Environment, Development and Sustainability (2024) 26:16267–16287

https://doi.org/10.1007/s10668-023-03237-6

CASE STUDY

Statistical analysis of rainfall and groundwater interaction

in Bhadra catchment

H. S. Nanditha1,2 · T. V. Reshmidevi1,2 · L. Udaya Simha1,2 · Parthan Kunhikrishnan1,2

Received: 8 August 2022 / Accepted: 4 April 2023 / Published online: 19 April 2023
© The Author(s), under exclusive licence to Springer Nature B.V. 2023

Abstract
Interaction between rainfall and groundwater is a complex phenomenon as it varies
spatially with respect to local hydrogeological characteristics and temporally with hydro
meteorological variations. To understand the resilience of the groundwater system to the
changing climate, rainfall-groundwater interaction needs to be analysed in detail. The
Rainfall and Time Trends (HARTT) have been widely used in recent studies to model
the impact of the rainfall and the groundwater level fluctuation. However, not many such
studies have been reported in tropical hilly areas. Bhadra reservoir catchment is a tropical
hilly catchment in India which is characterised by heavy rainfall, steep topography and
shallow water table conditions and significant subsurface flow conditions. The area is
largely dependent on groundwater for irrigation and drinking water supply. Nonparametric
statistical methods such as Mann–Kendall (MK) test, Sen’s slope estimator and Sequential
Mann–Kendall Test (SQMK) are used to study the temporal trends in the rainfall and static
groundwater levels. Further, contribution of rainfall to groundwater level fluctuation is
analysed using HARTT model. Study shows a decreasing trend in rainfall and groundwater
level in the upper catchment. Further, 1 mm increase or decrease in the rainfall anomaly is
estimated to cause 0.0012 m decrease or increase in the depth to water table. Only 36% of
the groundwater level fluctuation is explained by rainfall induced recharge; nevertheless,
the relation is non-uniform throughout the catchment. Spatial variation in topography
and hydrogeological conditions is suspected to have significant impact on the rainfall and
groundwater interaction leading to poor coefficient of determination for the regression
model between rainfall anomaly and the groundwater level fluctuations.

Keywords Mann–Kendall (MK) test · Sen’s slope estimator · Sequential Mann–Kendall

test (SQMK) · Hydrograph analysis: rainfall and time trends (HARTT) · Tropical
catchment · Rainfall and groundwater interaction

* H. S. Nanditha
[email protected]
Extended author information available on the last page of the article

