water

Article
Mapping Potential Water Resource Areas Using GIS-Based
Frequency Ratio and Evidential Belief Function
Yang Li 1 , Mohamed Abdelkareem 2, * and Nasir Al-Arifi 3, *

1 School of Civil Engineering and Geomatics, Southwest Petroleum University, Chengdu 610500, China
2 Geology Department, South Valley University, Qena 83523, Egypt
3 Chair of Natural Hazards and Mineral Resources, Geology and Geophysics Department,
King Saud University, Riyadh 68953, Saudi Arabia
* Correspondence: mohamed.abdelkareem@sci.svu.edu.eg (M.A.); nalarifi@ksu.edu.sa (N.A.-A.);
Tel.: +20-10-0800-1823 (M.A.)

Abstract: Groundwater is a critical freshwater resource that is necessary for sustaining life. Thus,
targeting prospective groundwater zones is crucial for the extraction, use, and management of
water resources. In this study, we combined the remote sensing, GIS-based frequency ratio (FR),
and evidential belief function (EBF) techniques into a model to delineate and quantify prospective
groundwater zones. To accomplish this, we processed Shuttle Radar Topography Mission (SRTM),
Landsat-8 Operational Land Imager (OLI), Sentinel-2, and rainfall data to reveal the geomorphic,
hydrologic, and structural elements and climatic conditions of the study area, which is downstream of
the Yellow River basin, China. We processed, quantified, and combined twelve factors (the elevation,
slope, aspect, drainage density, lineament density, distance to rivers, NDVI, TWI, SPI, TRI, land
use/cover, and rainfall intensity) that control the groundwater infiltration and occurrence using
the GIS-based FR and EBF models to produce groundwater potential zones (GWPZs). We used the
natural breaks classifier to categorize the groundwater likelihood at each location as very low, low,
moderate, high, or very high. The FR model exhibited a better performance than the EBF model, as
evidenced by the area under the curve (AUC) assessment of the groundwater potential predictions
(FR AUCs of 0.707 and 0.734, and EBF AUCs of 0.665 and 0.690). Combining the FR and EBF models
into the FR–EBF model increased the accuracy (AUC = 0.716 and 0.747), and it increased the areas of
very high and moderate potentiality to 1.97% of the entire area, instead of the 0.39 and 0.78% of the
Citation: Li, Y.; Abdelkareem, M.;
FR and EBF models, respectively. The integration of remote sensing and GIS-data-driven techniques
Al-Arifi, N. Mapping Potential Water is crucial for the mapping of groundwater prospective zones.
Resource Areas Using GIS-Based
Frequency Ratio and Evidential Belief Keywords: remote sensing; GIS; groundwater; Yellow River; China
Function. Water 2023, 15, 480.
https://doi.org/10.3390/w15030480

Academic Editor: Akira Kawamura

1. Introduction
Received: 4 December 2022 Water scarcity is a global issue because of the interdependence of water, food, and
Revised: 4 January 2023
energy, as well as the impact that these elements have on human livelihoods and national
Accepted: 13 January 2023
and international economies [1,2]. Water has always been the most crucial natural resource
Published: 25 January 2023
for maintaining life on Earth [3]. Groundwater is one of the most precious natural resources
in every climate zone on the globe and is a prominent water source [4,5]. Moreover, ground-
water is essential for socioeconomic development, even though its quality and quantity
Copyright: © 2023 by the authors.
widely vary, because it meets a variety of human demands, including those for drinking
Licensee MDPI, Basel, Switzerland. water, irrigation, forestry, industrial uses, and animal support [6]. Because groundwater is
This article is an open access article less susceptible to environmental degradation than surface water, we can use it in a more
distributed under the terms and sanitary and reliable manner [7,8], and particularly in areas with polluted water surfaces.
conditions of the Creative Commons The infiltration of river water improves the riparian groundwater capacity for denitrifica-
Attribution (CC BY) license (https:// tion. Therefore, the riverbank infiltration process is widely accepted. For the more than
creativecommons.org/licenses/by/ 2.5 billion people who live on the planet, these fresh water supplies are a daily require-
4.0/). ment [9], and the demand for groundwater will increase due to global population expansion

Water 2023, 15, 480. https://doi.org/10.3390/w15030480 https://www.mdpi.com/journal/water

Water 2023, 15, 480 2 of 24

and its effects on urbanization, industrialization, and agricultural operations, which may
be the main contributing factors to the world’s groundwater scarcity [10]. Moreover, the
increased need for groundwater and ongoing climatic change have inspired the research
community to develop quantitative methods to predict groundwater availability [11,12].
Groundwater is a type of water resource that fills joints, voids, and pore spaces in
the soil found inside geologic formations and layers. The presence and infiltration of
groundwater in rocks depends on the hydraulic conductivity of the lithologic materials,
which is a result of the materials’ porosity, permeability, and fluid flow through geologic
structures [2,13,14]. Researchers describe the use of groundwater mapping as a tool for
water resource development and planning in [15–17]. The topography, geological structure,
landform, drainage pattern, land use/cover, and climate are some of the geoenvironmental
factors that affect groundwater availability [18–20]. The zones with considerable ground-
water capacities that can be predicted and exploited are known as “groundwater potential
zones” (GWPZs) [21–23].
Because groundwater resources are buried beneath strata, we need to use predictive
models to investigate and uncover them [17,24,25]. The groundwater discovery techniques
that are currently used are time-consuming and costly. For groundwater exploration
and prediction, as well as for regional estimations, we can use remote sensing (RS) and
geographic information systems (GISs) [26,27]. We can use the GIS-based knowledge
and/or data-driven techniques to combine and analyze large volumes of geographical data
to predict and discover new water sources [28,29]. Researchers have demonstrated the
value of using RS and GISs to identify possible groundwater resource locations in several
studies [15,25,30,31]. For instance, multi-criteria decision making is a fast and cost-effective
method [17,25,29,32].
Researchers have effectively accomplished the exploration of GWPZ maps in various
environmental contexts using the knowledge-driven and data-driven methodologies. They
always employ GIS technology, which can process vast amounts of spatial data and com-
bine different data types to predict and locate more water supplies. The knowledge-driven
techniques include overlay analyses [16], the analytical hierarchy process (AHP) [27,29,33],
Boolean logic [34], index overlays, and fuzzy methods [35–37]. The data-driven techniques
include, among other, the frequency ratio [38–43], machine learning models [28], logis-
tic regression [38,44], random forests [41,45], the weight-of-evidence method [45–47], the
evidential belief function [48–56], support vector machines [57,58], artificial neural net-
works [46,58], and maximum entropy. We can use these techniques to successfully model
groundwater availability and flood hazards [55,59–61].
Researchers frequently use data-driven models and techniques (e.g., frequency ratio,
evidential belief function, WOE, machine learning models, logistic regression, weights
of evidence, linear regression decision tree analyses, and neural networks) [55,62]. We
used the frequency ratio (FR) technique, which is a data-driven model and bivariate
statistical method, to calculate the spatial relationships between the independent vari-
ables and classes of thematic layers and the dependent variable, which was the existing
wells (spring wells, observation wells), to assign rating (r) values to each class [6,63,64].
In many diverse situations, researchers commonly use the FR to map the groundwater
potential [6,50,62,63,65–68]. For predicting groundwater occurrences, they use several cri-
teria in this technique, including geological, topographic, climatic, and hydrologic data.
Several authors have applied the EBF to GWPZs [28,50,54,69].
In the present study, our aim was to apply the FR and EBF data-driven techniques
using GIS-based approaches for the delineation and identification of prospective groundwater
resource areas through multi-criteria analyses derived from remote sensing data. This tech-
nique was useful for groundwater extraction, management, and prediction, as well as for the
reliability assessments of the combined FR–EBF model compared with the individual models.
Water 2023, 15, x FOR PEER REVIEW 3 of 26

Water 2023, 15, 480

diction, as well as for the reliability assessments of the combined FR–EBF model com-
3 of 24
pared with the individual models.

2. Study Area
2. Study Area
The study area is situated in Shandong Province’s lower sections of the Yellow Riv-
er, andTheitstudy
bordersareathe
is situated
Bohai Sea in Shandong
to the eastProvince’s
(Figure 1), lower sections
covering of the
about Yellow
7438 km2.River,
The
and it borders the Bohai Sea to the east (Figure 1), covering about 7438 km 2 . The muddy
muddy waters of the Yellow River give the area its name. In 1958, 2.1 billion tons of silt
waters
were of the Yellow
dumped into theRiver
ocean, give
andthe area year
every its name.
since, In 1958, 30
around 2.1km
billion tons land
2 of new of silt were
forms.
dumped into the ocean, and every year since, around 30 km2 of new land forms. This
This region experiences a warm temperate semi-humid monsoon climate. The annual
region experiences a warm temperate semi-humid monsoon climate. The annual average
average rainfall is between 530 and 630 mm, which is considerably less than the yearly
rainfall is between 530 and 630 mm, which is considerably less than the yearly average
average evaporation of between 1470 and 2246 mm [70,71]. In western Shandong, the
evaporation of between 1470 and 2246 mm [70,71]. In western Shandong, the Yellow
Yellow River flows along a levee that is higher than the surrounding landscape. After
River flows along a levee that is higher than the surrounding landscape. After traversing
traversing various regions of the province, it drains into the sea around Shandong’s
various regions of the province, it drains into the sea around Shandong’s northern shore.
northern shore. The topography of the area is characterized by steep hills in the south-
The topography of the area is characterized by steep hills in the southwest and level
west and level terrain in the northeast. The alluvial marine plains in the coastal locations
terrain in the northeast. The alluvial marine plains in the coastal locations have a ground
have a ground elevation of only 2–5 m, compared with the northern Shandong hills,
elevation of only 2–5 m, compared with the northern Shandong hills, with elevations
with elevations around 10–30 m [48]. The average temperature ranges from 5 to 1 °C
around 10–30 m [48]. The average temperature ranges from 5 to 1 ◦ C (from 23 to 34 ◦ F) in
(from 23 to 34 °F) in January ◦ and from 24 ◦to 28 °C (from 75 to 82 °F) in July. The two
January and from 24 to 28 C (from 75 to 82 F) in July. The two main methods for shallow
main methodsrecharging
groundwater for shallow aregroundwater
precipitation recharging are precipitation
and river infiltration. and riversediment
The Quaternary infiltra-
tion. The Quaternary
thickness in the currentsediment
study areathickness in 350
is about the m,
current
and thestudy area is marine
Holocene about 350andm, and
deltaic
the Holocene marine and deltaic
sediment thickness is about 26 m [71]. sediment thickness is about 26 m [71].

Figure1.
Figure (a)Location
1.(a) Locationmap
mapof ofChina,
China,(b)
(b)location
locationmaps
mapsofofstudy
studyarea
areaininred
redpolygon,
polygon,(c)
(c)study
studyarea
area
that
thatoverlain
overlainbybywell
well locations
locations in
in red
red color
color.

3.Data
3. DataUsed
Usedand
andMethods
Methods
In this study, we integrated remote sensing data from a variety of sensors with topo-
graphic, hydrologic, and meteorological data to depict the possible water resource locations.
We used GIS techniques to merge several thematic maps (elevation, slope, aspect, terrain
 pect, terrain roughness index, SPI, lineaments, drainage density, distance to river, TWI,
NDVI, LU/LC, and rainfall intensity) produced from these data. These twelve factors are
important contributors to groundwater occurrence that represent the climatic, hydro-
logic, land-cover, and topographic factors. We describe these factors in Section 4. We
present the identification approaches of the FR and EBF models used to map the
Water 2023, 15, 480 groundwater potential zones in Figure 2. 4 of 24

We initiated the digital elevation models (DEMs) using SRTM (Shuttle Radar To-
pography Mission) data (90 m cell size). We used the Shuttle Radar Topographic Mis-
sion SRTMindex,
roughness 30 m SPI,
resolution NASADEM
lineaments, drainage1 arc second
density, WGS84
distance to data
river,to examine
TWI, NDVI,the topo-
LU/LC,
and rainfall
graphic intensity)
changes produced
(NASADEM 1 from these data.
arc second WGS84).These twelve
The DEMs factors are important
allowed con-
for the extrac-
tributors
tion of theto groundwater
slope, aspect,occurrence
and TRI. Wethat produced
represent the
theclimatic,
stream hydrologic,
network usingland-cover,
the 8D and
ap-
topographic
proach [72], factors.
and we Weuseddescribe
a GIS these factorstoinproduce
application Section 4. Westream
the present the identification
density map, TWI,
approaches
SPI, of the FR
and distance and EBF
to rivers. We models used the
calculated to map the groundwater
distances potential
to rivers using zones in
the Euclidean
Figure 2.
distance spatial tool in ArcGIS.

