You are on page 1of 14

Environ Earth Sci

DOI 10.1007/s12665-014-3476-2

ORIGINAL ARTICLE

Modeling of groundwater recharge using a multiple linear


regression (MLR) recharge model developed from geophysical
parameters: a case of groundwater resources management
Kehinde Anthony Mogaji Hwee San Lim
Khiruddin Abdullah

Received: 16 January 2014 / Accepted: 19 June 2014


Springer-Verlag Berlin Heidelberg 2014

Abstract In this paper we developed a simple multiple


linear regression (MLR) recharge model that relates the
recharge estimates obtained from rainfall to the geophysical parameters obtained from the interpretation of twodimensional (2D) resistivity imaging data for the purpose
of efficient groundwater resources management in the
southern part of Perak, Malaysia through recharge rate
estimation and prediction. Through application of linear
regression model, the estimated recharge from rainfall and
the corresponding estimated unsaturated layer resistivity
and its thickness (Depth to aquifer top) parameters
obtained from geophysical measurements were regressed in
R software written code environment for generating a MLR
recharge model. The sensitivity of analyzed results of the
MLR recharge model based on the parameter estimation of
the model predictors (resistivity and depth) evaluated at
Pr B 0.05 is 5.39 9 10-06 and 8.39 9 10-04, respectively.
The accuracy and predictive power test conducted on the
developed model using both t test and v2 distribution at
a = 5 % significance level established the model estimation and prediction capability. The obtained results of v2
distribution test and parameters estimation test confirmed
the reliability and accuracy of the proposed model in
K. A. Mogaji
Department of Applied Geophysics, Federal University
of Technology, P.M.B. 704, Akure, Nigeria
K. A. Mogaji (&)  H. S. Lim  K. Abdullah
School of Physics, Universiti Sains Malaysia,
11800 Georgetown, Penang, Malaysia
e-mail: mogakeh@yahoo.com
H. S. Lim
e-mail: hslim@usm.my
K. Abdullah
e-mail: khirudd@usm.my

recharge rate estimation and prediction in the area. The


application of the MLR recharge model gives estimate of
242.30 mm/year for regional groundwater recharge rate in
the area. Through GIS tool, the MLR recharge model was
used to produce groundwater recharge rate prediction map.
A quick and independent estimate of recharge by simple
geophysical measurement has been established based on
these results. The information on the prediction map could
serve as a scientific basis for groundwater resources management and exploration in the area. The approach suggests
a new application of geoelectric parameters in determining
recharge rate due to infiltration. The technique provides a
good alternative to other methods used for this purpose.
Keywords Multivariate regression recharges model 
Groundwater recharge prediction  2D resistivity imaging 
Geophysical parameters  2D resistivity imaging 
Hydrogeological

Introduction
The management of groundwater resources requires a
method to accurately calculate groundwater recharge rates
either on local scale or regional scale. As emphasized in the
studies according to Jyrkama and Sykes (2007) and Kaliraj
et al. (2014), the understanding of groundwater recharge is
fundamental to the management of groundwater resources.
In addition, for an effective watershed management strategy, the quantification of groundwater recharge is vital in
ensuring the protection of groundwater resources from an
unavoidable climate change impact and other stresses like
industrial revolution, urbanization, etc. (Robins 1998).
Moreover, it has also been reported in the study of Nolan
et al. (2007) that recharge is a major component of the

123

Environ Earth Sci

groundwater system as well as its importance in shallow


groundwater quality evaluation. The situation and fluctuation of groundwater quantity measurement required for an
exploitation of optimal groundwater resources are largely a
function of rate of groundwater recharge (Kumar 1977;
Omorinbola 2009). This was recently corroborated in the
study according to Gontia and Patil (2012), where it was
reported that locating groundwater potential zones for
groundwater resource development not only is enough to
salvage potable water resources supply demands for
domestic, agricultural, industrial, and other purposes, but
also requires quantifying the natural and artificial recharge
rates of an area of interest. Therefore, in lieu of sustaining
these precious natural resources, several studies have
emphasized on the estimation of groundwater recharge to
the aquifer system as very essential for the proper management of resources (Chandra et al. 2004; Leipnik and
Loaiciga 2006). However, according to Kumar and
Seethapathi (2002), in evaluation of groundwater resources, the rate of aquifer recharge is one of the most difficult
factors to measure and as such aquifer recharge rate estimation by any other method is always characterized with
uncertainties and errors. This was due to the fact that
recharge variable is a complex function of multiple factors
such as meteorological conditions, soil, vegetation, physiographic characteristics, and properties of the geologic
material within the paths of flow (Kumar and Ish 2012).
Besides, topography which is a factor that exerts more
influence on groundwater flow direction cannot be overemphasized in determining the rate of recharge of an area
(Akpan et al. 2013; Doumouya et al. 2012). Thus, an
approach that can reliably estimate and reduce uncertainty
in recharge rate estimation and prediction is needed.
Numerous studies have estimated groundwater recharge
rate via conventional methods such as water balance,
Darcian approach, lysimeter, water table fluctuation,
numerical simulation, etc. (Simmers 1998; Scanlon et al.
2002; Nimmo et al. 2005). The limitation of these prior
methods is their inadequacy in analyzing large volumes of
hydrological data including precipitation, surface runoff,
evapotranspiration, etc. over a considerable time span as
reported in the study of Chandra et al. (2004). On the other
hand, considering the different mode of recharge, the output of these aforementioned methods are limited for
regional scale application. Though, according to the report
of Izuka et al. (2010), the potentials of geographic information system (GIS) have been explored in enhancing the
capability of some analytical methods, including soil water
budget for estimating the rate of groundwater recharge at
any spatial scale. However, this latter approach still
requires the input of large and diverse spatial data sets that
may not be available for regions under investigation or
period of interest. Exploring an empirical model that uses