13
Vol.:(0123456789)
16268 H. S. Nanditha et al.

1 Introduction

Groundwater, though is only 13% of the total volume of fresh water in the hydrosphere,
contributes to almost 15% of the annual water consumption (Dragoni & Sukhija, 2008).
It is an important source of irrigation and drinking water supply, especially in the rural
areas. On an average, desert countries depend 90% on groundwater for irrigation, whilst
industrialised countries like France and Japan accounts for about 40% of groundwater.
Studies show that in the USA, groundwater share is about 25% of the total freshwater
withdrawal. In India, about 52% of the irrigation consumption and 85% of the rural water
supply are sourced by the groundwater (Panwar & Chakrapani, 2013). Groundwater
aquifers receive recharge from the precipitation and the interaction with the surface water
bodies. Any changes in the hydrologic cycle are expected to have an impact on the recharge
and hence the groundwater storage. Global warming and the projected changes in the
climate are observed to have profound impact on precipitation and evapotranspiration.
These components of the hydrological cycle directly affect the groundwater recharge
(Woldeamlak et al., 2007). Several studies have been done in the recent past focusing
on the impact of climate change on the surface water bodies (Chanapathi et al., 2018;
Overpeck & Udall, 2010; Reshmidevi et al., 2018; Seager et al., 2013; Vano et al., 2015)
and on the groundwater (Almedeij & Al-Ruwaih, 2006; Panwar & Chakrapani, 2013).
The historical relation between precipitation and the groundwater recharge indicates the
resilience of groundwater under the changing climate scenario and can be the basis for
understanding the climate change impact on the groundwater (Cuthbert et al., 2019; Fu
et al., 2019). Studies from different countries across the globe show the strong correlation
between rainfall and the groundwater recharge (Green et al., 2011; Ngongondo, 2006).
Detailed study on the precipitation and groundwater recharge has shown that the
relation between these two is not consistent. Response of a groundwater system to the
major climate variables like precipitation is slow compared to the surface water system,
generally resulting in temporal lag of a few months to a few years between the two (Green
et al., 2011; Hanson et al., 2004; Russo & Lall, 2017; Zhang et al., 2019). Studies show
that not only the present rainfall, but the antecedent or accumulated rainfall also influences
the groundwater (Jan et al., 2007). Whilst the groundwater hydrodynamic behaviour
is dependent on the rainfall anomaly, it is also influenced by other factors such as land
use and land cover, soil type, evaporation, storage, sea level fluctuations, distribution,
amount and timing of precipitation (Dong et al., 2012; Fu et al., 2019; Taminskas et al.,
2018). Amongst these factors, a large number of studies are focused on the response of
the groundwater levels to the seasonal precipitation (Lasagna et al., 2019; Lutz et al.,
2015; Tian & Liu, 2011). It is also well understood that the recharge varies with the type
of aquifer and the hydrogeology (Cuthbert et al., 2019; Green et al., 2011; Panwar &
Chakrapani, 2013). Different methods used for understanding the groundwater dynamics
and its relationship with the climate factors include numerical models, stochastic models
viz., ARMA, ARIMA models and statistical methods involving the time series analysis.
Numerical models are data intensive to simulate the amount and quality of groundwater
resources. However, collecting such information, particularly in developing nations, is
difficult. In the absence of detailed spatial data, statistical method is the better alternative.
Statistical methods are simple, yet effective methods to understand the long-term trends of
groundwater and its dependency on the climate factors (Reghunath et al., 2005), especially
when there is limited data available on the catchment properties (Yihdego et al., 2017).
Hence, time series analysis has been used in various research to understand the relationship

13
Statistical analysis of rainfall and groundwater interaction… 16269

between groundwater level fluctuations and climatic and non-climatic driving variables
(Crosbie et al., 2005; Zhang et al., 2019). Time series regression with least square error
method (Shalini, 2012) is one of the simplest statistical methods used to assess the
trends and interdependence of groundwater level and rainfall. The effect of seasonal
rainfall may be analysed by incorporating the seasonal-trend decomposition procedure to
the linear regression model (Zhang et al., 2019). Nonparametric trend analysis methods
such as Mann–Kendall test, Sen’s slope estimator (Mann, 1945), Mann–Whitney Pettit
and Mann–Whitney–Wilcoxon statistics (Ngongondo, 2006) are also widely adopted in
groundwater studies.
The Mann–Kendall (MK) test is a simple, and widely used method to identify the
monotonous trends in the time series data, and is effective even with time series of shorter
length (Pathak & Dodamani, 2019; Sonali & Nagesh Kumar, 2013; Vousoughi et al., 2013).
Sen’s slope estimator is used to determine the magnitude of the linear trend observed in the
data (Jan et al., 2007). These methods have been widely used to study the trends present
in various hydrological variables including precipitation and groundwater level (Jan et al.,
2007; Mack et al., 2013; Shamsudduha et al., 2009; Vousoughi et al., 2013; Zhang et al.,
2009). The sequential version of Mann–Kendall test statistic is used to identify the change
points in long-term time series. If the significant trend is present in the series, the point
of intersection of the prograde series and the retrograde series indicates the trend change
point (Patakamuri et al., 2020). These methods are used to study the trend in rainfall and
groundwater separately.
A few studies have made an attempt to simulate interaction between rainfall and
groundwater using numerical models like MODFLOW (Patil & Chetan, 2017). However,
these are highly data intensive which restrict their application. When there is insufficient
information on the physical and hydrogeological information about a catchment, statistical
methods are better alternative. Stochastic approaches like Autoregressive Moving Average
(ARMA) and ARIMA models are used to model the groundwater level fluctuations based
on the observed groundwater level (Gibrilla et al., 2018; Goodarzi, 2020). Limited studies
are available to say how the groundwater level is varying with respect to rainfall using
statistical models. A few studies have focused on the long-term trend in precipitation and
its impact on the groundwater levels fluctuations (Cai et al., 2020; Petpongpan et al., 2020).
A statistical method widely used to study the groundwater level fluctuations and
to separate the effects of the driving forces is Hydrograph Analysis: Rainfall and Time
trends (HARTT) model proposed by Ferdowsian et al. (2001). HARTT is the linear
regression model which uses the regression coefficients and the R2 values to establish the
relationship between groundwater, rainfall and the time lag between rainfall and its impact
on groundwater. The model explains the groundwater levels without adopting any random
non-climatic temporal trend (Yihdego et al., 2017). A few studies are reported from
Australia (Ferdowsian et al., 2001, 2002; Leong & Yokoo, 2019; Yihdego et al., 2017),
and the results are consistent with hydrogeological expectation (Yihdego et al., 2017).
A few studies using the HARTT model are reported from other countries as well. (Zeru
et al., 2020) used the HARTT model to determine the contribution of climatic and non-
climatic stresses on groundwater levels in the Lake Haramaya well-field, Ethiopia. It was
observed that variability in rainfall explained 81.3% of groundwater level using a 2-month
average time-delay. HARTT model can thus be used to differentiate the groundwater level
fluctuation over time from unit change in rainfall scenario.
In India, several studies have focused independently on rainfall and groundwater (Aslam,
2018; Jain & Kumar, 2012; Patakamuri et al., 2020; Salehi et al., 2020). However, not
many studies have modelled the groundwater level fluctuations using rainfall information.