Figure
Figure 2.
2. Data
Data and
and methods
methods used
used in
in present
present study.
study.

Landsat
We initiated 8, which was elevation
the digital deployedmodels
on 11 February
(DEMs) using2013,SRTM
carries the operational
(Shuttle land
Radar Topogra-
imager
phy (OLI) data
Mission) and thermal
(90 m cellinfrared
size). Wesensor
used(TIRS). In the
the Shuttle mapping
Radar of the land
Topographic cover
Mission and
SRTM
30 m the
use, resolution NASADEM 1 arc
visible-near-infrared andsecond WGS84 data to examine
shortwave-infrared wavelength the topographic changes
ranges are used. A
(NASADEM
Landsat-8 OLI 1 arc second
scene (pathWGS84).
121/row The DEMs
034) that allowed for the on
was acquired extraction of the slope,
30 December aspect,
2021 under-
and TRI.
went We alterations
image produced the andstream network procedures.
enhancement using the 8DWe approach
stacked[72], and2,we
bands 3, used a GIS
4, 5, and 7
application to produce the stream density map, TWI, SPI, and distance
to a 30 m spatial resolution and projected them to WGS-84, UTM Zone 50 N. We used to rivers. We calculated
the distances
data from to Landsat
rivers using
8 tothereveal
Euclidean distance spatial
the vegetation and tool
waterin ArcGIS.
resource signatures. We
Landsat
computed the8,normalized
which was difference
deployed vegetation
on 11 February
index2013,
(NDVI)carries the operational
to reveal land
the vegetation
imager[25
areas (OLI)
], andandwethermal infrared
estimated themsensor (TIRS). In
by applying thethe mapping
visible of thebands
infrared land cover
(NDVI and use,
= NIR
the visible-near-infrared and shortwave-infrared wavelength ranges are
(band 5)-R (band 4) and NIR (band 5) +R (band 4)). In addition to Landsat OLI, we used a used. A Landsat-8
OLI scene (path 121/row 034) that was acquired on 30 December 2021 underwent image
alterations and enhancement procedures. We stacked bands 2, 3, 4, 5, and 7 to a 30 m
spatial resolution and projected them to WGS-84, UTM Zone 50 N. We used the data
from Landsat 8 to reveal the vegetation and water resource signatures. We computed the
normalized difference vegetation index (NDVI) to reveal the vegetation areas [25], and we
estimated them by applying the visible infrared bands (NDVI = NIR (band 5) − R (band 4)
and NIR (band 5) + R (band 4)). In addition to Landsat OLI, we used a Landsat-5 scene
(path 121/row 034) that was acquired on 24 August 1992 (bands 7, 4, and 2 in R, G, and B,
respectively) in the change detection in the present study (Table 1).
Water 2023, 15, 480 5 of 24

Table 1. Data used in present study.

No Type of Data Source Date Resolution

bands 2, 3, 4, 5, 6, and
1 Landsat-8 OLI USGS 30 December 2021
7 (30 m)
bands 2, 3, 4, 8,
2 Sentinel-1 ESA/Copernicus 18 October 2022 (“10” m, 11, and 12
(“20” m)
3 SRTM DEM USGS 11–22 February 2000 C-band (30 m)
Climatic Research 1 January 2011– 0.5 degrees in latitude
4 crudata.uea.ac.uk
Unit data 1 January 2020 and longitudes

The Sentinel-2A satellite was launched on 23 June 2015, and the first data were collected
a few days later. The Sentinel-2 sensors collect the VNIR, SWIR, and TIR data. The spatial
resolution of these bands is 10–60 m. Sentinel-2 records a total of 13 bands of the VNIR
and SWIR spectra. The coastal B1 (443 nm) band has a 60 m pixel geometry; however,
the VNIR blue B2 (490 nm), green B3 (560 nm), red B4 (665 nm), and infrared B8 (842 nm)
bands have 10 m pixels. The SWIR bands (B11: 1610 nm; B12: 2190 nm) have 20 m wide
pixels. Sentinel-2 satellites have temporal resolutions of 10 and 5 days, which makes
them extremely valuable for future investigations. Sentinel-2 scenes are delivered as zip-
compressed files in Sentinel’s own SAFE format. The spectral bands are stored as jpg files
in this SAFE file at three different geometric resolutions (10, 20, and 60 m). The jpg files of
bands B2, B3, B4, and B8, with spatial resolutions of 10 m, and of bands B11 and B12, with
20 m resolutions, are stacked into a single GeoTIFF file of a uniform pixel size (10 m). We
obtained a subset of these data during preprocessing using SNAP software to minimize the
computational time and data. We extracted the LC/LU differences using images from two
dates (3 November 2018 and 18 October 2022).
The Climatic Research Unit obtained the data on the average rainfall with a 0.5◦
resolution. The period covered by the collected average rainfall data spanned from January
2011 to 2020, and we interpolated them using the kriging tool of Arc GIS (https://crudata.
uea.ac.uk/cru/data/hrg/ accessed on 01 November 2022).
In this study, we applied the data-driven application of the EBF and FR methodologies
to obtain more accurate estimates of the potential groundwater areas. We compared the
EBF results with the results of well-known statistical methods, including logistic regression
(LR) and the frequency ratio (FR). The EBF model produces better results in terms of
groundwater [50] and flood hazards [73], and it produces better results over fuzzy logic
models in mapping landslides [74]. Moreover, it produces better results in revealing land
subsidence [75]. We can use the model to analyze the effects of all classes of each factor and
the correlation between each component, which gives it an advantage over other statistical
methods [73]. Additionally, the frequency ratio (FR) is one of the better-known statistical
methods [76], and it is more efficient and dependable than the knowledge-driven AHP
model [77].
We produced the groundwater potential map using a data-driven FR–EBF model
built on a geographic information system (GIS), and we set or resampled all the input
data to a 30 m resolution following the spatial resolution of digital elevation data. In
numerous prediction techniques, researchers frequently employ this kind of multicriterion
decision-making process. The user chooses the relative importance of each observation in
the model, which is based on remote sensing, hydrologic, and geologic data. In the GIS
approach, a raster combination is used in which each layer’s pixel is associated with a
particular geographic location. As a result, the combination process is more suited to the
integration of traits from several datasets into an output layer. Moreover, for each pixel in
the study area, we can calculate the GWPZ indices using the EBF model. We calculated the
likelihood of a well occurrence in each unit cell to infer the spatial relationship between the
Water 2023, 15, 480 6 of 24

wells and each thematic evidentiary layer. To calculate the data-driven EBF application, we
used the number of well cells (with at least one cell) in each evidentiary layer.
The frequency ratio (FR) model, which is used to represent the incidence probability
for a specific attribute, is a straightforward geospatial assessment tool that we can use to
ascertain the final relationship between the GRW potential and its effective factors [62,68].
To determine the frequency ratio and to apply it to the overall recharge, we determined
the recharge occurrence ratio for each subclass of conditioning factors. We calculated the
surface ratio for each class and measured it against the total watershed area in Step 9. We
calculated the FR values for each subclass of GWR potential effective factors based on how
well they correlated with the GWR potential inventory.

4. Results
4.1. Elevation
The elevation is a topographical feature that is used as a surface indication to investi-
gate the groundwater potential. Elevation change can affect the climate, which can change
the rainfall, soil quality, vegetation, land uses, and vegetation types [62]. The elevation is an
important layer that governs the direction of the water flow over the land, and it controls
the groundwater occurrence and recharge potential [78]. We categorized the elevation map
of the study area (Figure 3a) into five zones: (1) 0–3; (2) 3–7; (3) 7–12; (4) 12–19; (5) 19–75,
which covered 29.35, 48.04, 19.55, 2.86, and 0.20% of the entire basin, respectively.

4.2. Slope
The slope is one of the indications of prospective groundwater occurrences, and
the topography and/or slope gradient have a direct impact on how much rainwater is
infiltrated [79]. Additionally, the slope can provide a general indication of the groundwater
movement direction [80]. While a moderate-to-steep slope increases the runoff water,
a low or almost level slope has a high infiltration rate and little runoff, which leads to
good groundwater recharging [81]. We can determine the slope via either digital elevation
models from physical measurements, such as STRM DEMs [82–85], or from survey base
points and topographical contours, both of which allow for the creation of a DEM [63].
We classified the slope angles of the study area, which ranged from 0 to 44 degrees, into
four classes: (1) 0–1; (2) 1–2; (3) 2–3.6; (4) 3.6–44 (Figure 3b), which covered 40.87, 36.36,
19.26, and 3.51% of the entire area, respectively (Table 1). Areas with low slope angles
are groundwater potential areas with high potentiality; however, steep slopes promote
runoff [86].

4.3. Aspect
The infiltration rate is substantially influenced by the slope aspect, which also affects
the GWR, which is a measure of the solar radiation. The slope aspect, which ranges from 0
to 360 degrees in a clockwise rotation, is the slope’s orientation. We created the slope factor
using the DEM in ArcGIS 10.6, and we used the DEM in ArcGIS 10.8 to calculate the aspect.
The aspect measurements included ten categories: (1) flat (or no aspect direction); (2) north;
(3) northeast; (4) east; (5) southeast; (6) south; (7) southwest; (8) west; (9) northwest; (10) north.
The northern hemisphere has more water resources on the slopes that face north and east
than on the slopes that face south and west. The slopes of the mountains to the east and north
receive less sunshine than the slopes to the south and west [87]. Despite the high soil moisture
levels on the slopes that face north and east, the transpiration is modest, and vegetation
has increased on the northern and eastern faces as a result. An increase in vegetation can
improve the surface infiltration and groundwater recharge in some areas [26,87]. We divided
the aspect map (Figure 3c) into ten classes: (1) flat (−1); (2) N (0–22.5); (3) NE (22.5–67.5); (4) E
(67.5–112.5); (5) SE (112.5–157.5); (6) S (157.5–202.5); (7) SW (202.5–247.5); (8) W (247.5–292.5);
(9) NW (292.5–337.5); (10) north (337.5–360), which covered 8.77, 9.68, 8.74, 8.89, 10.52, 8.24,
9.91, 9.49, 9.91, and 15.84% of the entire area, respectively.
 assess landscape heterogeneities, and we can use it to identify groundwater [4,88]:

𝑇𝑅𝐼 = (𝑚𝑎𝑥 − 𝑚𝑖𝑛 ) (1)

We classified the TRI result map of the study area, which ranged from 0 to 0.89, into
five classes: (1) 0.00–0.11; (2) 0.11–0.31; (3) 0.32–0.47; (4) 0.48–0.62; (5) 0.63–0.89 (Figure
Water 2023, 15, 480 7 of 24
3d), which covered 6.12, 8.84, 31.30, 38.06, and 15.68% of the entire area, respectively
(Table 1).

Figure 3.
Figure (a) Elevation;
3. (a) Elevation; (b)
(b) slope;
slope; (c)
(c) slope
slope aspects;
aspects; (d)
(d)TRI.
TRI.
4.4. TRI
The topography roughness index (TRI), which is a geomorphometric statistic that
we used to describe and quantify the spatial distributions of the hills and valleys in the
research area, also has an impact on the groundwater potential. The TRI was created to
assess landscape heterogeneities, and we can use it to identify groundwater [4,88]:
q
TRI = (max2 − min2 ) (1)

We classified the TRI result map of the study area, which ranged from 0 to 0.89, into
five classes: (1) 0.00–0.11; (2) 0.11–0.31; (3) 0.32–0.47; (4) 0.48–0.62; (5) 0.63–0.89 (Figure 3d),
which covered 6.12, 8.84, 31.30, 38.06, and 15.68% of the entire area, respectively (Table 1).