123

the available data that are applicable for both local and
regional scales recharge estimate is the quest of this study.
According to Kumar (2000), empirical method can be used
to conveniently estimate groundwater recharge rate from a
few input variables that are relatively easy to obtain for
most regions. As such, the usefulness of empirical methods
in estimating groundwater recharge rate using the basic
theory of regression model has been documented in the
studies of Gontia and Patil (2012), Nolan et al. (2007),
Shuy et al. (2007), Chandra et al. (2004), and Rangarajan
and Athavale (2000). In accordance with the previous
studies including Misstear et al. (2009), Xi et al. (2008),
Kumar and Seethapathi (2002) and Rangarajan and Athavale (2000), the precipitation/rainfall variable was used as
the major input variable in the adopted rainfall recharge
model for the estimation of recharge rate in the investigated area. Consequently, this study will explore the concept of estimating recharge rate from precipitation/rainfall
data of the area using the rainfallrecharge relationship
model applicable for the area (Gontia and Patil 2012;
Yusoff et al. 2013). The determined recharge values will be
correlated with the interpreted geoelectric parameters
obtainable in the area. A multiple linear regression (MLR)
equation where the computed recharge rates (dependent
variable) and the interactive model regression between the
geoelectric parameters (layer resistivity and layer thickness) (independent variables) will form the underlying
recharge model. However, to actualize this objective,
detailed geophysical survey for hydrogeological evaluation
must be carried out.
The determination of geoelectrical parameters for
hydrogeological evaluation can only be mapped by subsurface investigation. The usefulness of the non-invasive
geophysical prospecting methods in delineating subsurface
layers and determining their geoelectric parameters has
been documented in the studies of Aizebeokhai et al.
(2010), Mogaji et al. (2011), and Oladapo et al. (2009).
Establishing also from the Mohamed et al.s (2012) report,
the geophysical techniques together with geological techniques have gained widespread acceptance in groundwater
exploration. As such, previous researchers have exploited
geoelectrical method among others to quantitatively estimate the water-transmitting properties of aquifers,
groundwater recharge, and so on (Mufid al-hadithi et al.
2006; Louis et al. 2004; Chandra et al. 2004; Cook et al.
1992; Barker 1990). In addition to this, the direct-current
(DC) electrical resistivity method has been reported by
researchers as a very powerful and cost-effective technique
in groundwater studies (Rubin and Hubbard 2005; Koefoed 1979; Jupp and Vozoff 1975). The lithological properties including resistivity, depth to water table, soil types,
water content, etc. which influence groundwater flow and
percolation to subsurface were easily mapped with this

Environ Earth Sci

geoelectrical method as emphasized in the priors studies.


The 2D resistivity imaging technique of geoelectrical
method which has been applied in various domains of
groundwater studies including groundwater pollution (Islami et al. 2012; Bahaa-eldin et al. 2011), salinity mapping
(Pujari and Soni 2009; Sathish et al. 2011), aquifer
potential mapping (Asry et al. 2012; Ewusi et al. 2009),
and aquifer parameter estimation (Niwas et al. 2011) was
efficiently explored for the used geophysical data in those
studies. However, little research has been conducted
exploring geophysical data in groundwater recharge rate
estimation such as the works according to Mufid al-hadithi
et al. (2006) and Chandra et al. (2004). The determined
geophysical parameters from the surface electrical method
acquired in those studies were correlated with the estimated recharge rates in the investigated area via regression
analysis. The application of regression model analysis in
modeling recharge equations has also been documented in
studies according to Nolan et al. (2007), Izuka (2006) and
Shuy et al. (2007) with appeal results. The output of their
recharge models is feasibly useful to provide independent
estimates of recharge.
In this study, we proposed modeling groundwater
recharge on regional scales by correlating the estimated
recharge rate due to rainfall with multiple lithological
parameters (layer resistivity and layer thickness) derived
from geophysical data. The empirical method was exploited
to develop the proposed multiple linear regression (MLR)

recharge model that can allow interactive modeling of


multiple factors that are significant for modeling recharge
rate with reasonable certainty. The proposed MLR recharge
model is different from the previous empirical approaches,
such as those of Chaturvedi (1973) and Kumar and
Seethapathi (2002), where recharge estimation was done
from rainfall data but lithological control on recharge was
ignored. Although, Mufid al-hadithi et al. (2006) and
Chandra et al. (2004) considered lithological control measures, however, the interactive significance of multiple
factors is overlooked. In our proposed MLR recharge
model, the recharge model depends simultaneously on two
factors namely resistivity of the material composition of
vadose zone and its thickness which is accordance with the
reports of the studies of Wang et al. (2008) and Nolan et al.
(2007). The developed MLR recharge model can yield more
reliable groundwater recharge rate estimates than the conventional methods. The proposed model may be applicable
in any other areas with similar geology.

Materials and methods


Geography, hydrology, and hydrogeology of the study
area
The study area is situated between the boundary of Perak
and Selango in Peninsular Malaysia. Figure 1 presents the

Fig. 1 Location of the study area in Peninsular Malaysia

123

Environ Earth Sci

2,884 km2 study area showing other important features.


The site lies between longitudes 10100 E and 101400 E and
between latitudes 3370 N and 4180 N in the southern part
of Perak. The area has four major rock types, namely,
QUA: Quaternary (mainly recent alluvium), DEV: Devonian (sedimentary rocks), SIL: Silurian (sedimentary rocks
with lava and tuff), and ING: acidic and undifferentiated
granitoids (Fig. 2). The region is characterized by an
equatorial maritime climate with nearly uniform air temperatures throughout the year. The average daily temperature is approximately 27 C, and relative humidity has a
monthly mean value of 62 and 78 % for the dry period and
peak of the rainy season, respectively. The regional topographic elevation variation in the area is in the range
792,131 m as extracted from the world topo map. The
general annual precipitation in Perak state ranges from 830
to 3,000 mm. However, the daily records of rainfall
amount for several locations within the study area for
periods of 10 years (20002010) were extracted from the
Tropical Rainfall Measuring Station (TRMM) database
acquired at 0.25 resolution (0.25) using MatLAB software and analyzed for this study (see Table 1).

