater (Jica 2012): 39–51.
(6), 1590–1598.
Chezgi, J., Pourghasemi, H.R., Naghibi, S.A., Moradi, H.R., Kheirkhah Zarkesh, M., 2016. As-
Saravanan, S., Jacinth, J.J., Singh, L., Saranya, T., Sivaranjani, S., 2019. “Impact of land-use
sessment of a spatial multi-criteria evaluation to site selection underground dams in
change on soil erosion in the coonoor watershed, Nilgiris Mountain range, Tamil
the Alborz Province, Iran. Geocarto International 31 (6), 628–646.
Nadu, India”. In Advances in Remote Sensing and Geo Informatics Applications (pp.
Dempster, A.P., 1967. Upper and lower probabilities induced by a multivalued mapping.
109-111). Springer. Cham. https://doi.org/10.1007/978-3-030-01440-7_26.
Ann. Math. Stat. 38 (2), 325–339.
Schapire, Robert E. 2003. “The Boosting Approach to Machine Learning: An Overview.”:
Drobne, S., Lisec, A., 2009. Multi-attribute decision analysis in GIS: weighted linear com-
149–71.
bination and ordered weighted averaging. Informatica (Ljubljana) 33 (4), 459–474.
Şener, Ş., Sener, E., Karagüzel, R., 2011. Solid waste disposal site selection with GIS and
Duan, H., Deng, Z., Deng, F., Wang, D., 2016. Assessment of groundwater potential based
AHP methodology: a case study in Senirkent-Uluborlu (Isparta) Basin, Turkey. Envi-
on multicriteria decision making model and decision tree algorithms. Math. Probl.
ron. Monit. Assess. 173 (1–4), 533–554.
Eng. 1, 1–11.
Shah, T., 2005. Groundwater and human development: challenges and opportunities in
Espinoza, F.P., Minsker, B.S., Goldberg, D.E., 2005. Adaptive hybrid genetic algorithm for
livelihoods and environment. Water Sci. Technol. 51 (8), 27–37.
groundwater remediation design. J. Water Resour. Plan. Manag. 131 (1), 14–24.
14 D. Abijith et al. / HydroResearch 3 (2020) 1–14

Shimpi, S., Rokade, V.M., Upasani, K., 2019. Application of remote sensing and GIS for Taweesin, K., Seeboonruang, U., Saraphirom, P., 2018. The influence of climate variability
groundwater potential zonation: a case study of Bori-Chikli watershed, Maharashtra, effects on groundwater time series in the lower Central Plains of Thailand. Water
India. Bulletin of Pure and Applied Sciences 38 (1), 114–127. (Switzerland) 10 (3).
Siddha, S., Sahu, P., 2018. Assessment of groundwater potential of Gandhinagar region, Thapa, R., Gupta, S., Guin, S., Kaur, H., 2017. Assessment of groundwater potential zones
Gujarat. J. Geol. Soc. India 91 (1), 91–98. using Multi-Influencing Factor (MIF) and GIS: a case study from Birbhum District,
Singh, L., Katpatal, Y.B., 2018. Wetland change analysis and their impact on dense vegeta- West Bengal. Appl Water Sci 7 (7), 4117–4131.
tion by spatial approach. Journal of Urban & Environmental Engineering 12 (1), Umar, Z., et al., 2014. Earthquake induced landslide susceptibility mapping using an inte-
70–76. grated ensemble frequency ratio and logistic regression models in west Sumatera
Singh, S.K., Zeddies, M., Shankar, U., Griffiths, G.A., 2019. Potential groundwater recharge Province, Indonesia. Catena 118 (September 2009), 124–135. https://doi.org/
zones within New Zealand. Geosci. Front. 10 (3), 1065–1072. 10.1016/j.catena.2014.02.005.
Snyder, Shannyn. 2019. “Water In Crisis - Spotlight India.” https://thewaterproject.org/ Venkatesan, G., Pitchaikani, S., Saravanan, S., 2019. Assessment of groundwater vulnera-
water-crisis/water-in-crisis-india (Accessed 11September, 2019). bility using GIS and DRASTIC for upper Palar River basin, Tamil Nadu. J. Geol. Soc.
Stephenson, David A., 1979. Rockfill in Hydraulic Engineering. New York. India 94 (4), 387–394.
Subba Rao, N., Devadas, D.J., Srinivasa Rao, K.V., 2006. Interpretation of groundwater qual- Wang, G., Yang, L., Chen, L., Kubota, J., 2005. Impacts of land use changes on groundwater
ity using principal component analysis from Anantapur District, Andhra Pradesh, resources in the Heihe River basin. Dili Xuebao/Acta Geographica Sinica 60 (3),
India. Environ. Geosci. 13 (4), 239–259. 456–466.
Tang, Y., Zang, C., Wei, Y., Jiang, M., 2019. Data-driven modeling of groundwater level with Yu, H.L., Lin, Y.C., 2015. Analysis of space-time non-stationary patterns of rainfall-
Least-Square support vector machine and spatial–temporal analysis. Geotech. Geol. groundwater interactions by integrating empirical orthogonal function and cross
Eng. 37 (3), 1661–1670. wavelet transform methods. J. Hydrol. 525, 585–597.

You might also like