4.5. Drainage Density (Dd)

Drainage is a crucial hydrogeological control mechanism. Surface and subsurface
structures are reflections of drainage patterns [89,90], while the relationship between the
Water 2023, 15, 480 8 of 24

permeability and drainage density is inverse. The type of vegetation, the soil’s ability to
absorb rainwater, infiltration, and the slope gradient are all factors that influence an area’s
drainage system. The bedrock type and structure also have impacts [91]. When there is a
high drainage density in a low-permeable-surface region, there is also high precipitation
runoff from that area [89,92]. A high-drainage-density zone contributes to a high amount
of surface runoff with a low groundwater recharge volume [93,94], and the overall length
of the drainage densities correlates with the groundwater recharge volume [95]. Areas
with less drainage density have greater infiltration and less surface runoff, which means
that groundwater development is appropriate in places with low drainage densities [96,97].
Furthermore, because it is used to measure the surface runoff, the drainage density indi-
rectly reveals the groundwater recharge [98]. Low infiltration and higher surface runoff
due to a high Dd result in lower groundwater potential [63,99]. We classified the Dd result
map of the study area, which ranged from 5.82 to 244.6, into five classes: (1) 5.82–53.57;
Water 2023, 15, x FOR PEER REVIEW(2) 53.58–101.3; (3) 101.3–149.1; (4) 149.1–196.8; (5) 196.9–244.6 (Figure 4a,b), which
9 of covered
26

12.26, 33.02, 32.53, 18.48, and 3.71% of the entire area, respectively (Table 1).

(a)Stream
Figure4.4.(a)
Figure Streamnetworks;
networks;(b)(b) drainage
drainage density;
density; (c) TWI;
(c) TWI; (d) SPI.
(d) SPI.

4.8. Distance to River

In hydrogeological research, the distance from hydrographic networks is crucial
because the local alluvial layers are typically found close to river courses, and especially
in semiarid environments [107]. Riverfront locations are ideal for efficient infiltration,
Water 2023, 15, 480 9 of 24

4.6. Topographic Wetness Index (TWI)

We used the TWI map as an indicator of the slope, elevation, and landform effects on
the groundwater resources [100]. In the TWI, the upslope area is an indicator of flowing
water (local slope) or a gauge of the subsurface lateral transmissivity [45,50]. The regional
diversity of the hydrological conditions, including the soil moisture, substantially depends
on the TWI [101]. The extent and zoning of saturated areas are influenced by the TWI,
which, in turn, affects the incidence of springs [44,45]; thus, the higher the TWI, the greater
the groundwater potential [7]. The TWI calculations include a summary of the topographic
roughness, hillslope, and foothill effects on the lateral groundwater flow [102]. Areas with
high TWIs allow us to locate areas with infiltration potential and soil moisture accumulation,
which are unusual in foothills [103,104].

Ac
TW I = ln( ) (2)
tans
We classified the TWI result map of the study area into three classes: (1) from −8.16
to −2.97; (2) from −2.97 to 0.25; (3) from 0.25 to 13.51 (Figure 4c), which covered areas of
56.19, 31.89, and 11.92% of the entire area, respectively (Table 1).

4.7. SPI
The SPI is an important index for mapping groundwater infiltration zones. The SPI is
an indicator of the slope, elevation, and landform effects on the groundwater resources [62].
Using slope and flow accumulation parameters, we computed the SPI using Arc GIS
software version 10.5. The runoff influence increased as the SPI value increased [105,106].
We classified the SPI result map of the study area, which ranged from 0 to 34.002, into three
classes: (1) 0–0.001; (2) 0.001–0.1; (3) 0.1–34.002 (Figure 4d), which covered areas of 94.67,
4.79, and 0.54% of the entire area, respectively (Table 1).

4.8. Distance to River

In hydrogeological research, the distance from hydrographic networks is crucial
because the local alluvial layers are typically found close to river courses, and especially in
semiarid environments [107]. Riverfront locations are ideal for efficient infiltration, and
thus, for groundwater recharge [44,55]. The groundwater potential is stronger in areas near
rivers, and particularly those with continuous or extended flows [43,59,108]. However, in
places farther than 600 m, it is challenging to locate the alluvial layers [4]. Rivers contribute
to the GWR, and hence, affect its potential in watersheds. To initiate the distance categories,
we utilized the Euclidean distance tool from the spatial analyst tools in Arc GIS 10.6 [62].
We classified the map of the distance to river results of the study area, which ranged from 0
to 0.019, into five classes: (1) 0.00–0.0039; (2) 0.004–0.0078; (3) 0.0079–0.012; (4) 0.013–0.016;
(5) 0.017–0.019 (Figure 5a), which covered 28.96, 34.66, 21.77, 31.12, and 1.48% of the entire
area, respectively (Table 1).

4.9. Lineaments
Linear earth features that result from geologic structures, which are known as lin-
eaments, serve as overall surface depictions of the subsurface cracks [109]. Geological
features with linear or curved shapes play a major role in the occurrence and transport
of groundwater over crystalline terrain. Lineaments, such as cracks, fissures, and joints,
cause the infiltration of surface runoff and the replenishment of hard-rock aquifers, and
they often form because of tectonic stress/strain. Numerous academics have highlighted
the connection between lineaments and the presence of groundwater to emphasize how
the lineament density affects well yields [110–112]. Lineaments fall under the category of
secondary porosity, and we can distinguish them from other terrain characteristics on satel-
lite imagery by tonal differences. A lineament can be a fault, fracture, or master joint, or a
lengthy linear geological structure, topographic linearity, or a stream’s straight course [113].
Lineaments have a substantial impact on the groundwater storage and flow and surface
 groundwater likelihood in a given land area increases with the rainfall [124]. Research-
ers can track, observe, and measure the analyzed watershed’s downpour rates using
rainfall data from the TRMM satellite. Due to the substantial precipitation and subse-
quent flooding from storms, the infrastructure in the area has been severely damaged,
Water 2023, 15, 480 and there have been losses in the provision of water resources. The identification of 10 are-
of 24
as that are prone to water accumulation may be possible using rainfall information be-
cause precipitation is crucial for recharging groundwater [125]. We classified the mean
annual rainfall intensity (2011–2020) map of the study area (Figure 5f) into five zones: (1)
runoff infiltration into the subsurface [114]. We classified the lineament density map of the
567.4–575.8; (2) 575.9–584.2; (3) 584.3–592.5; (4) 592.6–600.9; (5) 601–609.3, which covered
study area (Figure 5b,c) into five zones: (1) 0–12; (2) 12–24; (3) 24–35; (4) 35–47; (5) 48–59,
32.05, 38.89, 11.33, 9.31, and 8.42% of the entire basin, respectively. We gave the dense
which covered 34.09, 22.17, 23.50, 14.40, and 5.84% of the entire basin, respectively. We gave
rainfall zones that were deemed potential groundwater recharge zones and that pro-
dense lineament zones that were deemed potential groundwater recharge zones and that
moted well production to a higher rank (“numeric value”).
promoted well production to a higher rank (“numeric value”).

Figure 5. (a) Distance to rivers; (b) lineaments; (c) lineament density; (d) LC/LC; (e) NDVI; (f) rainfall
intensity.

4.10. Land Use/Cover

Land use/cover includes a region’s vegetation cover, aquatic bodies, residential neigh-
borhood distribution, urbanization, and soil deposit types [115]. Land use/cover has a
substantial impact on the groundwater recharge, occurrence, and availability [116,117]. For
instance, agricultural and plantation lands are good locations for groundwater investiga-
tions because trees and plants can collect water on their foliage and allow it to seep into the
earth through their roots and biological processes, which replenishes the water table [118].
However, because of the usage of concrete floors, which only allow for surface drainage,
settlements are poor groundwater areas [85]. Typically, researchers use topographic maps
Water 2023, 15, 480 11 of 24

and a combination of field verification and remotely sensed data to map land-use and
land-cover patterns. False color composites (FCCs) are typically created from satellite
images utilizing a variety of band combinations [84,119].
The recharge in vegetated areas is substantially lower than that in non-vegetated areas,
according to Gee et al. [120]. Furthermore, compared with permanent lands, such as shrub
and forest areas, recharging is higher in agricultural and grassland areas [121]. To identify
the different types of LUs, we used Landsat-8 data. Additionally, using a supervised
classification strategy and the maximum likelihood approach, we classified the LU/LC
result map of the study area into three classes: (1) water, (2) vegetation, and (3) urban
(Figure 5d), which covered 13.19, 20.24, and 46.57% of the entire area, respectively (Table 1).

4.11. NDVI
Researchers frequently use the NDVI for groundwater potential zones [122,123]. Using
a map of the normalized difference vegetation index (NDVI), we depicted the vegetation
density and coverage. We used Envi 5.3 to prepare the NDVI layer. The NDVI has a value
range from −1 to 1 (often known as from 0 to 255). A greater NDVI value indicates thick
vegetation. We classified the NDVI result map of the study area, which ranged from 0 to 255,
into five classes based on the natural breaks method: (1) 0.00–66; (2) 66–131; (3) 131–174;
(4) 174–210; (5) 210–255 (Figure 5e), which covered 8.52, 8.61, 28.13, 39.45, and 15.28% of
the entire area, respectively (Table 1).

4.12. Rainfall
The rainfall amount is one of the key elements in the identification of groundwater
potential zones. Rainfall is a hydrologic technique for refilling aquifers [97]. The ground-
water likelihood in a given land area increases with the rainfall [124]. Researchers can
track, observe, and measure the analyzed watershed’s downpour rates using rainfall data
from the TRMM satellite. Due to the substantial precipitation and subsequent flooding
from storms, the infrastructure in the area has been severely damaged, and there have
been losses in the provision of water resources. The identification of areas that are prone
to water accumulation may be possible using rainfall information because precipitation is
crucial for recharging groundwater [125]. We classified the mean annual rainfall intensity
(2011–2020) map of the study area (Figure 5f) into five zones: (1) 567.4–575.8; (2) 575.9–584.2;
(3) 584.3–592.5; (4) 592.6–600.9; (5) 601–609.3, which covered 32.05, 38.89, 11.33, 9.31, and
8.42% of the entire basin, respectively. We gave the dense rainfall zones that were deemed
potential groundwater recharge zones and that promoted well production to a higher rank
(“numeric value”).

5. Results
5.1. Application of FR Model
Researchers use the FR to describe the likelihood that a phenomenon possesses a
particular characteristic [28]. We entered the calibration well locations (the dependent
variable) and conditioning factors (such as the lithology, slope, distance to streams, and
distance to faults) using the FR approach [126]. We calculated the well occurrence chance
in each class for all the parameters using the FR [28]. We used the following equation to
determine each class’s impact on the independent variable:

( Pt /T pt )
Fr = (3)
( Nc /TNc )

where FR is the rate at which each class of each parameter impacts the parameter; Pt denotes
the proportion of points in each class; Tpt is the total number of points across all classes; NC
is the number of classes; TNC is the total number of pixels. The FR value produced for each
class of conditioning factors determines the weight of each class of conditioning factors in
the subject layers for mapping the groundwater potential. After applying the FR to each
class, the new thematic layer is the input value for the hybrid model.
Water 2023, 15, 480 12 of 24

We identified the spatial correlations between the well locations (a proxy for the ground-
water potential) and conditioning factors using the FR model (Table 1). According to Lee and
Pradhan [127], FR values below 1 indicate low correlations, and those above 1 indicate strong
correlations. The elevation classes from 19 to 75 had the highest FRs (9.06), followed by the
elevation classes from 12 to 19 (3.81), from 7 to 12 (1.48), from 3 to 7 (0.87), and from 0 to 3
(0.55). Because this range indicates a flat plateau, the areas with high elevations (between 19
and 75) are more likely to retain groundwater. The zones with slopes from 0 to 1 had high FRs
(1.157), as did the E, NE, Flat, and SE zones (2.65, 1.45, 1.24, and 1.03, respectively), as the area’s
general slope is due east. Additionally, the areas with higher TRIs (from 0.63 to 0.89) and low
terrains (from 0.11 to 0.31) had FRs of 1.27 and 1.23, respectively.
In terms of the hydrologic characteristics, the zones with low Dd values (from 53.58
to 101.3) had higher FRs (1.26); however, the areas of high Dd values had the lowest FRs
(Table 1). The groundwater potential probability was the highest for the TWI class from
−2.97 to 0.25 (FR = 1.026), followed by the TWI classes from −8.16 to 2.97 (FR = 1.003) and
from 0.25 to 13.51 (FR = 0.915). The SPI classes from 0.0 to 0.001 (FR = 1.018) and from
0.001 to 0.1 (FR = 0.760) had high probabilities; however, the areas of high SPIs (from 0.1 to
34.002) had no potentiality (FR = 0). Furthermore, the areas nearest to the rivers from 0 to
0.0039 had the highest FRs (1.130), followed by those from 0.004 to 0.0078 (0.997); however,
the areas farther from the rivers had no potentiality (FR = 0).
The lineament density values from 36 to 47 (FR = 1.136) were the highest, followed by
those from 0 to 12 (FR = 1.120); however, the moderate LD areas (from 25 to 35) had the
lowest FR values (FR = 0.851). Using the LU/LC, the groundwater potential probability
was the highest for the vegetation class (FR = 1.08), as well as for areas with high NDVIs
(from 210 to 255) (FR = 1.42); however, the areas from 0 to 66 had low GWPs (FR = 0.427).
In terms of the rainfall classes, the areas that received from 114.42 to 148.13 mm had the
highest precipitation (FR = 1.56), followed by those that received from 148.13 to 244.56 mm
(FR = 1.512) and from 570 to 580 mm (FR = 1.36); however, the areas that received from 580
to 590 mm had the lowest FRs (0.608).