13
16270 H. S. Nanditha et al.

Therefore, in this study, an attempt has been to model the groundwater level fluctuations
from the rainfall data using the HARTT model for tropical hilly catchment in India.

2 Study area and data description

Bhadra river originates from the Gangamoola in the Western Ghats in the peninsular India.
It flows around 190 km through the state of Karnataka before joining the river Tunga to
form river Tunga-Bhadra, which is one of the major tributaries of river Krishna. Bhadra
dam with a gross storage capacity of 71.60 thousand million cubic feet (TMC) was com-
missioned across river Bhadra, near Bhadravathi and Tarikere, in the year 1965 (Water
Resources Department, 2021). Catchment area of the Bhadra reservoir is spread across
2048 ­km2 in the Chickmagalur district in Karnataka. In addition to providing irrigation
water for a gross command area of 162,818 ha, the project also generates 40.4 MW of
hydropower. The catchment is in the tropical humid region with annual average rainfall
varying from 3500 mm at the hilly areas to nearly 1500 mm towards the catchment outlet.
Almost 80% of this rainfall is received during the 6 months from June to November (Jayas-
ree, 2008). The catchment is distinguished by steep and undulating terrain and a high sub-
surface flow condition. Shallow wells (dug wells) and bore wells are used to extract ground
water for irrigation throughout the catchment. In this study, the relationship between the
rainfall and the groundwater level fluctuations in the Bhadra reservoir catchment area is
analysed using statistical methods. The area lies between Latitudes 13.124 and 13.750°N
and Longitudes 75.157 and 75.750° E. Location map of the study area is shown in Fig. 1.

Fig. 1  Location map of the Bhadra reservoir catchment area with Location of the rainfall grid points and
Groundwater observation wells