Table 1 The estimated average annual rainfall amount and recharge


rate estimated in the area
Loc nos

Latitude

Longitude

Estimated average
rainfall amount
(mm/year)

Estimated
recharge rate
(mm/year)

418242.6

735417.6

1,382

249.66

428841.4

735594.2

1,082

220.59

455868.3

742483.4

2,083

460107.8

745663.1

829

192.7

511511.9

735770.9

1,294

241.5

401637.8

762974.4

1,430

254

429018.4

763504.4

1,150

227.51

456574.9

763327.7

2,161

312.77

467173.7

763857.7

920

203.17

10

511865.2

763151.1

1,315

243.47

11

414179.7

783818.7

1,510

261.08

12

429371.3

786097.0

1,248

237.12

13

453218.6

781168.0

2,168

313.28

14
15

461344.3
511688.6

775516.3
790707.9

992
1,345

211.09
246.26

307.03

2D resistivity imaging data acquisition, processing,


and interpretation

Fig. 2 Geological map of the area showing the rock types and 2D
locations

123

The 2D resistivity imaging data acquisition was carried out


with the use of modern field equipment, ABEM SAS 4000
which is a multi-electrode resistivity meter system highly
suitable for high-resolution 2D resistivity surveys. The
established 30 survey lines cutting across the diverse
geological settings in the area (see Table 2; Fig. 2) were
combed using both WennerSchlumberger array and Schlumberger array with the ABEM SAS 4000 system. Data
were recorded automatically along the profile at the electrode stations. The obtained data on each profile were
processed and inverted using the RES2DINV software
developed by Loke and Barker (1996). The software uses a
least squares optimization technique to invert the 2Dacquired apparent resistivity pseudosections to define true
resistivity distribution in the subsurface (Loke 2004; Sasaki
1992). Least squares optimization minimizes the square of
the differences between the observed and the calculated
apparent resistivity values. The program automatically
creates a 2D model by dividing the subsurface into rectangular blocks (Loke 2004), and the resistivity of the
blocks is iteratively adjusted to reduce the difference
between the measured and the calculated apparent resistivity values. The program calculates the apparent resistivity values and compares these to the measured data.
During the iteration, the modeled resistivity values are
adjusted until the calculated apparent resistivity values of
the model agree with the actual measurements. The iteration is stopped when the inversion process converges (i.e.,

Environ Earth Sci


Table 2 Summary of geophysical survey
Survey line

Geology

Array-type used

Loc 7, Loc 12, Loc 17, Loc 18 Loc 22, Loc24 and
Loc 23

QUA: Quaternary (mainly recent


Alluvium)

Wenner-Schlum and Schlumberger a = 5.0 m


e = 200 m, 400 m, d = 40 m, 80 m

Loc 19, Loc 20 Loc 21 Loc 25 and Loc 26

DEV: (sedimentary rocks)

Wenner-Schlum and Schlumberger, a = 5.0 m,


e = 200 m, 400 m d = 40 m, 80 m

Loc 1, Loc 2, Loc 5 Loc 8, Loc 9, Loc 10, Loc 11,


Loc 13, Loc 15 and Loc 16,

SIL: (sedimentary rock with


associated lava and tuff)

Wenner-Schlum and Schlumberger a = 5.0 m,


e = 200 m, 400 m d = 40 m, 80 m

Loc 3, Loc 4, Loc 6, Loc 14, Loc 27, Loc 28 Loc 29


and Loc 30

ING: Igneous rock (granitoid)

Wenner-Schlum and Sch lumberger, a = 5.0 m,


e = 200 m, 400 m d = 40 m, 80 m

a The electrode spacing, e the spread length, d the expected depth

either when the root mean square error falls to acceptable


limits or when the change between RMS errors for consecutive iterations becomes negligible). However, before
the geoelectric parameters were determined, the subsurface
layers were first delineated. It should be noted that the
boundary of lithology units in the subsurface is fuzzy in
nature where there is no clear demarcation boundary to
define the extent of each underlying lithologies as depicted
by the inverted 2D imaging sections (see Fig. 3). One of
the efficient technique of salvaging this challenge and
correctly interpreting these 2D imaging sections is by
constraining them with an in situ hydrogeological data
measurement obtainable from borehole logs. Hence, the
litho-logs of the available boreholes drilled within the
vicinity or on the 2D profiles were used as constraints that
guided the interpretation of the inverted 2D sections. We
achieved this by studying and interpreting the logs considering the borehole log description and Gamma log
lithology description for the various subsurface layers
mapping at varying depths. Thereafter, the subsurface
parameters such as resistivity of the overburden layer
(Unsaturated layer) and its thickness (Depth to aquifer top)
were determined. The delineated overburden layer is
regarded as the vadose zone (Unsaturated layer) in this
study. Since the geology of the study area is largely heterogeneous, the unsaturated (vadose zone) layers in most
cases are multilayered as depicted by the 2D sections
where resistivity section labeled (a) is located at 12; (b) is
located at 19; (c) is located at 9; (d) is located at 14; (e) is
located at 11 and (f) is located at 29. The resistivity of the
vadose zone (overburden layer) was estimated by first
saving the 2D section in XYZ format where the Z represents the resistivity parameters at varying depth Y.
Thereafter, the mean of the multilayered resistivity values
was estimated with reference to the depth Y of the
delineated aquifer top. Typical examples of the inverted 2D
resistivity sections showing various subsurface strata are
presented in Fig. 3af.
The interpretation shows that resistivity of the unsaturated layer in the area varies from 7 to 2,301 Xm. The