5.2. Application of EBF Model

For the spatial prediction of possible groundwater areas, we used the Dempster–Shafer-
based evidential belief function (EBF) [128–130]. This strategy is a potentially effective way
to address some of the challenges that are associated with the combination of evidence and
data fusion [131]. We can calculate the EBF as follows [132]:

1WCij D
BeICij = (4)
∑m
j = 1WCij D



N Cij ∩ D /N Cij
WCij D =

(5)
N ( D ) − N Cij ∩ D /N ( T ) − N Cij ]
The numerator used to calculate the WCij D is the likelihood that the D (groundwa-
ter) occurs in the presence of Cij (a particular class of one conditioning factor), which
demonstrates that the D more frequently occurs (or is more present) in the Cij than would
be predicted by chance. The denominator used to calculate the WCij D is the conditional
probability that D exists given the lack of Cij , which indicates that D more frequently occurs
outside of Cij than would be predicted by chance. Therefore, the parameter WCij D is the
weight of the Cij in terms of D being more present than absent as anticipated by chance.
Therefore, the relative strength of the WCij D for each jth Cij class of evidence in map Xi is
BeI Cij , as indicated in the equations above. N (Cij∩D) represents the number of wells that
occur in a subclass; N (Cij) is the number of pixels in a subclass; N (D) is the total number
of wells; N (T) is the sum of the pixel domain for a class. Accordingly, for the WCij D , we can
calculate the BeI Cij, as shown in Table 1. The values of the Bel range are between 0 and 1
for each subclass.
Water 2023, 15, 480 13 of 24

The groundwater potential probability was the highest for the elevation classes from
19 to 75 m with the highest EBFs (0.565); however, it was the lowest for the elevation classes
from 0 to 3 m (EBF = 0.029) (Table 2). The zones with the highest slopes (from 3.6 to 44) and
lowest slope angles (from 0 to 1) had the highest EBFs (0.43 and 0.26, respectively). The east,
northeast, and flat zones had the highest EBFs (0.29, 0.14, and 0.11, respectively); however,
the southwest zone had the lowest (EBF = 0.095). Furthermore, the higher TRI areas (from
0.63 to 0.89) had high EBFs (0.256). In the case of the Dd, EBF values = 0.35 were higher
in the classes of low density (from 53.58 to 101.3). We assigned the highest EBF values to
the two TWI classes: from −2.97 to 0.25 (EBF = 0.35) and from −8.16 to −2.97 (EBF = 0.34).
Furthermore, we assigned the SPI classes from 0.0 to 0.001 and from 0.001 to 0.1 EBFs of
0.66 and 0.33, respectively, as well as groundwater probabilities. According to the LD,
the values from 0 to 12 had the highest EBFs (0.24). Using the LU/LC, the groundwater
potential probability was the highest for the vegetation class (EBF = 0.415), as well as for
the areas with high NDVIs (from 210 to 255), with the highest EBF 0.32. The class from
592.5 to 600.9 mm had the highest EBF value (0.285).

Table 2. Groundwater factors of the adopted models in the study area.

No. N N (D) N (D) −N

Domain No. No. N (T)
Topography Pixels in FR (Cij∩D)/N −N (Cij∩D)/N (T) WcijD Bel
% Wells Wells % −N (Cij)
Domain (Cij) (Cij∩D) −N (Cij)
0 to 3 2,883,463 0.29345745 9 0.164 0.558 0.0000031 46 6,942,367 6.62598 × 10−6 0.471 0.029
3 to 7 4,720,324 0.48039952 23 0.418 0.870 0.0000049 32 5,105,506 6.26774 × 10−6 0.777 0.048
7 to 12 1,921,284 0.19553402 16 0.291 1.488 0.0000083 39 7,904,546 4.93387 × 10−6 1.688 0.104
12 to 19 281,047 0.02860288 6 0.109 3.814 0.0000213 49 9,544,783 5.13369 × 10−6 4.159 0.255
19 to 75 19,712 0.00200614 1 0.018 9.063 0.0000507 54 9,806,118 5.50677 × 10−6 9.212 0.565
Slope
0 to 1 4,010,495 0.40868454 26 0.473 1.157 0.0000065 29 5,802,685 4.99769 × 10−6 1.297 0.260
1 to 2 3,568,474 0.36364094 16 0.291 0.800 0.0000045 39 6,244,706 6.24529 × 10−6 0.718 0.144
2 to 3.6 1,889,910 0.19258895 9 0.164 0.850 0.0000048 46 7,923,270 5.80568 × 10−6 0.820 0.164
3.6 to 44 344,301 0.03508557 4 0.073 2.073 0.0000116 51 9,468,879 5.38607 × 10−6 2.157 0.432
Slope ASPECT
Flat 860,817 0.08772049 6 0.109 1.244 0.0000070 49 8,952,363 5.47342 × 10−6 1.273 0.117
North 950,349 0.09684414 4 0.073 0.751 0.0000042 51 8,862,831 5.75437 × 10−6 0.731 0.067
Northeast 857,199 0.08735181 7 0.127 1.457 0.0000082 48 8,955,981 5.35955 × 10−6 1.524 0.140
East 872,576 0.08891878 13 0.236 2.658 0.0000149 42 8,940,604 4.69767 × 10−6 3.171 0.291
Southeast 1,032,724 0.10523847 6 0.109 1.037 0.0000058 49 8,780,456 5.58058 × 10−6 1.041 0.095
South 808,454 0.08238451 3 0.055 0.662 0.0000037 52 9,004,726 5.77475 × 10−6 0.643 0.059
Southwest 972,047 0.09905525 2 0.036 0.367 0.0000021 53 8,841,133 5.99471 × 10−6 0.343 0.031
West 931,399 0.09491307 4 0.073 0.766 0.0000043 51 8,881,781 5.74209 × 10−6 0.748 0.069
Northwest 972,718 0.09912363 5 0.091 0.917 0.0000051 50 8,840,462 5.65581 × 10−6 0.909 0.083
North 1,554,897 0.15844986 5 0.091 0.574 0.0000032 50 8,258,283 6.05453 × 10−6 0.531 0.049
Terrain Roughness Index (TRI)
0.0 to 0.11 602,302 0.06124132 3 0.055 0.891 0.0000050 52 9,232,594 5.63222 × 10−6 0.884 0.169
0.11 to 0.31 869,074 0.08836636 6 0.109 1.235 0.0000069 49 8,965,822 5.4652 × 10−6 1.263 0.241
0.32 to 0.47 3,078,693 0.31303768 15 0.273 0.871 0.0000049 40 6,756,203 5.92049 × 10−6 0.823 0.157
0.48 to0.62 3,743,124 0.3805962 20 0.364 0.955 0.0000053 35 6,091,772 5.74545 × 10−6 0.930 0.177
0.63 to 0.89 1,541,703 0.15675844 11 0.200 1.276 0.0000071 44 8,293,193 5.30556 × 10−6 1.345 0.256
Water 2023, 15, 480 14 of 24

Table 2. Cont.

No. N N (D) N (D) −N

Domain No. No. N (T)
Topography Pixels in FR (Cij∩D)/N −N (Cij∩D)/N (T) WcijD Bel
% Wells Wells % −N (Cij)
Domain (Cij) (Cij∩D) −N (Cij)
Drainage Density
5.821–53.57 10,128 0.12264619 6 0.109 0.889 0.0005924 49 72,451 0.000676319 0.876 0.210
53.58–101.3 27,266 0.3301808 23 0.418 1.267 0.0008435 32 55,313 0.000578526 1.458 0.350
101.4–149.1 26,864 0.32531273 16 0.291 0.894 0.0005956 39 55,715 0.000699991 0.851 0.204
149.2–196.8 15,257 0.18475642 10 0.182 0.984 0.0006554 45 67,322 0.000668429 0.981 0.235
196.9–244.6 3064 0.03710386 0 0.000 0.000 0.0000000 55 79,515 0.000691693 0.000 0.000
TWI
−8.16 to
5,514,106 0.56190817 31 0.564 1.003 0.0000056 24 4,299,074 5.5826 × 10−6 1.007 0.341
−2.97
−2.97 to
3,128,986 0.31885546 18 0.327 1.026 0.0000058 37 6,684,194 5.53545 × 10−6 1.039 0.352
−0.25
0.25 to 13.51 1,170,088 0.11923637 6 0.109 0.915 0.0000051 49 8,643,092 5.66927 × 10−6 0.904 0.307
SPI
0 to 0.001 9,290,340 0.94672063 53 0.964 1.018 0.0000057 2 522,840 3.82526 × 10−6 1.491 0.665
0.001 to 0.1 469,594 0.0478534 2 0.036 0.760 0.0000043 53 9,343,586 5.67234 × 10−6 0.751 0.335
0.1 to 34.002 53,246 0.00542597 0 0.000 0.000 0.0000000 55 9,759,934 5.63528 × 10−6 0.000 0.000
Dist to Rivers
0–400 23,750 0.28963768 18 0.327 1.130 0.0007579 37 58,249 0.000635204 1.193 0.294
400–800 28,423 0.34662618 19 0.345 0.997 0.0006685 36 53,576 0.000671943 0.995 0.246
800–1200 17,855 0.21774656 11 0.200 0.918 0.0006161 44 64,144 0.000685957 0.898 0.222
1200–1600 10,759 0.13120892 7 0.127 0.970 0.0006506 48 71,240 0.000673779 0.966 0.238
1600–1900 1212 0.01478067 0 0.000 0.000 0.0000000 55 80,787 0.000680803 0.000 0.000
Lineaments
0–12 3,273,187 0.34086999 21 0.382 1.120 0.0000064 34 6,329,263 5.37187 × 10−6 1.194 0.240
13–24 2,128,686 0.22168155 11 0.200 0.902 0.0000052 44 7,473,764 5.88726 × 10−6 0.878 0.176
25–35 2,256,503 0.23499242 11 0.200 0.851 0.0000049 44 7,345,947 5.9897 × 10−6 0.814 0.163
36–47 1,382,833 0.14400835 9 0.164 1.136 0.0000065 46 8,219,617 5.59637 × 10−6 1.163 0.234
48–59 561,241 0.05844769 3 0.055 0.933 0.0000053 52 9,041,209 5.75144 × 10−6 0.929 0.187
Land cover
Water 1,090,317 0.13192203 4 0.073 0.551 0.0000037 51 7,174,542 7.10847 × 10−6 0.516 0.186
Vegetation 3,325,930 0.40241824 24 0.436 1.084 0.0000072 31 4,938,929 6.27666 × 10−6 1.150 0.415
Urbans 3,848,612 0.46565973 27 0.491 1.054 0.0000070 28 4,416,247 6.34023 × 10−6 1.107 0.399
NDVI
0–66 729,152 0.08520368 2 0.036 0.427 0.0000027 53 7,828,600 6.77005 × 10−6 0.405 0.084
66–131 736,815 0.08609913 4 0.073 0.845 0.0000054 51 7,820,937 6.52096 × 10−6 0.833 0.172
131–174 2,407,709 0.2813483 18 0.327 1.163 0.0000075 37 6,150,043 6.01622 × 10−6 1.243 0.257
174–210 3,376,062 0.39450337 19 0.345 0.876 0.0000056 36 5,181,690 6.94754 × 10−6 0.810 0.167
210–255 1,308,014 0.15284551 12 0.218 1.427 0.0000092 43 7,249,738 5.93125 × 10−6 1.547 0.320
Rainfall
567.4–575.8 2,944,227 0.32047756 24 0.436 1.362 0.0000082 31 6,242,772 4.96574 × 10−6 1.642 0.282
575.9–584.2 3,573,148 0.38893528 13 0.236 0.608 0.0000036 42 5,613,851 7.4815 × 10−6 0.486 0.083
584.2–592.5 1,041,162 0.11332994 3 0.055 0.481 0.0000029 52 8,145,837 6.38363 × 10−6 0.451 0.077
592.5–600.9 855,039 0.09307054 8 0.145 1.563 0.0000094 47 8,331,960 5.64093 × 10−6 1.659 0.285
601–609.3 773,423 0.08418669 7 0.127 1.512 0.0000091 48 8,413,576 5.70507 × 10−6 1.586 0.272
Water 2023, 15, 480 15 of 24

We evaluated the EBF weights (Bel values) for the conditioning factors and their
classes, and according to the results, the trends for the EBF values for the classes were
Water 2023, 15, x FOR PEER REVIEW 15 of 26
correlated as for the FR values (Figure 6), with the exception of a few classes in the Dd,
NDVI, lineaments, and rainfall.