13
Statistical analysis of rainfall and groundwater interaction… 16271

Various spatial information required for the hydrological analysis of the area is assimi-
lated from various agencies and is shown in Fig. 2. Catchment boundary and the stream
network of the area are generated using Cartosat-1 (V.3R1) Digital Elevation Model
(DEM) with 32 m spatial resolution (Fig. 2a), downloaded from the Bhuvan geoportal of
Indian Space Research Organisation (ISRO). Bhadra reservoir catchment area is character-
ised by the undulating hilly terrain with elevation varying from 1880 m (above mean sea
level) in the upper catchment to around 600 m near the reservoir. The average elevation of
the Bhadra catchment is about 900 m above MSL, and average slope is 6%. Land use/land
cover map of the area at 1:50,000 scale is also obtained from the Bhuvan portal for the
period of 2015–16 (Fig. 2b). Predominant land cover in the area is dense forest followed by
plantation crops. Major crops cultivated in the area include coffee, pepper, cardamon and
arecanut in the hilly zones and paddy in the plains within the hilly zones. The soil map of
the study area is downloaded from FAO (Food and Agricultural Organisation, UN) depart-
ment (Fig. 2c), which shows Acrisols and Nitosols as the predominant soil type present in
the catchment. These are typically sandy clay loam type of soil with more than 50% sand
content in the top layer of thickness 300 mm and below a deep, clayey soil. Lithology
map of the area is obtained from the web enabled Water Resources Information System,
(India-WRIS), of the Central Water Commission. The major geological formation in the
area is of Gneiss followed by Schist as shown in Fig. 2d. The aquifer system in the area is
an unconfined aquifer with an average thickness of 35 m. The top layer is a weathered zone

Fig. 2  Spatial information of the catchment a DEM b Land use/Land cover Map c Soil map d Lithology
map

13
16272 H. S. Nanditha et al.

that extends 10–15 m below ground level followed by a fractured zone. Transmissivity var-
­ 2/day.
ies from 2 to 100 m
Daily rainfall data for the study area are collected from the Indian Meteorological
Department (IMD) at grid points in and around the catchment area at a spatial resolution of
0.25 × 0.25 degrees for a period of 24 years (1995–2018). The six grid points considered in
this study are marked in Fig. 1. Monthly static groundwater levels from 5 observation wells
are collected from the Central Groundwater Board for the period 1995–2018. Locations
of these observation wells are shown in Fig. 1. Dug wells having 24 years of data are
considered for the study with highest depth to groundwater level of 9.01 m recorded at
Niduvale and lowest 1.92 m at Kalasa.
The main occupation is agriculture in the Bhadra river catchment. Irrigation and
drinking water supply is sourced by the groundwater. Due to intensive agricultural activity
and a rise in demand for domestic water supply, groundwater resources are drastically
declining. Therefore, in this study, attempts are made to analyse the impact of rainfall
variations on the groundwater table levels in the Bhadra catchment.

3 Methodology

The proposed methodology consists of 2 steps.

Step 1: Identify potential temporal pattern and starting point of the changes.
Step 2: Modelling groundwater level as a function of rainfall.

3.1 Mann–Kendall test

Mann–Kendall (MK) test is a nonparametric statistical method (Kendall, 1975) used to

identify the monotonic trends present in time series data. The trend may be monotonically
increasing or decreasing, but need not be linear. It follows the principles of hypothesis
analysis. The null hypothesis (H0) is assumed as the presence of no trend in the time series
data. Hence, all the data from the population are assumed to be random variables that
are independent and identically distributed. Accordingly, the alternate hypothesis (H1) is
assumed to be the presence of an increasing or decreasing trend in the data. The hypothesis
is tested using the standard normal test statistics Zc, which is computed as shown in Eq. 1
(Vousoughi et al., 2013):

⎧ √S−1 when S > 0

⎪ Var(S)
Zc = ⎨ 0 when S = 0 (1)
⎪ √S+1 when S < 0
⎩ Var(S)

where S is the MK test statistics calculated using Eq. 2. Each observation in the time series
(xj) is compared with observations in all the previous time steps (xi). Thus, if a time series
has n number of observations, n (n−1)/2 numbers of data pairs are generated. In each of
these pairs, the data observed at the later time period are compared with the data observed
in the earlier time period using a Sign function as given in Eq. 3:

13
Statistical analysis of rainfall and groundwater interaction… 16273

∑
n−1
∑
n
S= . sign(xj − xi ) (2)
i=1 j=i+1

⎧ +1 when �x − x � > 0
� � ⎪ � j �i
Sign xj − xi = ⎨ 0 when x�j − xi �= 0 (3)
⎪ −1 when xj − xi < 0
⎩