depth to aquifer top is also in the range 273 m (see


Table 3). The large variation in resistivity of the unsaturated layer generally indicates the varying nature of this
layer. These results suggest that there is existence of various heterogeneities which typifies the presence of lineaments, intrusive, differential weathering, fractured rocks,
and changes in the mineralogical composition of the rocks
constituting this soil formation column (Ewusi et al. 2009;
Chandra et al. 2008). The delineated unsaturated layer
which is a regolith column of soil formation lying just over
the aquifer has direct bearing on the moisture flux movement or recharge to aquifer. Hence, its lithological properties namely resistivity and depth to aquifer top as
discussed above were determined to remodel the groundwater recharge rate in the area.
Recharge estimation
The recharge rate estimation due to rainfall was carried out
using an existing model equation. This was done using one
of the empirical models of groundwater recharge namely the
UP Irrigation Research Institute, Roorkee (1954) [Chaturvedi formula] documented in Gontia and Patil (2012)
study. The ad hoc approach used by Yusoff et al. (2013) was
adopted to modify the renowned Chaturvedic formula to be
applicable in tropical zones area. The modification is in
agreement with the report of Kumar and Seethapathi (2002)
who said that this equation can be suitably altered for a
specific hydrogeological condition. The modified version of
the Chaturvedi formula is expressed as
R 6:75P  140:5

where R is the recharge due to precipitation during the


year, and P is the annual precipitation. Based on Eq. (1),
which can be referred as a rainfallrecharge model, the
recharge rate within the area was estimated and presented
in Table 1. Thereafter, the results of the estimated recharge
rate at various locations in Table 1 were processed with
kriging interpolation technique in GIS environment to
produce the potential recharge map shown in Fig. 4.

123

Environ Earth Sci

Aquifer layer

Unsaturated layer
Vadose zone

Unsaturated layer
Vadose zone

Aquifer layer

Unsaturated layer
Vadose zone
Aquifer layer
Fig. 3 Examples of 2D sections showing how geoelectrical layers were delineated

123

Environ Earth Sci

Aquifer layer

Unsaturated layer
Vadose zone

Unsaturated layer
Aquifer layer

Vadose zone

Unsaturated layer
Vadose zone

Aquifer layer

Fig. 3 continued

Geoelectric parametersrecharge relationship


To establish the influence of lithology control on the rate
of recharge due to rainfall, a relationship between the

recharge rate estimated and the determined geoelectric


parameters was established. Based on GIS analysis, the
actual estimated recharge rate and the corresponding
interpreted geoelectric parameters obtained for each 2D

123

Environ Earth Sci


Table 3 Results of the
estimated recharge rate and the
corresponding geoelectric
parameters

2D Loc. no

Easting

Northing

Recharge estimate

Resistivity of
unsaturated
layer (Xm) q

Depth to aquifer
top (m) [D]

Recharge values
(mm/year) [RE]
211.95

Loc 1

4.1,878

101.2168

94

Loc 2

4.2064

101.2983

178

237.43

Loc 3
Loc 4

4.2270
4.1920

101.3820
101.4540

50
126

3
13

203.57
225.27

Loc 5

4.0869

101.2458

445

257.98

Loc 6

4.1180

101.4180

1,202

35

278.37

Loc 7

4.0280

101.1338

359

15

256.61

Loc 8

4.0490

101.2376

391

20

259.48

Loc 9

4.0690

101.2416

472

25.1

258.62

Loc 10

4.0390

101.2956

559

18

265.39

Loc 11

4.0323

101.3675

645

23

243.03

Loc 12

3.9549

101.2017

219

15.4

252.88

Loc 13

3.9519

101.3629

363

15.7

250.04

Loc 14

3.8920

101.5230

201

14

251.08

Loc 15

3.7401

101.4558

355

18

242.57

Loc 16

3.7160

101.4830

334

15

247.14

Loc 17

3.7176

101.3295

408

15

228.37

Loc 18
Loc 19

3.6840
3.8720

101.3070
101.2740

312
222

15
9

260.04
243.24

Loc 20

3.7230

101.2340

316

15

243.93

Loc 21

3.7720

101.3040

180

11

255.37

Loc 22

3.9200

101.1440

86

12

245.45

Loc 23

3.7652

101.1270

73

248.96

Loc 24

3.8857

101.1078

251

20

221.83

Loc 25

3.9841

101.2658

532

12

250.54

Loc 26

3.7730

101.5817

141

6.2

257.9

Loc 27

3.8661

101.3327

562

25

233.99

Loc 28

4.0462

101.4455

562

38

267.85

Loc 29

4.0747

101.5551

2,301

301.74

Loc 30

4.1921

101.4984

95

5.7

220.19

resistivity location within the study area were determined. The results of the estimated recharge rate and the
corresponding geoelectric parameters values are presented
in Table 3. The obtained results in Table 3 were used to
generate linear graphs showing relationship of the estimated recharge values versus the unsaturated layer
resistivity values and thickness of the unsaturated layer
(depth to aquifer top) as shown in Fig. 5a, b. However, it
is important to note that we log the resistivity variable in
the regression equations, because the measured resistivity
values in the subsurface are often changes from low
magnitude to high magnitude. Besides, the complexity of
the subsurface is non-linear and requires the use of nonlinear equation to resolve the subsurface features correctly (Loke 2014). Therefore, computing the log values

123

Geoelectrical parameters

of the measured resistivity data is to enhance the linear


scaling representation of the resistivity data.
Multiple linear regression (MLR) recharge model
Consider the following generalized multiple linear regression model:
Y b0 b1 x1 b2 x2    bn xn i

where b0 is the intercept; b1 and b2 are the slopes of the


regression line with x1 and x2 , respectively; i is the error
terms; and Y is the dependent variable (response) as
reported in Koutsoyiannis (1977).
The estimated recharge (RE) values using the rainfall
recharge model (Eq. 1) were used as the dependent