Variationsinincomputed
Figure6.6.Variations
Figure computedclasses
classes
inin
FRFR and
and EBF
EBF models.
models.
Water 2023, 15, 480 16 of 24

5.3. Groundwater Potential Mapping and Validation of Built Models

We classified the combined groundwater potential zone (GWPZ) map using the FRs
(Figure 7a) combined from twelve thematic maps using the natural breaks approach into
five classes: (1) very low (23.67%); (2) low (40.60%); (3) moderate (26.44%); (4) high (8.90%);
(5) very high (0.39%). Furthermore, the GWPZ map using the EBF (Figure 7b) that ranged
from very low to very high was occupied by 23.60, 41.62, 24.44, 9.57, and 0.78%, respectively.
We combined the FR and EBF models to produce the combined FR–EBF model (Figure 7c),
Water 2023, 15, x FOR PEER REVIEW 18 of 26
which we divided into five classes ranging from very low to very high, and covering areas
of 22.89, 39.94, 26.22, 8.98, and 1.97% of the entire area (Table 3). In the combined FR–EBF
model, the areas of the very high and moderate potentiality increased (Figure 8). Based on
model). Due toband
the Landsat-8 the uncertainty
composites 7,around
5, andthe presence
3, most of theofvery
groundwater in the GWPZs
high to extreme results, the
are
combined
occupied bymodel had better
vegetation and accuracy (AUC =(Figure
water resources 0.716). 7e,f).

Figure
Figure 7.
7. GWPZs:
GWPZs: (a)
(a)Fr
Frmodel;
model;(b)
(b)EBF
EBFmodel;
model;(c)
(c)FR–EBF
FR–EBF model;
model; (d–f)
(d–f) subsets
subsets of
of Sentinel-2
Sentinel-2 (2022).
(2022).

Table 3. GWPZ
Table 3. GWPZ areas.
areas.
GWPZ
GWPZ
Area FR %
Area FR %
Area EBF %
Area EBF %
Area EBF+FR %
Area EBF + FR %
Very low 23.67 23.60 22.89
Very low 23.67 23.60 22.89
Low 40.60 41.62 39.94
Low 40.60 41.62 39.94
Moderate 26.44 24.44 26.22
Moderate 26.44 24.44 26.22
High 8.90 9.57 8.98
Highhigh
Very 8.90
0.39 9.57
0.78 8.98
1.97
Very high 0.39 0.78 1.97
Water 2023,
Water 2023, 15,
15, 480
x FOR PEER REVIEW 19
17of 26
of 24

Figure 8. GWPZ areas of three models.

We present the ROC curves for the predicted groundwater map in Figure 8. We can use
6. Discussion
the AUC to describe the system’s capacity to precisely anticipate both the “groundwater”
Water scarcityofhas
and “no-existence now becomewhich
groundwater”, a major global issue,
highlights with substantial
the usefulness implications
of its prediction. The
for the long-term sustainability of water supplies. In the research
AUC ranges from 0 to 1, with values smaller than 0.5 indicating model integrity, and models on spatial groundwater
recharge
with largerpotential zones, researchers
values indicating emphasize
higher accuracy. theAUC
The importance
results of ofthe
themodels,
efficientincluding
planning
and FR,
the management
EBF, and combined of water FR–EBF
resources. According
models, to remote
were 0.707, 0.665, sensing
and 0.716, data analyses,
respectively.
We achieved better accuracy by fusing the two models (the RF–EBF model). Due condi-
GWPZs are controlled by topographic, geomorphic, hydrological, and climatic to the
tions [28,110,133].
uncertainty aroundAccording
the presence to the applied FR and
of groundwater EBFresults,
in the models, thethe rainfall, model
combined NDVI, had lin-
eament density, and
better accuracy (AUC = 0.716). distance to river were positively correlated, and there were more
water well locations in the high-altitude areas, which consist of plateaus that are en-
6. Discussion
riched in vegetation and annual precipitation, as well as surface water, near to the Yel-
low and
Water Zihe Rivershas
scarcity (innow
the become
south). The highest
a major wellissue,
global locations
with were on theimplications
substantial east, north-
east, and flat slope aspects, which is because the northern
for the long-term sustainability of water supplies. In the research on spatial hemisphere and north- and
ground-
east-facing slopes contain more water resources than the
water recharge potential zones, researchers emphasize the importance of the efficient south- and west-facing slopes.
Less sunlight
planning and hits the northern
management and eastern
of water resources.mountain
According slopes than it does
to remote sensing thedata
southern
anal-
and western
yses, GWPZsones. are On the north-
controlled by and east-facinggeomorphic,
topographic, inclines, the transpiration
hydrological,isand lowclimatic
despite
the high soil
conditions moisture levels.
[28,110,133]. The effect
According to theisapplied
vegetation growth
FR and EBFon the northern
models, and eastern
the rainfall, NDVI,
faces. In certain
lineament density, places, the increased
and distance to river vegetation enhancescorrelated,
were positively the groundwaterand there recharge
were moreand
surfacewell
water infiltration
locations[26,87]. The LC/LU types
in the high-altitude areas,have
which a substantial impact on
consist of plateaus thatthearepermeabil-
enriched
ity,vegetation
in evapotranspiration,
and annualand runoff. The
precipitation, as recharge in vegetated
well as surface water, near areas to is
thesubstantially
Yellow and
lowerRivers
Zihe than that(in thein south).
non-vegetated
The highest areas, according
well locationstowere Gee on et the
al. [120]. Furthermore,
east, northeast, the
and flat
slope aspects,
recharging which isinbecause
is higher the northern
agricultural hemisphere
and grassland areasandthan north- and east-facing
on permanent lands,slopes
such
contain
as shrubmore and water
forest resources
areas [121]. than the south- and west-facing slopes. Less sunlight hits the
northern and eastern
The trends of themountain
EBF values slopes
for thethan it does
classes the southern
were related toand the western
FR values, ones. On the
according
north-
to the and east-facing
evaluation of theinclines, the transpiration
EBF weights (Bel values)is lowfordespite the high soil
the conditioning moisture
factors andlevels.
their
The effect is vegetation growth on the northern and eastern
classes (Figure 6). The only exception to the similarity were a few classes in the Dd, faces. In certain places, the
increased vegetation enhances the groundwater recharge
NDVI, lineaments, and rainfall. The applied FR was better validated than the EBF basedand surface infiltration [26,87].
The
on theLC/LU
AUCtypes values have a substantial
of 0.665 and 0.707, impact on the permeability,
respectively. evapotranspiration,
In various studies in the literature,and
runoff.
including The[75],
recharge in vegetated
researchers areas is substantially
have demonstrated that thelowerweights than that in non-vegetated
generated using the FR
areas,
and EBF according
models to areGee et al. [120].
consistent [134].Furthermore,
Furthermore, the recharging
the FR surpasses is higher
the EBF in agricultural
techniques,
and grassland areas than on permanent lands, such as
according to numerous studies [28,50]. Combining the two models (FR–EBF model) shrub and forest areas [121].
The trends
produced betterofresults
the EBF values
than thoseforof theeach
classes were related
individual model, to the
as FR values, by
validated according
the AUC to
the evaluation of the EBF weights (Bel values) for the conditioning
value of 0.716, which was higher than those of the FR and EBF models (Figure 9). More- factors and their classes
(Figure
over, the6).combined
The only FR–EBFexception to theincreased
model similaritythe were
areasa few classes
of very high in perspectivity
the Dd, NDVI, tolinea-
1.97,
ments,
compared andwithrainfall.
the 0.39The and
applied
0.78 FR wasFR
of the better
and EBFvalidated
models, than the EBF based on the AUC
respectively.
values of 0.665 and 0.707, respectively. In various studies in the literature, including [75],
Water 2023, 15, 480 18 of 24

researchers have demonstrated that the weights generated using the FR and EBF models are
consistent [134]. Furthermore, the FR surpasses the EBF techniques, according to numerous
studies [28,50]. Combining the two models (FR–EBF model) produced better results than
those of each individual model, as validated by the AUC value of 0.716, which was higher
than those of the FR and EBF models (Figure 9). Moreover, the combined FR–EBF model
Water 2023, 15, x FOR PEER REVIEW 20 of 26
increased the areas of very high perspectivity to 1.97, compared with the 0.39 and 0.78 of
the FR and EBF models, respectively.

Figure 9. AUCs
Figure AUCsofofpredicted
predictedGWPZ
GWPZmodels:
models:(a) (a)
FR, FR,
(b) EBF, and and
(b) EBF, (c) FR–EBF testing
(c) FR–EBF models;
testing (d–f)
models;
using training samples.
(d–f) using training samples.

The Yellow
The Yellow River
River is
is extensively
extensively contaminated
contaminated by by factory
factory waste
waste and
and sewage
sewage from
from the
the
expanding cities, and hence, it is polluted.
expanding polluted. The main sources of pollution are industrial,
agricultural, and domestic sewage, along with industrial waste gas [135]. Additionally,
agricultural,
the wastewaterdischarge
the increasing wastewater dischargeinto
intothe
theriver
river [136]
[136] threatens
threatens thethe sustainability
sustainability of
of the
the surface
surface water
water use.use.
As a As a result,
result, groundwater
groundwater has become
has become an appropriate
an appropriate and water
and clean clean
water resource
resource for thefor the long-term
long-term sustainability
sustainability of water
of water supplies
supplies in areas
in areas withwater
with low low water
qual-
quality.
ity. OurOur increasing
increasing understanding
understanding of of thespatial
the spatialgroundwater
groundwaterrecharge
recharge potential
potential zones
zones
highlights
highlightsthe
theimportance
importanceofofthe effective
the effectiveplanning
planning andand
management
management of water resources
of water [62].
resources
[62].
7. Conclusions
In this article, we applied the FR and EBF models, and we integrated them into a
7. Conclusions
hybridInensemble FR–EBF
this article, model that
we applied we and
the FR can use
EBFtomodels,
reveal groundwater potential
and we integrated areas.
them Wea
into
prepared, analyzed, and integrated twelve evidential factors. We investigated
hybrid ensemble FR–EBF model that we can use to reveal groundwater potential areas. the spatial
recharge basedanalyzed,
We prepared, on the topographic, hydrologic,
and integrated twelveclimatic, andfactors.
evidential land-use
Weelements, and we
investigated the
assessed their potential impacts on GWPZs. According to the findings, the AUC function
spatial recharge based on the topographic, hydrologic, climatic, and land-use elements,
for the hybrid model (0.716) was reasonably better than those of the individual FR (0.707)
and we assessed their potential impacts on GWPZs. According to the findings, the AUC
and EBF (0.665) models. According to the results, we classified the GWPZs using the natural
function for the hybrid model (0.716) was reasonably better than those of the individual
breaks classifier into five classes that ranged from very low to very high, and that covered
FR (0.707) and EBF (0.665) models. According to the results, we classified the GWPZs
areas of 22.89, 39.94, 26.22, 8.98, and 1.97% of the entire area, respectively. The applied
using the natural breaks classifier into five classes that ranged from very low to very
modeling techniques and output maps can considerably improve decision makers’ abilities
high, and that covered areas of 22.89, 39.94, 26.22, 8.98, and 1.97% of the entire area, re-
to manage groundwater resources sustainably, to advance efforts to accurately predict the
spectively. The applied modeling techniques and output maps can considerably improve
promising areas for extraction and recharge capability. This study presented valuable and
decision makers’ abilities to manage groundwater resources sustainably, to advance ef-
better information about the utilize of ensemble FR–EBF model for revealing groundwater
forts to accurately predict the promising areas for extraction and recharge capability. This
potential zones compared with the findings of individual models of FR or/and EBF. Overall,
study presented valuable and better information about the utilize of ensemble FR–EBF
model for revealing groundwater potential zones compared with the findings of indi-
vidual models of FR or/and EBF. Overall, the findings of this study may be applied to the
management, planning, and regulation of water resources under various conditions.
Author Contributions: Conceptualization, M.A.; methodology, M.A.; validation, M.A. and Y.L.;
Water 2023, 15, 480 19 of 24

the findings of this study may be applied to the management, planning, and regulation of
water resources under various conditions.