For number of data points n greater than 10, S can be approximated as normally
distributed with mean 0 and variance as given in Eq. 4 (Kendall, 1975):
[ ]
1 ∑ q
( )( )
VAR(S) = n(n − 1)(2n + 5) − tp tp − 1 2tp + 5 (4)
18 p=1

where q is the number of data groups that are tied in the data series, and tp is the number of
ties for the xi value of the pth tied group.
A positive or negative value of Zc indicates an increasing or decreasing trend,
respectively (Zhang et al., 2009). A two-tailed distribution is assumed for Zc, and the
hypothesis is tested for the selected level of significance α. The alternate hypothesis (H1)
is accepted if the absolute value of Zc is more than the Z1−α/2 obtained from the standard
normal cumulative probability table. Generally, in hydrological studies, α is taken between
5 and 10% (Patakamuri et al., 2020; Vousoughi et al., 2013). Failure to accept the alternate
hypothesis does not indicate the absence of any trend, nevertheless, the trends, if any, are
not statistically significant at the threshold significance level α (Meals et al., 2010).
Most of the hydrological studies, using the MK tests, are based on the assumption that
observed data are serially independent. However, presence of serial correlation in time
series can lead to the false positive outcome from the hypothesis testing, particularly when
the MK test is used (A Kulkarni & Storch, 1995a, 1995b; Storch, 1995; Yue et al., 2002).
Hence, in this study, before applying the MK test, pre-whitening method, suggested by
Storch (1995), is applied to remove the effect of lag-1 serial correlation, significant at the
95% confidence level (Hamed, 2009; Marofi et al., 2012; Sayemuzzaman & Jha, 2014).

3.2 Sen’s slope estimator test

Sen’s slope estimator (bsen) is used to determine the actual change per unit time (magnitude
of the trend) for rainfall and groundwater level data (Patakamuri et al., 2020). In this
method, each data of the time series is compared with all the earlier observations and the
slope (bi) of the line connecting these two observations are determined as given in Eq. 5:
xj − x k
bi = for i = 1, 2, … , N (5)
j−k

where xj and xk are the observations at time period j and k, respectively, for which j > k and
N is the numbers of data pairs in each time series.
Sen’s slope estimator bsen is the median of these bi values calculated from the data series.
Positive or negative values of bsen indicate the magnitude of a monotonously increasing or
decreasing trend, respectively.

13
16274 H. S. Nanditha et al.

3.3 Sequential Mann–Kendall test

The sequential form of the MK (SQMK) test is a useful tool for determining the beginning
of any increasing or decreasing trend in the time series. The series are arranged in the rank
order to calculate the test statistic for prograde series and retrograde series and are plotted
(Patakamuri et al., 2020). If any significant trend is present in the data, these two curves cross
each other, and the point of intersection presents the point of change.

3.4 HARTT model

In HARTT analysis, the underlying multiple linear regression model is used to establish the
relationship between the groundwater level and a set of three variables viz., accumulative
residual rainfall, time and a model dependent third variable (Ferdowsian et al., 2001; Leong &
Yokoo, 2019; Zeru et al., 2020). The main aim of this model is to determine the contribution
of precipitation and non-climatic factors on groundwater levels (Leong & Yokoo, 2019).
The impact of rainfall on groundwater is calculated by representing rainfall as an
Accumulation of deviations from Average Rainfall (ARR). Accumulative Monthly Residual
Rainfall (AMRR) and Accumulative Annual Residual Rainfall (AARR) are the two forms of
ARR which are used in the model and are calculated as given in Eq. 6 and 7 (Ferdowsian
et al., 2001, 2002):
∑
t
AMRRt = (Mi,m − M m ) (6)
i=1

( )
∑
t
Ā
AARRt = Mi− (7)
i=1
12

where Mi,m is rainfall (in mm) in mth month of a year, and i is the time in months since the
start of the data set. M m is mean monthly rainfall (in mm) for the mth month of the year, and
t is total months since the start of the data set. A̅ is mean annual rainfall (in mm).
Finally, a relation between ARR and groundwater depth ­ (Deptht) is developed using
regression model (Ferdowsian et al., 2001). ARR at different lag time is used in the regression
model as shown in Eq. 8:
Deptht = K0 + K1 ∗ ARRt−L1 + K2 t (8)
where L1 is length of time lag in months between rainfall and its impact on groundwater
level. The parameters K0, K1 and K2 indicate initial depth to groundwater level, the impact
of ARR on the groundwater level and groundwater trend over time, respectively. The
coefficient of determination (R2) explains the degree of fit between estimated and recorded
water levels (Ferdowsian & Pannell, 2009; Zeru et al., 2020).