Environ Earth Sci

Results and discussion


Sensitivity analysis result of the developed MLR
recharge model

(response) variable for the development of multiple linear


regression model in this study. The combination of resistivity of unsaturated layer (q) and thickness of the unsaturated layer (depth to aquifer top, D) was used as the
independent (predictors) variables.
Therefore, the multiple linear regression model in the
present phase is expressed as

The sensitivity analysis carried out on the developed MLR


recharge model enabled parameters significance evaluation. This analysis was carried out using the R software.
Table 4 presents the parameters evaluation results of the
developed MLR recharge models. This is in line with the
view of evaluating the essentiality of the independent
variables (predictors) in modeling the recharge estimate in
the area. The results in Table 4 show that the evaluated
resistivity and depth parameters have a significant relationship to the response variable (RE) at Pr B 0.05 (5 %)
in the area. This implies that the significance of both
predictors (resistivity of unsaturated layer q) and thickness
of the unsaturated layer (depth to aquifer top, D) at
probability Pr C 95 % can explain the RE in the MLR
recharge model (Eq. 4). Beside this, the computed statistically t test at a = 0.05 for both q and D gives the results
of the calculated values to be 9.27 and 7.09 representing
the predictors, respectively, which are lesser compared to
the tabulated values of 18.49. The latter results also confirmed the significance of the considered predictors in the
recharge model (Eq. 4). Hence, the considered geoelectrical parameters of varying influences and their interactive effect on recharge due to rainfall are significant for
estimating and predicting recharge rate in the study area.
This finding is in agreement with the reports of the studies
of Kumar and Ish (2012), Wang et al. (2008), and Nolan
et al. (2007). The RE model can reliably be used for
estimating and predicting recharge rate in the area if the
required geoelectric parameters in the study area are
known.

RE b0 b1 log10 q b2 D i :

Appraisal of model prediction accuracy

Fig. 4 Spatial distribution of recharge estimate using UPRI groundwater recharge model

The coefficients b0 , b1 and b2 were determined through


interactive model regression analysis of the records in
Table 3 using R software. From the interactive model
regression analysis, b1 = 25.83, b2 = 0.57 and b0 =
175.12 and R2 = 0.84 are obtained. By substituting the
obtained results in Eq. (3), the RE model becomes
RE 175:12 25:83 log10 q 0:57D:

Equation (4) is a multiple linear regression (MLR)


equation having (RE) as the dependent variable and (q) and
(D) as the multiple independent variables. Based on the
submission of Mazac et al. (1985) reported in the study
according Mufid al-hadithi et al. (2006), Eq. (4) is referred
to as a model. Hence, Eq. (4) is established as the multiple
linear regression (MLR) recharge model developed for the
study area.

The predictive power of the developed MLR recharge


model needs to be apprised to determine the feasibility of
using the RE model to predict and estimate groundwater
recharge in the area. Neil (1990) and Koutsoyiannis (1977)
suggested a systematic measure of accuracy for any forecast obtained from a model. This measure is called the
Theil inequality coefficient, which is given by
0
1
^ 2
n
X
y
y

i
@
A
5
K
^
y
l1
where yi is the actual estimated recharge rate observed in
the area, y^ is the corresponding predicted recharge rate of yi
from the RE model (see Table 5), and K the Theil
inequality coefficient.

123

Environ Earth Sci

a
Estimated recharge rate (mm/yr)

Fig. 5 Linear relationships


between recharge rate and
geophysical parameters

330
310
290
270
250
230

y = 1.1725x + 227.5

210

R 2 = 0.6477,R=0.8048

190
170
150

10

20

30

40

50

60

70

80

Depth to aquifer top (m)

Estimated recharge rate (mm/yr)

310
290
270
250
230

y = 37.556x + 156.59

210

R2 = 0.7541, R=0.8684

190
170
150
0

0.5

1.5

2.5

3.5

Log resistivity of unsaturated layer (Ohm-m)

Table 4 Parameter estimation analysis of the developed multiple linear regression (MLR) recharge model developed in the area
Developed recharge model

RE 175:12 25:83 log10 q 0:57D

Parameters

q
D

Standard error and parameters significance testing using


standard value of a at 5 % significance level

Remark: parameters
significance OK
at Pr \ 5 %

t value

Pr ([ijtj) value

4.5746

5.39 9 10-06

OK

-04

OK

3.757

8.39 9 10

q Resistivity of unsaturated layer (Xm), D depth to aquifer top (m)

Equation (5) was used to measure the prediction accuracy of the RE model. The determined Theil inequality
coefficient K value was then gaged by the critical value of
v2p ; a, where P = n - 1, n is the number of occupied 2D
locations, and a at 5 % significance level. The smaller the
value of K compared with the v2 -tabulated value, the better
the prediction accuracy of the model under investigation
(Neil 2003). The accuracy appraisal result of the MLR
recharge model is shown in Table 6. The result obtained
confirmed the reliability and accuracy of using the RE
model to predict groundwater recharge in the non-investigated part in the area. Therefore, the output of this recharge

123

rate predictive model can be harnessed for groundwater


resources evaluation and management in the study area.
Groundwater recharge rate estimation using the MLR
recharge model
The MLR recharge model in Table 6 was used to estimate
the mean groundwater recharge rate in the area. The
obtained result of the mean regional groundwater recharge
rate in the area was estimated to be 242.30 mm/year.
Thereafter comparisons of this result with the mean
groundwater recharge rate results obtained with the

Environ Earth Sci


Table 5 The records of the actual estimated recharge rate and the
predicted recharge rate
2D location
nos

Actual estimated
recharge rates observed
in the area yi

Predicted recharge
rates from the
RE model (^
y)