Author Contributions: Conceptualization, M.A.; methodology, M.A.; validation, M.A. and Y.L.;
investigation, M.A.; resources, N.A.-A.; writing—original draft preparation, M.A., Y.L. and N.A.-A.;
writing—review and editing, M.A.; visualization, M.A. All authors have read and agreed to the
published version of the manuscript.
Funding: This research received no external funding.
Data Availability Statement: Not applicable.
Acknowledgments: The authors thank editors and anonymous reviewers. We thank F. Abdalla.
Nasir Al-Arifi extends his grateful to the Deanship of Scientific Research, King Saud University for
funding through the Vice Deanship of Scientific Research Chairs.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Owolabi, S.T.; Madi, K.; Kalumba, A.M. Comparative evaluation of Spatio-temporal attributes of precipitation and streamflow in
Buffalo and Tyume Catchments, Eastern Cape, South Africa. Environ. Dev. Sustain. 2020, 23, 4236–4251. [CrossRef]
2. Owolabi, S.T.; Madi, K.; Kalumba, A.M.; Alemaw, B.F. Assessment of recession flow variability and the surficial lithology impact:
A case study of Buffalo River catchment, Eastern Cape, South Africa. Environ. Earth Sci. 2020, 79, 187. [CrossRef]
3. Abdelkareem, M.; Al-Arifi, N. The use of remotely sensed data to reveal geologic, structural, and hydrologic features and predict
potential areas of water resources in arid regions. Arab. J. Geosci. 2021, 14, 704. [CrossRef]
4. Moghaddam, M.; Rezaei, H.R.; Pourghasemi, Z.S.; Pourtaghie, B.; Pradhan. A Modeling Comparison of Groundwater Potential
Mapping in a Mountain Bedrock Aquifer: QUEST, GARP, and RF Models. Arab. J. Geosci. 2015, 8, 913. [CrossRef]
5. Chaminé, J.M.; Carvalho, M.J.; Afonso, J.; Teixeira, L.; Freitas. On a dialogue between hard-rock aquifer mapping and hydrogeo-
logical conceptual models: Insights into groundwater exploration. Eur. Geol. 2013, 35, 25.
6. Odhiambo, O.G. Water scarcity in the Arabian Peninsula and socio-economic implications. Appl. Water Sci. 2017, 7, 2479–2492.
[CrossRef]
7. Kim, J.C.; Jung, H.-S.; Lee, S. Spatial mapping of the groundwater potential of the geum river basin using ensemble models based
on remote sensing images. Remote Sens. 2019, 11, 2285. [CrossRef]
8. Lee, S.; Hyun, Y.; Lee, S.; Lee, M.-J. Groundwater Potential Mapping Using Remote Sensing and GIS-Based Machine Learning
Techniques. Remote Sens. 2020, 12, 1200. [CrossRef]
9. Hasanuzzaman, M.; Mandal, M.H.; Hasnine; Shit, P.K. Groundwater potential mapping using multi-criteria decision, bivariate
statistic and machine learning algorithms: Evidence from Chota Nagpur Plateau, India. Appl. Water Sci. 2022, 12, 58. [CrossRef]
10. Nguyen, P.T.; Ha, D.H.; Jaafari, A.; Nguyen, H.D.; Van Phong, T.; Al-Ansari, N.; Prakash, I.; Van Le, H.; Pham, B.T. Groundwater
Potential Mapping Combining Artificial Neural Network and Real AdaBoost Ensemble Technique: The DakNong Province
Case-study, Vietnam. Int. J. Environ. Res. Public Health 2020, 17, 2473. [CrossRef]
11. Rahmati, O.; Melesse, A.M. Application of Dempster–Shafer theory, spatial analysis and remote sensing for groundwater potential
and nitrate pollution analysis in the semi-arid region of Khuzestan, Iran. Sci. Total Environ. 2016, 568, 1110–1123. [CrossRef]
[PubMed]
12. Golkarian, A.; Naghibi, S.A.; Kalantar, B.; Pradhan, B. Groundwater potential mapping using C5.0, random forest, and multivariate
adaptive regression spline models in GIS. Environ. Monit. Assess. 2018, 190, 149. [CrossRef] [PubMed]
13. Barlow, P.M.; Leake, S.A. Streamflow Depletion by Wells: Understanding and Managing the Effects of Groundwater Pumping on
Streamflow; US Geological Survey: Reston, VA, USA, 2012.
14. Barker, J. A generalized radial flow model for hydraulic tests in fractured rock. Water Resour. Res. 1988, 24, 1796–1804. [CrossRef]
15. Abdelkareem, M.; El-Baz, F.; Askalany, M.; Akawy, A.; Ghoneim, E. Groundwater prospect map of Egypt’s Qena Valley using
data fusion. Int. J. Image Data Fusion 2012, 3, 169–189. [CrossRef]
16. Elbeih, S.F. An overview of integrated remote sensing and GIS for groundwater mapping in Egypt. Ain Shams Eng. J. 2015, 6,
1–15. [CrossRef]
17. Dušek, P.; Velísková, Y. Comparison of the MODFLOW modules for the simulation of the river type boundary condition. Pollack
Period. 2017, 12, 3–13. [CrossRef]
18. Zhu, Q.; Abdelkareem, M. Mapping Groundwater Potential Zones Using a Knowledge-Driven Approach and GIS Analysis.
Water 2021, 13, 579. [CrossRef]
19. Rahmati, O.; Naghibi, S.A.; Shahabi, H.; Bui, D.T.; Pradhan, B.; Azareh, A.; Rafiei-Sardooi, E.; Samani, A.N.; Melesse, A.M.
Groundwater spring potential modelling: Comprising the capability and robustness of three different modeling approaches. J.
Hydrol. 2018, 565, 248–261. [CrossRef]
20. Shenga, Z.D.; Baroková, D.; Šoltész, A. Modeling of groundwater extraction from wells to control excessive water levels. Pollack
Period. 2018, 13, 125–136. [CrossRef]
Water 2023, 15, 480 20 of 24

21. Mandel, S. Groundwater Resources: Investigation and Development; Elsevier: Amsterdam, The Netherlands, 2012.
22. Waikar, M.L.; Nilawar, A.P. Identification of groundwater potential zone using remote sensing and GIS technique. Int. J. Innov.
Res. Sci. Eng. Technol. 2014, 3, 12163–12174.
23. Thompson, S.A. Hydrology for Water Management; CRC Press: Boca Raton, FL, USA, 2017.
24. Abdelkareem, M.; El-Baz, F. Analyses of optical images and radar data reveal structural features and predict groundwater
accumulations in the central Eastern Desert of Egypt. Arab. J. Geosci. 2014, 8, 2653–2666. [CrossRef]
25. Abdelkareem, M.; Abdalla, F. Revealing potential areas of water resources using integrated remote-sensing data and GIS-based
analytical hierarchy process. Geocarto Int. 2021, 1–25. [CrossRef]
26. Arulbalaji, P.; Padmalal, D.; Sreelash, K. GIS and AHP Techniques Based Delineation of Groundwater Potential Zones: A case
study from Southern Western Ghats, India. Sci. Rep. 2019, 9, 2082. [CrossRef] [PubMed]
27. Kumar, V.A.; Mondal, N.C.; Ahmed, S. Identification of Groundwater Potential Zones Using RS, GIS and AHP Techniques:
A Case Study in a Part of Deccan Volcanic Province (DVP), Maharashtra, India. J. Indian Soc. Remote Sens. 2020, 48, 497–511.
[CrossRef]
28. Yariyan, P.; Avand, M.; Omidvar, E.; Pham, Q.B.; Linh, N.T.T.; Tiefenbacher, J.P. Optimization of statistical and machine learning
hybrid models for groundwater potential mapping. Geocarto Int. 2020, 37, 3877–3911. [CrossRef]
29. Sun, T.; Cheng, W.; Abdelkareem, M.; Al-Arifi, N. Mapping Prospective Areas of Water Resources and Monitoring Land Use/Land
Cover Changes in an Arid Region Using Remote Sensing and GIS Techniques. Water 2022, 14, 2435. [CrossRef]
30. Melese, T.; Belay, T. Groundwater Potential Zone Mapping Using Analytical Hierarchy Process and GIS in Muga Watershed,
Abay Basin, Ethiopia. Glob. Chall. 2021, 6, 2100068. [CrossRef]
31. Priya, U.; Iqbal, M.A.; Salam, M.A.; Alam, N.E.; Uddin, M.F.; Islam, A.R.M.T.; Sarkar, S.K.; Imran, S.I.; Rak, A.E. Sustainable
Groundwater Potential Zoning with Integrating GIS, Remote Sensing, and AHP Model: A Case from North-Central Bangladesh.
Sustainability 2022, 14, 5640. [CrossRef]
32. Lee, S.; Hyun, Y.; Lee, M.-J. Groundwater potential mapping using data mining models of big data analysis in Goyang-si, South
Korea. Sustainability 2019, 11, 1678. [CrossRef]
33. Singh, L.K.; Jha, M.K.; Chowdary, V. Assessing the accuracy of GIS-based Multi-Criteria Decision Analysis approaches for
mapping groundwater potential. Ecol. Indic. 2018, 91, 24–37. [CrossRef]
34. Riad, P.; Billib, M.; Hassan, A.; Salam, M.; El Din, M. Application of the overlay weighted model and boolean logic to determine
the best locations for artificial recharge of groundwater. J. Urban Environ. Eng. 2011, 5, 57–66. [CrossRef]
35. Mallick, J.; Abad Khan, R.; Ahmed, M.; Alqadhi, S.D.; Alsubih, M.; Falqi, I.; Abul Hasan, M. Modeling Groundwater Potential
Zone in a Semi-Arid Region of Aseer Using Fuzzy-AHP and Geoinformation Techniques. Water 2019, 11, 2656. [CrossRef]
36. Muthumaniraja, C.; Anbazhagan, S.; Jothibasu, A.; Chinnamuthu, M. Remote Sensing and Fuzzy Logic Approach for Artificial
Recharge Studies in Hard Rock Terrain of South India. In GIS and Geostatistical Techniques for Groundwater Science; Elsevier:
Amsterdam, The Netherlands, 2019; pp. 91–112. [CrossRef]
37. Shahid, S.; Nath, S.K.; Kamal, A.S.M. GIS Integration of Remote Sensing and Topographic Data Using Fuzzy Logic for Ground
Water Assessment in Midnapur District, India. Geocarto Int. 2014, 17, 69–74. [CrossRef]
38. Ozdemir, A. GIS-based groundwater spring potential mapping in the Sultan Mountains (Konya, Turkey) using frequency ratio,
weights of evidence and logistic regression methods and their comparison. J. Hydrol. 2011, 411, 290–308. [CrossRef]
39. Manap, M.A.; Nampak, H.; Pradhan, B.; Lee, S.; Sulaiman, W.N.A.; Ramli, M.F. Application of probabilistic-based frequency ratio
model in groundwater potential mapping using remote sensing data and GIS. Arab. J. Geosci. 2012, 7, 711–724. [CrossRef]
40. Guru, B.; Seshan, K.; Bera, S. Frequency ratio model for groundwater potential mapping and its sustainable management in cold
desert, India. J. King Saud. Univ.—Sci. 2017, 29, 333–347. [CrossRef]
41. Naghibi, S.A.; Pourghasemi, H.R.; Dixon, B. GIS-based groundwater potential mapping using boosted regression tree, clas-
sification and regression tree, and random forest machine learning models in Iran. Environ. Monit. Assess. 2015, 188, 44.
[CrossRef]
42. Muavhi, N.; Thamaga, K.H.; Mutoti, M.I. Mapping groundwater potential zones using relative frequency ratio, analytic hierarchy
process and their hybrid models: Case of Nzhelele-Makhado area in South Africa. Geocarto Int. 2022, 37, 6311–6330. [CrossRef]
43. Oh, H.-J.; Kim, Y.-S.; Choi, J.-K.; Park, E.; Lee, S. GIS mapping of regional probabilistic groundwater potential in the area of
Pohang City, Korea. J. Hydrol. 2011, 399, 158–172. [CrossRef]
44. Chen, W.; Li, H.; Hou, E.; Wang, S.; Wang, G.; Panahi, M.; Li, T.; Peng, T.; Guo, C.; Niu, C.; et al. GIS-based groundwater potential
analysis using novel ensemble weights-of-evidence with logistic regression and functional tree models. Sci. Total. Environ. 2018,
634, 853–867. [CrossRef]
45. Arabameri, A.; Rezaei, K.; Cerda, A.; Lombardo, L.; Rodrigo-Comino, J. GIS-based groundwater potential mapping in Shahroud
plain, Iran. A comparison among statistical (bivariate and multivariate), data mining and MCDM approaches. Sci. Total. Environ.
2019, 658, 160–177. [CrossRef] [PubMed]
46. Lee, S.; Song, K.-Y.; Kim, Y.; Park, I. Regional groundwater productivity potential mapping using a geographic information
system (GIS) based artificial neural network model. Hydrogeol. J. 2012, 20, 1511–1527. [CrossRef]
47. Corsini, A.; Cervi, F.; Ronchetti, F. Weight of evidence and artificial neural networks for potential groundwater mapping: An
application to the Mt. Modino area (Northern Apennines, Italy). Geomorphology 2009, 111, 79–87. [CrossRef]
Water 2023, 15, 480 21 of 24