4 Results and discussion

Daily gridded rainfall data and the static groundwater level data (depth below ground
level in metres) from the study area are analysed to find the relation between rainfall and
groundwater level fluctuations. MK test and Sen’s slope are used to detect the statistically

13
Statistical analysis of rainfall and groundwater interaction… 16275

significant trend and the magnitude of the trend, respectively, for annual and seasonal time
scale (winter, pre-monsoon, monsoon and post-monsoon). SQMK test is used to detect the
change point. Prior to performing the trend analysis, pre-whitening is done to remove the
statistically significant lag-1 serial correlation present in the data. Further, the interaction
between groundwater level fluctuations and rainfall is modelled using HARTT.

4.1 Rainfall trend analysis

Annual average rainfall over the catchment is estimated using the rainfall data from the
6 IMD grids and is spatially interpolated. Figure 3 is the graphical representation of the
results. Annual average rainfall in catchment is estimated to be 1750 mm during the study
period.
Grid 1 of the catchment receives the highest rainfall, and it decreases towards the east
direction in the catchment. Governed by the local orography in the catchment, grids 1 and
2, located at the upper catchment, receives more rainfall than the lower catchment. The
highest rainfall of 3116 mm is recorded at grid 1. Rainfall recorded at grid 5 and grid 6
is less than the annual average rainfall of the catchment. These stations are located in the
eastern part of the catchment with some smaller hillocks. The rainfall data at different grids
in the Bhadra Catchment are analysed both at seasonal and annual levels. Seasonal analysis
shows that monsoon season contributes the highest rainfall (73%), and the winter season
contributes the least rainfall (1%), as shown in Fig. 4. Pre-monsoon and post-monsoon sea-
son contributes 12% and 14% rainfall, respectively.
Standard normal test statistics (Zc) and Sen’s slope (bsen) are estimated for the seasonal
and annual rainfall series at the 6 grids, and the test statistics are shown in Table 1. Rainfall
data for grid 2, which covers large part of the upper catchment, show a decreasing trend at
95% confidence interval for all the seasons, except during the pre-monsoon period. During
pre-monsoon period, the rainfall data are observed to have a significant increasing trend
for all the grids, except grid 5. At grid 5, the data do not show empirical evidences of
statistically significant trend in any of the seasons under considerations. Further, during the
post-monsoon period, a decreasing trend in rainfall is observed at grid 3.

Fig. 3  Annual average rainfall

distribution map

13
16276 H. S. Nanditha et al.

Fig. 4  Seasonal rainfall contribu-

tion

The spatial interpolation of Sen’s slope (bsen) over the catchment is depicted in Fig. 5.
Only in the upper catchment, annual and monsoon rainfall show a statistically significant
trend. In the lower catchment, no substantial evidences are observed to establish the pres-
ence of any trend at 95% confidence interval at annual and monsoon scale. Further, sea-
sonal analysis during winter, which accounts for only 1% of the annual rainfall, is observed
to have no noticeable trend in rainfall. The pre-monsoon rainfall shows a significant
increasing trend, whereas the post-monsoon rainfall shows a significant decreasing trend in
the upper catchment.

4.2 Trend change‑point analysis for rainfall

The SQMK test is used to determine the change point in case of any significant trend
detected in the data. The test statistics for the prograde and retrograde series are plotted
for the annual rainfall data at all the 6 grids (Fig. 6a–f). The SQMK plots of all the grids,
except grid 2, overlap at multiple points and show no evidences of the significant trend.
At grid 2, the prograde and retrograde series meet in 2010 and deviate from each other, as
shown in Fig. 6b, indicating a possible change point in 2010.