211.95

231.22

237.43

234.38

203.57

220.74

225.27

236.77

257.98

245.81

278.37

274.63

256.61

249.67

259.48

253.48

258.62

256.35

10

265.39

260.80

11

243.03

250.19

12

252.88

251.25

13

250.04

251.10

14
15

251.08
242.57

248.09
240.86

16

247.14

248.25

17

228.37

231.93

18

260.04

258.49

19

243.24

244.34

20

243.93

243.16

21

255.37

248.86

22

245.45

239.64

23

248.96

248.51

24

221.83

199.80

25

250.54

252.37

26

257.9

260.40

27

233.99

234.19

28

267.85

267.81

29
30

301.74
220.19

303.57
229.45

Fig. 6 Groundwater recharge prediction map of the area produced


from MLR recharge model

integrate the influences of multiple factors that have direct


bearing on moisture flux movement or recharge to aquifer
is thus established in this study with intrinsic property of
evaluating recharge rate with reasonable certainty.
Modeling groundwater recharge prediction
with the MLR recharge model

Table 6 MLR recharge model prediction accuracy analysis


S/n

Proposed MLR
recharge models

Nos of 2D
locations

v2p ; a 5 %

K value

RE 175:12
25:83 log10 q
0:57D

30

17.70

0.000423

q Resistivity of unsaturated layer (Xm), D depth to aquifer top (m)

rainfallrecharge relationship in Eq. (1) were analyzed.


The comparison result shows that an underestimation of
*5.78 mm/year was observed in the model (Table 1). This
difference may be due to the fact that the rainfallrecharge
model did not account for lithological factors in the model.
The use of recharge model that can simultaneously

Based on the records in Table 5, the predicted recharge rate


from the MLR recharge model was interpolated using
Kriging technique to produce the recharge rate prediction
map of the area (Fig. 6). It was observed from the model
map that the recharge rate for the study area varies between
199.81 and 303.55. The visual interpretation of Fig. 6 using
the legend zoning classes values shows that the eastern arm
of the area is mostly characterized with moderately high
recharge rate with few isolated patches of high, moderate
and moderate to low recharge rate. However, the areas with
moderate and few isolated patches of low to moderate and
moderately high recharge rate cover the southeastern parts
of the area. The western part and the central are mostly
covered with moderate and patches of moderately high
recharge rate. Whereas, the northern arm is found to be
overlain mostly with moderate to low and a noticeable low

123

Environ Earth Sci

recharge rate. The moderately high to high recharge rate


zones in the area are observed to be characterized by the
presence of porous and permeable unsaturated columns as
indicated by the layer resistivity. On the other hand, the
low to moderately recharge rate areas could be attributed to
the presence of hardpan or high clay content in a highly
weathered/low-resistive unsaturated layer. By hydrogeological implication, any aquifer units found associated with
moderately high to high recharge rate zones will be greatly
recharged for high groundwater potential due to direct
rainfall infiltration. Conversely, the aquifer units associated
with very low to moderate recharge rate zones can only be
potentially accumulated through indirect recharge from the
sources like stream, topographic depression, and spring for
producing good groundwater potential in the area (De
Vries and Simmers 2002). However, the very low to
moderate recharge zones are area characterized with high
protective capacity against impending groundwater contamination compared to the moderate high to high recharge
rate zones. Therefore, in terms of shallow groundwater
aquifers, the groundwater quality will be more protected in
the very low to moderate recharge zones area and vice
versa for other zones. Hence, the produced groundwater
recharge rate prediction map (Fig. 6) is a viable tool for
monitoring assessment of groundwater quality status which
can enhance groundwater resources management in the
area. In summary, the study area is underlain by both
unconfined and confined aquifers where water can mostly
be potentially accumulated through direct and indirect
modes of recharge (Scanlon et al. 2002; Nolan et al. 2007).
Although, the accuracy of the developed recharge model is
site specific; however, it can be reliably applied for
recharge rate assessment in other areas of similar geology
for the purpose of groundwater resources exploration and
management.

provided an excellent insight into assessing the varying


susceptibility of the underlain aquifers to potentially
recharge as well as its vulnerability to pollution in the area.
The information on this prediction map can serve as a
scientific basis for groundwater resource exploration and
management in the area. Furthermore, the proposed MLR
recharge model which was developed with variables from
multifaceted geologic settings can be used in any area with
similar geology for groundwater resource potential evaluation and groundwater quality status monitoring if the
required geoelectrical parameters are known.
Compared with other recharge estimation models, the
approach used in this study can provide a quick, independent, and cost-effective estimation of recharge by simple
geophysical measurement. Despite the advantages of the
proposed model, its recharge estimates should still be
corroborated by estimates from other methods because
multiple techniques are highly recommended in the estimation of any groundwater recharge. However, further
improvement on the accuracy of the MLR recharge model
can be achieved if the RE component of the model which
was obtained basically from climate data is re-evaluated
from a natural groundwater recharge in situ measurement
using an injected tracer technique in the area. In the same
hand, more predictor variable can be evaluated by carrying
out a survey to determine the water-level fluctuation
measurement from groundwater wells in the area at two
different seasons. Such water-level fluctuation parameter
which has a direct relationship with recharge rate of an area
can form an additional independent variable component of
the model to enhance the future accuracy output of this
recharge model.
Acknowledgments This project was carried out using the financial
support from RUI, Investigation Of The Impacts Of Summertime
Monsoon Circulation To The Aerosols Transportation And Distribution In Southeast Asia Which Can Lead To Global Climate Change,
1001/PFIZIK/811228.