48. Moghaddam, D.; Rezaei, M.; Pourghasemi, H.R.; Pourtaghi, Z.S.; Pradhan, B. Groundwater spring po-tential mapping using
bivariate statistical model and GIS in the Taleghan watershed Iran. Arab. J. Geosci. 2013, 8, 913–929. [CrossRef]
49. Kaliraj, S.; Chandrasekar, N.; Magesh, N.S. Idnetification of potential groundwater recharge zones in Vaigai upper basin, Tamil
Nadu, using GISbased analytical hierarchical process (AHP) technique. Arab. J. Geosci. 2014, 7, 1385–1401. [CrossRef]
50. Nampak, H.; Pradhan, B.; Manap, M.A. Application of GIS based data driven evidential belief function model to predict
groundwater potential zonation. J. Hydrol. 2014, 513, 283–300. [CrossRef]
51. Park, I.; Kim, Y.; Lee, S. Groundwater Productivity Potential Mapping Using Evidential Belief Function. Groundwater 2014, 52,
201–207. [CrossRef]
52. Mogaji, K.A.; Lim, H.S.; Abdullah, K. Regional prediction of groundwater potential mapping in a multifaceted geology terrain
using GIS-based Dempster–Shafer model. Arab. J. Geosci. 2015, 8, 3235–3258. [CrossRef]
53. Al-Abadi, A.M. Groundwater potential mapping at northeastern Wasit and Missan governorates, Iraq using a data-driven
weights of evidence technique in framework of GIS. Environ. Earth Sci. 2015, 74, 1109–1124. [CrossRef]
54. Mogaji, K.A.; Omosuyi, G.O.; Adelusi, A.O.; Lim, H.S. Application of GIS-Based Evidential Belief Function Model to Regional
Groundwater Recharge Potential Zones Mapping in Hardrock Geologic Terrain. Environ. Process. 2016, 3, 93–123. [CrossRef]
55. Rahmati, O.; Pourghasemi, H.R.; Melesse, A.M. Application of GIS-based data driven random forest and maximum entropy
models for groundwater potential mapping: A case study at Mehran Region, Iran. Catena 2016, 137, 360–372. [CrossRef]
56. Pourghasemi, H.R.; Beheshtirad, M. Assessment of a data-driven evidential belief function model and GIS for groundwater
potential mapping in the Koohrang Watershed, Iran. Geocarto Int. 2014, 30, 662–685. [CrossRef]
57. Naghibi, S.A.; Ahmadi, K.; Daneshi, A. Application of support vector machine, random forest, and genetic algorithm optimized
randomforest models in groundwater potential mapping. Water Resour. Manag. 2017, 31, 2761–2775. [CrossRef]
58. Lee, S.; Hong, S.-M.; Jung, H.-S. GIS-based groundwater potential mapping using artificial neural network and support vector
machine models: The case of Boryeong city in Korea. Geocarto Int. 2017, 33, 847–861. [CrossRef]
59. Arnold, R.S. Groundwater Potential Mapping Using Maximum Entropy. In Water Resources Management and Sustainability; Kumar,
P., Nigam, G.K., Sinha, M.K., Singh, A., Eds.; Springer: Berlin/Heidelberg, Germany, 2022.
60. Zeng, Z.; Li, Y.; Lan, J.; Hamidi, A.R. Utilizing User-Generated Content and GIS for Flood Susceptibility Modeling in Mountainous
Areas: A Case Study of Jian City in China. Sustainability 2021, 13, 6929. [CrossRef]
61. Lin, J.; He, P.; Yang, L.; He, X.; Lu, S.; Liu, D. Predicting future urban waterlogging-prone areas by coupling the maximum entropy
and FLUS model. Sustain. Cities Soc. 2022, 80, 103812. [CrossRef]
62. Jaafarzadeh, M.S.; Tahmasebipour, N.; Haghizadeh, A.; Pourghasemi, H.R.; Rouhani, H. Groundwater recharge potential zonation
using an ensemble of machine learning and bivariate statistical models. Sci. Rep. 2021, 11, 5587. [CrossRef] [PubMed]
63. Razandi, Y.; Pourghasemi, H.R.; Neisani, N.S.; Rahmati, O. Application of analytical hierarchy process, frequency ratio, and
certainty factor models for groundwater potential mapping using GIS. Earth Sci. Inform. 2015, 8, 867–883. [CrossRef]
64. Arshad, A.; Zhang, Z.; Zhang, W.; Dilawar, A. Mapping favorable groundwater potential recharge zones using a GIS-based
analytical hierarchical process and probability frequency ratio model: A case study from an agro-urban region of Pakistan. Geosci.
Front. 2020, 11, 1805–1819. [CrossRef]
65. Neshat, A.; Pradhan, B. An integrated DRASTIC model using frequency ratio and two new hybrid methods for groundwater
vulnerability assessment. Nat. Hazards 2015, 76, 543–563. [CrossRef]
66. Das, S. Comparison among influencing factor, frequency ratio, and analytical hierarchy process techniques for groundwater
potential zonation in Vaitarna basin, Maharashtra, India. Groundw. Sustain. Dev. 2019, 8, 617–629. [CrossRef]
67. Urqueta, H.; Jódar, J.; Herrera, C.; Wilke, H.G.; Medina, A.; Urrutia, J.; Custodio, E.; Rodríguez, J. Land surface temperature as an
indicator of the unsaturated zone thickness: A remote sensing approach in the Atacama Desert. Sci. Total Environ. 2018, 612,
1234–1248. [CrossRef] [PubMed]
68. Siahkamari, S.; Haghizadeh, A.; Zeinivand, H.; Tahmasebipour, N.; Rahmati, O. Spatial prediction of food-susceptible areas using
frequency ratio and maximum entropy models. Geocarto Int. 2018, 33, 927–941. [CrossRef]
69. Liu, G. On the geo-basis of river regulation in the lower reaches of the Yellow River. Sci. China Earth Sci. 2012, 55, 530–544.
[CrossRef]
70. Zhi, C.; Cao, W.; Wang, Z.; Li, Z.; Ren, Y. Genesis of As in the groundwater with extremely high salinity in the Yellow River Delta,
China. Appl. Geochem. 2022, 139, 105229. [CrossRef]
71. Liu, Y.; Huang, H.-J. Characterization and mechanism of regional land subsidence in the Yellow River Delta, China. Nat. Hazards
2013, 68, 687–709. [CrossRef]
72. O’Callaghan, J.F.; Mark, D.M. The extraction of drainage networks from digital elevation data. Comput. Vis. Graph. Image Process.
1984, 28, 323–344. [CrossRef]
73. Tehrany, M.S.; Kumar, L. The application of a Dempster–Shafer-based evidential belief function in flood susceptibility mapping
and comparison with frequency ratio and logistic regression methods. Environ. Earth Sci. 2018, 77, 490. [CrossRef]
74. Bui, D.T.; Pradhan, B.; Lofman, O.; Revhaug, I.; Dick, O.B. Spatial prediction of landslide hazards in Hoa Binh province (Vietnam):
A comparative assessment of the efficacy of evidential belief functions and fuzzy logic models. Catena 2012, 96, 28–40.
75. Pradhan, B.; Abokharima, M.H.; Jebur, M.N.; Tehrany, M.S. Land subsidence susceptibility mapping at Kinta Valley (Malaysia)
using the evidential belief function model in GIS. Nat. Hazards 2014, 73, 1019–1042. [CrossRef]
Water 2023, 15, 480 22 of 24

76. Devkota, K.C.; Regmi, A.D.; Pourghasemi, H.R.; Yoshida, K.; Pradhan, B.; Ryu, I.C.; Dhital, M.R.; Althuwaynee, O.F. Landslide
susceptibility mapping using certainty factor, index of entropy and logistic regression models in GIS and their comparison at
Mugling–Narayanghat road section in Nepal Himalaya. Nat. Hazards 2013, 65, 135–165. [CrossRef]
77. Ahmadi, H.; Kaya, O.A.; Babadagi, E.; Savas, T.; Pekkan, E. GIS-Based Groundwater Potentiality Mapping Using AHP and FR
Models in Central Antalya, Turkey. Environ. Sci. Proc. 2021, 5, 11.
78. Mallick, J.; Al-Wadi, H.; Rahman, A.; Ahmed, M. Landscape dynamic characteristics using satellite data for a mountainous
watershed of Abha, Kingdom of Saudi Arabia. Environ. Earth Sci. 2014, 72, 4973–4984. [CrossRef]
79. Al Saud, M. Mapping potential areas for groundwater storage in Wadi Aurnah Basin, western Arabian Peninsula, using remote
sensing and geographic information system techniques. Hydrogeol. J. 2018, 18, 1481–1495. [CrossRef]
80. Gupta, M.; Srivastava, P.K. Integrating GIS and remote sensing for identification of groundwater potential zones in the hilly
terrain of Pavagarh, Gujarat, India. Water Int. 2010, 35, 233–245. [CrossRef]
81. Rawat, K.S.; Singh, S.K.; Singh, M.I.; Garg, B. Comparative evaluation of vertical accuracy of elevated points with ground control
points from ASTERDEM and SRTMDEM with respect to CARTOSAT-1DEM. Remote Sens. Appl. Soc. Environ. 2018, 13, 289–297.
[CrossRef]
82. Jasrotia, A.S.; Bhagat, B.D.; Kumar, A.; Kumar, R. Remote sensing and GIS approach for delineation of groundwater potential and
ground water quality zones of western Doon Valley, Uttarakhand, India. J. Indian Soc. Remote Sens. 2012, 41, 365–377. [CrossRef]
83. Deepika, B.; Avinash, K.; Jayappa, K.S. Integration of hydrological factors and demarcation of groundwater prospect zones:
Insights from remote sensing and GIS techniques. Environ. Earth Sci. 2013, 70, 1319–1338. [CrossRef]
84. Fashae, O.A.; Tijani, M.N.; Talabi, A.O.; Adedeji, O.I. Delineation of groundwater potential zones in the crystalline basement
terrain of SW-Nigeria: An integrated GIS and remote sensing approach. Appl. Water Sci. 2013, 4, 19–38. [CrossRef]
85. Sikakwe, G.U.; Ntekim, E.E.U.; Obi, D.A.; George, A.M. Geohydrological study of weathered basement aquifers in Oban massif
and environs southeastern Nigeria: Using remote sensing and geographic information system techniques. IOSR J. Appl. Geol.
Geophys. 2015, 3, 321–990.
86. Ahmad, N.; Khan, S.; Ehsan, M.; Rehman, F.U.; Al-Shuhail, A. Estimating the Total Volume of Running Water Bodies Using
Geographic Information System (GIS): A Case Study of Peshawar Basin (Pakistan). Sustainability 2022, 14, 3754. [CrossRef]
87. Gaur, S.; Chahar, B.; Graillot, D. Combined use of groundwater modeling and potential zone analysis for management of
groundwater. Int. J. Appl. Earth Obs. Geoinf. 2011, 13, 127–139. [CrossRef]
88. Kalantar, B.; Al-Najjar, H.A.H.; Pradhan, B.; Saeidi, V.; Halin, A.A.; Ueda, N.; Naghibi, S.A. Optimized Conditioning Factors
Using Machine Learning Techniques for Groundwater Potential Mapping. Water 2019, 11, 1909. [CrossRef]
89. Jaiswal, R.K.; Mukherjee, S.; Krishnamurthy, J.; Saxena, R. Role of remote sensing and GIS techniques for generation of
groundwater prospect zones towards rural development-an approach. Int. J. Remote Sens. 2003, 24, 993–1008. [CrossRef]
90. Prasad, R.K.; Mondal, N.C.; Banerjee, P.; Nandakumar, M.V.; Singh, V.S. Deciphering potential groundwater zone in hard rock
through the application of GIS. Environ. Geol. 2007, 55, 467–475. [CrossRef]
91. Manap, M.A.; Sulaiman, W.N.A.; Ramli, M.F.; Pradhan, B.; Surip, N. A knowledge-driven GIS modeling technique for groundwa-
ter potential mapping at the Upper Langat Basin, Malaysia. Arab. J. Geosci. 2013, 6, 1621–1637. [CrossRef]
92. Murthy, K.S.R. Groundwater potential in a semi-arid region of Andhra Pradesh: A geographical information system approach.
Int. J. Remote Sens. 2000, 21, 1867–1884. [CrossRef]
93. Yeh, H.-F.; Lee, C.-H.; Hsu, K.-C.; Chang, P.-H. GIS for the assessment of the groundwater recharge potential zone. Environ. Geol.
2009, 58, 185–195. [CrossRef]
94. Pinto, D.; Shrestha, S.; Babel, M.S.; Ninsawat, S. Delineation of groundwater potential zones in the Comoro watershed, Timor
Leste using GIS, remote sensing and analytichierarchy process (AHP) technique. Appl. Water Sci. 2017, 7, 503–519. [CrossRef]
95. Selvam, S.; Dar, F.A.; Magesh, N.S.; Singaraja, C.; Venkatramanan, S.; Chung, S.Y. Application of remote sensing and GIS for
delineating groundwater recharge potential zones of Kovilpatti Municipality, Tamil Nadu using IF technique. Earth Sci. Inform.
2016, 9, 137–150. [CrossRef]
96. Kumar, P.K.D.; Gopinath, G.; Seralathan, P. Application of remote sensing and GIS for the demarcation of groundwater potential
zones of a river basin in Kerala, southwest coast of India. Int. J. Remote Sens. 2007, 28, 5583–5601. [CrossRef]
97. Magesh, N.; Chandrasekar, N.; Soundranayagam, J.P. Delineation of groundwater potential zones in Theni district, Tamil Nadu,
using remote sensing, GIS and MIF techniques. Geosci. Front. 2012, 3, 189–196. [CrossRef]
98. Chowdhury, A.; Jha, M.K.; Chowdary, V.M.; Mal, B.C. Integrated remote sensing and GIS-based approach for assessing
groundwater potential in West Medinipur district, West Bengal, India. Int. J. Remote Sens. 2009, 30, 231–250. [CrossRef]
99. Winzeler, H.E.; Owens, P.R.; Read, Q.D.; Libohova, Z.; Ashworth, A.; Sauer, T. Topographic Wetness Index as a Proxy for Soil
Moisture in a Hillslope Catena: Flow Algorithms and Map Generalization. Land 2022, 11, 2018. [CrossRef]
100. Pourali, S.H.; Arrowsmith, C.; Chrisman, N.; Matkan, A.A.; Mitchell, D. Topography wetness index application in flood-risk-based
land use planning. Appl. Spat. Anal. Policy 2016, 9, 39–54. [CrossRef]
101. Lee, S.; Pradhan, B. Landslide hazard mapping at Selangor, Malaysia using frequency ratio and logistic regression models.
Landslides 2007, 4, 33–41. [CrossRef]
102. Moore, I.D.; Grayson, R.B.; Ladson, A.R. Digital terrain modelling: A review of hydrological, geomorphological, and biological
applications. Hydrol. Process. 1991, 5, 3–30. [CrossRef]
Water 2023, 15, 480 23 of 24

103. Hojati, M.; Mokarram, M. Determination of a topographic wetness index using high-resolution digital elevation models. Eur. J.
Geogr. 2016, 7, 41–52.
104. Neilson, B.T.; Cardenas, M.B.; O’Connor, M.T.; Rasmussen, M.T.; King, T.V.; Kling, G.W. Groundwater flow and exchange across
the land surface explain carbon export patterns in continuous permafrost watersheds. Geophys. Res. Lett. 2018, 45, 7596–7605.
[CrossRef]
105. Cao, C.; Xu, P.; Wang, Y.; Chen, J.; Zheng, L.; Niu, C. Flash Flood Hazard Susceptibility Mapping Using Frequency Ratio and
Statistical Index Methods in Coalmine Subsidence Areas. Sustainability 2016, 8, 948. [CrossRef]
106. Alarifi, S.S.; Abdekareem, M.; Abdalla, F.; Alotaibi, M. Mapping Susceptible Areas to Flash Flood Hazards Using Remote Sensing
and GIS Techniques in the Southwest Part of Saudi Arabia. Sustainability 2022, 14, 14145. [CrossRef]
107. Benjmel, K.; Amraoui, F.; Boutaleb, S.; Ouchchen, M.; Tahiri, A.; Touab, A. Mapping of groundwater potential zones in crystalline
terrain using remote sensing, GIS techniques, and multicriteria data analysis (case of the Ighremregion, western Anti-Atlas,
Morocco). Water 2020, 12, 471. [CrossRef]
108. Golkarian, A.; Rahmati, O. Use of a maximum entropy model to identify the key factors that influence groundwater availability
on the Gonabad Plain, Iran. Environ. Earth Sci. 2018, 77, 369. [CrossRef]
109. Abdelkareem, M.; El-Baz, F. Mode of formation of the Nile Gorge in northern Egypt: A study by DEM-SRTM data and GIS
analysis. Geol. J. 2015, 51, 760–778. [CrossRef]
110. Achu, A.L.; Reghunath, R.; Thomas, J. Mapping of Groundwater Recharge Potential Zones and Identification of Suitable Site
Specific Recharge Mechanisms in a Tropical River Basin. J. Earth Syst. Environ. 2020, 4, 131–145. [CrossRef]
111. Hung, L.Q.; Batelaan, O.; de Smedt, F. Lineament extraction and analysis, comparison of LANDSAT ETM and ASTER imagery.
Case study: Suoimuoi tropical karst catchment, Vietnam. Proc. SPIE 2005, 5983, 59830–59831.
112. Assatse, W.T.; Nouck, P.N.; Tabod, C.T.; Akame, J.M.; Biringanine, G.N. Hydrogeological activity of lineaments in Yaoundé
Cameroon region using remote sensing and GIS techniques. Egypt. J. Remote Sens. Space Sci. 2016, 19, 49–60. [CrossRef]
113. Pradhan, B. Groundwater potential zonation for basaltic watersheds using satellite remote sensing data and GIS techniques. Cent.
Eur. J. Geosci. 2009, 1, 120–129. [CrossRef]
114. Abdelkareem, M.; El-Baz, F. Remote sensing of Paleodrainage systems west of the Nile River, Egypt. J. Geocarto Int. 2017, 32,
541–555. [CrossRef]
115. Dawood, F.; Akhtar, M.M.; Ehsan, M. Evaluating urbanization impact on stressed aquifer of Quetta Valley, Pakistan. Desalination
Water Treat. 2021, 222, 103–113. [CrossRef]
116. Hussein, A.-A.; Govindu, V.; Nigusse, A.G.M. Evaluation of groundwater potential using geospatial techniques. Appl. Water Sci.
2016, 7, 2447–2461. [CrossRef]
117. Kumar, P.; Herath, S.; Avtar, R.; Takeuchi, K. Mapping of groundwater potential zones in Killinochi area, Sri Lanka, using GIS and
remote sensing techniques. Sustain. Water Resour. Manag. 2016, 2, 419–430. [CrossRef]
118. Sajjad, H.; Iqbal, M.; Bhat, F.A. Integrating geospatial and geophysical information for deciphering groundwater potential zones
in Dudhganga catchment, Kashmir Valley, India. Am. J. Water Resour. 2014, 2, 18–24. [CrossRef]
119. Fenta, A.A.; Kifle, A.; Gebreyohannes, T.; Hailu, G. Spatial analysis of groundwater potential using remote sensing and GIS-based
multi criteria evaluation in Raya Valley, northern Ethiopia. Hydrogeol. J. 2014, 23, 195–206. [CrossRef]
120. Gee, G.W.; Wierenga, P.J.; Andraski, B.; Young, M.H.; Fayer, M.J.; Rockhold, M.L. Variations in Water Balance and Recharge
Potential at Three Western Desert Sites. Soil Sci. Soc. Am. J. 1994, 58, 63–72. [CrossRef]
121. Prych, E.A. Using Chloride and Chlorine-36 as Soil-Water Tracers to Estimate Deep Percolation at Selected Locations on the US Department
of Energy Hanford Site, Washington; US Geological Survey: Reston, VA, USA, 1998; Volume 2481.
122. Senthilkumar, M.; Gnanasundar, D.; Arumugam, R. Identifying groundwater recharge zones using remote sensing & GIS
techniques in Amaravathi aquifer system, Tamil Nadu, South India. Sustain. Environ. Res. 2019, 29, 15. [CrossRef]
123. Singh, S.K.; Zeddies, M.; Shankar, U.; Griffiths, G.A. Potential groundwater recharge zones within New Zealand. Geosci. Front.
2019, 10, 1065–1072. [CrossRef]
124. Abdekareem, M.; Al-Arifi, N.; Abdalla, F.; Mansour, A.; El-Baz, F. Fusion of Remote Sensing Data Using GIS-Based AHP-Weighted
Overlay Techniques for Groundwater Sustainability in Arid Regions. Sustainability 2022, 14, 7871. [CrossRef]
125. Adiat, K.A.N.; Nawawi, M.N.M.; Abdullah, K. Assessing th e accuracy of GIS-based elementary multi criteria de-cision analysis
as a spatial prediction tool—A case of predicting potential zones of sustainable groundwater re-sources. J. Hydrol. 2012, 440,
75–89. [CrossRef]
126. Janizadeh, S.; Avand, M.; Jaafari, A.; Van Phong, T.; Bayat, M.; Ahmadisharaf, E.; Prakash, I.; Pham, B.T.; Lee, S. Prediction Success
of Machine Learning Methods for Flash Flood Susceptibility Mapping in the Tafresh Watershed, Iran. Sustainability 2019, 11, 5426.
[CrossRef]
127. Lee, S.; Pradhan, B. Probabilistic landslide hazards and risk mapping on Penang Island, Malaysia. J. Earth Syst. Sci. 2006, 115,
661–672. [CrossRef]
128. Dempster, A.P. Upper and lower probabilities induced by a multivalued mapping. In Classic Works of the Dempster–Shafer Theory
of Belief Functions; Springer: Berlin/Heidelberg, Germany, 2008; pp. 57–72.
129. Shafer, G. A Mathematical Theory of Evidence; Princeton University Press: Princeton, NJ, USA, 1976.
130. Dempster, A.P. New approaches for reasoning towards posterior distributions based on sample data. Ann. Math. Statist. 1966, 37,
355–374. [CrossRef]
Water 2023, 15, 480 24 of 24

131. Tehrany, M.S.; Pradhan, B.; Jebur, M.N. Flood susceptibility mapping using a novel ensemble weights-of-evidence and support
vector machine models in GIS. J. Hydrol. 2014, 512, 332–343. [CrossRef]
132. Carranza, E.J.M.; Hale, M. Evidential belief functions for datadriven geologically constrained mapping of gold potential, Baguio
district, Philippines. Ore Geol. Rev. 2003, 22, 117–132. [CrossRef]
133. Hong, Y.; Abdelkareem, M. Integration of remote sensing and a GIS-based method for revealing prone areas to flood hazards and
predicting optimum areas of groundwater resources. Arab. J. Geosci. 2022, 15, 114. [CrossRef]
134. Zhang, Z.; Yang, F.; Chen, H.; Wu, Y.; Li, T.; Li, W.; Wang, Q.; Liu, P. GIS-based landslide susceptibility analysis using frequency
ratio and evidential belief function models. Environ. Earth Sci. 2016, 75, 948. [CrossRef]
135. Zhao, M.M.; Wang, S.-M.; Chen, Y.-P.; Wu, J.-H.; Xue, L.-G.; Fan, T.T. Pollution status of the Yellow River tributaries in middle and
lower reaches. Sci. Total. Environ. 2020, 722, 137861. [CrossRef]
136. Yellow River Conservancy Commission, Yellow River Water Resources Bulletin, 2017. Yellow River Conservancy Commission of
MWR, China. Available online: http://www.yrcc.gov.cn/other/hhgb/ (accessed on 1 December 2022).

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.