Conclusion and future works


In this study, groundwater recharge rate assessment was
adequately evaluated on regional scale using a developed
MLR recharge model. The newly proposed recharge model
was based on relating recharge estimated from a rainfall
recharge model to geoelectric parameters interpreted from
2D resistivity imaging acquired in the area. Sensitivity and
prediction accuracy analyses of the newly proposed models
using t test and v2 distribution at a = 0.05 significance
level were conducted using R software. The MLR recharge
model was used to estimate recharge rate and to produce
groundwater recharge rate prediction map using GIS
techniques for the area. The regional recharge rate in the
area was estimated to be 242.30 mm/year. The regional
groundwater recharge rate prediction map produced

123

References
Aizebeokhai AP, Olayinka AI, Singh VS (2010) Application of 2D
and 3D geoelectrical resistivity imaging for engineering site
investigation in a crystalline basement terrain, Southwestern
Nigeria. Environ Earth Sci 61:14811492. doi:10.1007/s12665010-0464
Akpan AE, Ugbaja AN, George NJ (2013) Integrated geophysical,
geochemical and hydrogeological investigation of shallow
groundwater resources in parts of the Ikom-Mamfe Embayment
and the adjoining areas in Cross River State, Nigeria. Environ
Earth Sci 70:14351456. doi:10.1007/s12665-013-2232-3
Asry Z, Samsudin AR, Yaacom WZ, Yaakub J (2012) Groundwater
investigation using electrical resistivity imaging technique at Sg.
Udang, Malaka, Malaysia. Bull Geol Survey Malays 58:5558
Bahaa-eldin EAR, Yusoff I, Abdul Rahim SWY, Wan Zuhairi MR,
Ghani A (2011) Tracing subsurface migration of contaminants

Environ Earth Sci


from an abandoned municipal landfill. Environ Earth Sci
63:10431055. doi:10.1007/s12665-010-0780-3
Barker R (1990) Investigation of groundwater salinity by geophysical
methods. Geotech Environ Geophys 201212. SBN:978-156080-278-5 print ISBN: 978-0-931830-99-0
Chandra S, Chand R, Rao VA, Singh VS, Jain SC (2004) Estimation
of natural recharge and its dependency on sub-surface geoelectric parameters. J Hydrol 299(1):6783. doi:10.1016/j.jhydrol.
2004.04.001
Chandra S, Kumar D, Ahmed S, Perrin J, Dewandel B (2008)
Contribution of geophysical methods in exploration and assessment of groundwater in hard rock aquifers. The 3rd International
Conference on Water Resources and Arid Environments (2008)
and the 1st Arab Water Forum 299 (2004), pp 6783
Chaturvedi RS (1973) A note on the investigation of ground water
resources in Western Districts of Uttar Pradesh. Annual Report,
UP Irrigation Research Institute, pp 86122
Cook PG, Walker GR, Buselli G, Potts IAR (1992) The application of
electromagnetic techniques to groundwater recharges investigations. J Hydrol 130(14):201229
De Vries JJ, Simmer I (2002) Groundwater recharge: an overview of
processes and challenge. Hydrogeol J 10:517. doi:10.1007/
s10040-001-0171-7
Doumouya I, Dibi B, Kouame KI, Saley B, Jourda JP, Savane I,
Biemi J (2012) Modeling of favourable zones for the establishment of water points by geographical information systems (GIS)
and multi-criterial analysis (MCA) in the Aboisso area (Southeast of Cote dIvoire). Environ Earth Sci 67(6):17631780.
doi:10.1007/s12665-012-1622-2
Ewusi A, Kuma JS, Voigt HJ (2009) Utility of the 2-D multi-electrode
resistivity imaging technique in groundwater exploration in the
Voltaian sedimentary basin, Northern Ghana. Nat Resour Res
18(4):267275
Gontia NK, Patil PY (2012) Assessment of groundwater recharge
through rainfall and water harvesting structures in Jamka micro
watershed using remote sensing and GIS. J Indian Soc Remote
Sens 40(4):639648. doi:10.1007/s12524-011-0176
Islami N, Taib SH, Yusoff I, Ghani AA (2012) Integrated geoelectrical resistivity, hydrochemical and soil property analysis
methods to study shallow groundwater in the agriculture area,
Machang, Malaysia. Environ Earth Sci 65:699712
Izuka SK (2006) Effects of irrigation, drought, and ground-water
withdrawals on groundwater levels in the southern Lihue Basin,
Kauai, Hawaii. US Geological Survey Scientific Investigations
Report 2006-5291, p 42
Izuka SK, Oki DS, Engott JA (2010) Simple method for estimating
groundwater recharge on tropical islands. J Hydrol
387(2010):8189. doi:10.1016/j.jhydrol.2010.03.034
Jupp DLB, Vozoff K (1975) Joint inversion of geophysical data.
Geophys J Royal Astron Soc 42(977991):195
Jyrkama MI, Sykes JF (2007) The impact of climate change on spatial
varying groundwater recharge in the grand river watershed
(Ontario). J Hydrol 338:237250. doi:10.1016/j.jhydrol.2007.02.
036
Kaliraj S, Chandrasekar N, Magesh NS (2014) Idnetification of
potential groundwater recharge zones in Vaigai upper basin,
Tamil Nadu, using GIS-based analytical hierarchical process
(AHP) technique. Arab J Geosci 7:13851401. doi:10.1007/
s12517-013-0849-x
Koefoed O (1979) Geosounding principles 1: resistivity sounding
measurements. Elsevier Science Publishing Co, Amsterdam
Koutsoyiannis A (1977) Theory of econometrics, 2nd edn. Palgrave,
New York
Kumar CP (1977) Estimation of natural ground water recharge. ISH J
Hydraulic Eng 3(1):6174

Kumar CP (2000) Groundwater assessment methodology. National


Institute of Hydrology, Roorkee, p 21
Kumar CP, Ish M (2012) Estimation of ground water recharge due to
rainfall by modeling of soil moisture movement. The Indian
Society for hydraulics. J Hydraulic Eng. doi:10.1080/09715010.
1998.10514633. http://www.tandfonline.com/loi/tish20
Kumar CP, Seethapathi PV (2002) Assessment of natural ground water
recharge in upper Ganga Canal command area. J Appl Hydrol
15(4):1320. http://www.angelfire.com/nh//publication/ugcm.pdf
Leipnik RB, Loaiciga HA (2006) Radially convergent groundwater
flow in sloping terrain. Hydrol Sci J (des Sciences Hydrologiques) 51(4):700712. doi:10.1623/hysj.51.4.700
Loke MH (2004) Geoelectrical imaging 2-D & 3-D. Malaysia,
Geotomo Software, p 133
Loke MH (2014) Inversion and interpretation of multi-dimensional
resistivity surveys. Geotomo Software, Malaysia
Loke MH, Barker RD (1996) Rapid least-squares inversion of
apparent resistivity pseudosections by a Quasi-Newton method.
Geophys Prospect 44:131152
Louis I, Karantonis G, Voulgaris N, Louis F (2004) Geophysical
methods in the determination of aquifer parameters: the case
of Mornos river delta, Greece. Res J Chem Environ
18(4):4149
Mazac O, Kelly WE, Landa I (1985) A hydrogeophysical model for
relation between electrical and hydrualic properties of aquifers.
J Hydrol 79:19
Misstear BDR, Brown L, Daly D (2009) A methodology for making
initial estimates of groundwater recharge from groundwater
vulnerability mapping. Hydrogeol J 17(2):275285
Mogaji KA, Omosuyi GO, Olayanju GM (2011) Groundwater system
evaluation and protective capacity of overburden material at IleolujI, Southwestern Nigeria. J Geol Min Res 3(11):294304
Mohamed SEJ, Shaharin I, Wan NAS, Puziah AL (2012) Groundwater resources assessment using integrated geophysical techniques in the southwestern region of Peninsular Malaysia. Arab J
Geosci 6(11):41294144. doi:10.1007/s12517-012-0700-9
Mufid Al-Hadithi DCS, Israil M, Kumar B (2006) Groundwaterrecharge estimation using a surface electrical resistivity method
in the Himalayan foothill region, India. Hydrogeol J 14:4450.
doi:10.1007/s10040-004-0391-8
Neil HT (1990) Multivariate analysis with application in education
and psychology, 2nd edn. Wiley, New York
Neil HT (2003) Multivariate analysis with application in Education
and Psychology, Revised edition. Wadsworth publishing Company Inc, Belmont
Nimmo JR, Healy RW, Stonestrom DA (2005) Aquifer recharge. In:
Anderson MG, Bear J (eds) Encyclopedia of hydrological science,
vol 4. Universiti Sains Malaysia, Wiley, UK, pp 22292246
Niwas S, Tezkan B, Israil M (2011) Aquifer hydraulic conductivity
estimation from surface geoelectrical measurements for Krauthausen test site, Germany. Hydrogeol J19(2):307315. doi:10.
1007/s10040-010-0689-7
Nolan BT, Healy RW, Taber PE, Perkins K, Hitt KJ, Wolock DM
(2007) Factors influencing ground-water recharge in the eastern
United States. J Hydrol 332(12):1187. http://dx.doi.org/10.
1016/j.jhydrol.2006.06.029
Oladapo MI, Adeoye-Oladapo OO, Mogaji KA (2009) Hydrogeophysical study of the groundwater potential of Ilara-Mokin
Southwestern Nigeria. Glob J Pure Appl Sci 15(2):195204
Omorinbola EO (2009) Empirical equations of groundwater patterns/
equations. Hydrol Sci J 211. http://www.tandfonline.com/loi/
thsj20
Pujari PR, Soni AK (2009) Sea water intrusion studies near Kovaya
limestone mine, Saurashtra coast, India. Environ Monit Assess
154:93109. doi:10.1007/s10661-008-0380-9

123

Environ Earth Sci


Rangarajan R, Athavale RN (2000) Annual replenishable groundwater potential of India-an estimate based on injection tritium
studies. J Hydrol 234:3853
Robins NS (ed) (1998) Ground-water pollution aquifer recharge and
vulnerability. Geological Society, London, Special Publications
130, pp 107115
Rubin Y, Hubbard S (2005) Hydrogeophysics. Springer, Dordrecht,
p 523
Sasaki Y (1992) Resolution of resistivity tomography inferred from
numerical simulation. Geophys Prospect 40:453463
Sathish S, Elango L, Rajesh R, Sarma VS (2011) Assessment of
seawater mixing in a coastal aquifer by high resolution electrical
resistivity tomography. Int J Environ Sci Tech 8(3):483492
Scanlon BR, Healy RW, Cook PG (2002) Choosing appropriate
techniques for quantifying groundwater recharge. Hydrogeol J
10:1839. doi:10.1007/s10040-001-0176-2
Shuy BE, Tan SBS, Chua CHL (2007) Regression method for
estimating rainfall at unconfined sandy aquifers with an

123

equatorial climate. Hydrol Process 21:35143526. doi:10.1002/


HYP.6552
Simmers (1998) Groundwater recharge: an overview of estimation
problems and recent developments. Geological Society, London, Special Publications 1998, vol 130, pp 107115. doi:10.
1144/GSL.SP.1998.130.01.10
Wang I, Jin M, Nimmo JR, Yang l, Wang W (2008) Estimating
groundwater recharge in Hebei Plain, China under varying land
use practices using tritium and bromide tracers. J Hydrol
356(12):209222
Xi C, Zhang Z, Zhang X, Chen Y, Qian M, Peng S (2008) Estimation
of recharge from precipitation and evapotranspiration by
Lysimeter measurement and soil moisture model. J Hydrol Eng
13(50):333340
Yusoff I, Saghravani SR, Mustapha S, Saghravan SF (2013)
Estimating groundwater recharge using empirical method: a
case study in the Tropical Zone. Sains Malaysiana
42(5):553560

You might also like