4.3 Groundwater trend analysis

The groundwater data from various well locations in the Bhadra Catchment are examined
separately. The Bhadra catchment has an average annual groundwater level of 4.6 m. The
lowest depth to groundwater level is 1.92 m at Kalasa, which is located at an elevation of
712 m, whilst the highest is 9.01 m at Niduvale, which is located at an elevation of 902 m.
Figure 7 depicts the average yearly groundwater level distribution map.
MK test is performed for all the wells at annual and seasonal time scales. The test sta-
tistics are presented in Table 2, and the values significant at 95% confidence level are high-
lighted. The corresponding Sen’s slope values are also presented in Table 2. From the anal-
ysis, evidences for a statistically significant trend are obtained only for the wells located at
the upper catchment. Figure 8 depicts the spatial interpolation of Sen’s slope (bsen) over the
catchment. Annual depth to groundwater table shows a positive trend at Kalasa, whereas it
is negative at Sunksale. A decreasing trend indicates a rise in groundwater level and vice
versa. Thus, groundwater levels at Kalasa are dropping, whereas it is rising at Sunksale

13
 Table 1  Results of the Mann–Kendall (Zc) test and Sen’s Slope test (bsen) for Rainfall
Grid no Lat/long Annual Winter Pre-monsoon Monsoon Post-monsoon
Zc bsen Zc bsen Zc bsen Zc bsen Zc bsen

Grid 1 75.25E, 13.25N 0.620 15.445 0.075 0 2.356 8.521 0.372 10.205 −0.719 −2.498
Grid 2 75.5E, 13.25N −2.483 −57.638 −0.45 −0.106 2.555 8.088 −2.113 −41.106 −2.853 −7.487
Statistical analysis of rainfall and groundwater interaction…

Grid 3 75.5E, 13.5N −1.004 −13.668 −0.352 0 3.398 8.807 −1.321 −15.473 −2.604 −6.102
Grid 4 75.5E, 13.75N −0.211 −2.469 −0.155 0 2.356 5.464 −0.158 −3.924 −1.463 −4.796
Grid 5 75.75E, 13.25N 0.471 3.891 1.521 0.295 1.712 5.871 −0.124 −0.743 −1.315 −3.121
Grid 6 75.75E, 13.5N −0.273 −2.958 0.228 0 2.059 6.533 0 −0.165 −1.91 −6.249
*
Bold value indicates significant trend at the 95% confidence level
16277

13
16278 H. S. Nanditha et al.

Fig. 5  Maps showing Sen’s Slope (bsen) for rainfall for different seasons

annually. Seasonal analysis of the depth to water level shows a similar pattern. The analy-
sis shows a decreasing trend at Sunksale, but significant only during winter and pre-mon-
soon seasons. During winter and post-monsoon season, an increasing trend is observed in
the upper catchment, indicating drop in the groundwater level. The evidence gained from
the analysis is insufficient to prove the presence of any meaningful trend in the depth to
groundwater level data during monsoon and pre-monsoon season over the entire catch-
ment, except at Sunksale.
From the analysis, a decreasing trend in the rainfall is observed in the upper catchment
for all the seasons, except during the pre-monsoon. Similarly, the depth to groundwater
level indicates an increasing trend annually in the upper catchment. However, the
groundwater level fluctuations are not directly proportional to the variations observed in
the rainfall. Hence, to study the impact of rainfall on the groundwater level fluctuations,
HARTT model is used.

4.4 HARTT model

The HARTT model is used to study the long-term association between rainfall and
groundwater level. From the SQMK test, since a change point is identified at the
upper catchment (grid 2) in the year 2010, further analysis is performed for the period
2010–2018. The multiple regression analysis is performed between groundwater level,
ARR and time. Both AMRR and AARR at different lag time are used to represent the
ARR, and corresponding R2 of the regression relation is determined. For all the observa-
tion wells, since model regression relation using AARR is observed to have higher R2

13
Statistical analysis of rainfall and groundwater interaction… 16279

Fig. 6  Sequential Mann–Kendall plots for a grid1, b grid 2 and c grid 3 d grid 4 e grid 5 f grid 6

Fig. 7  Average annual ground-

water level distribution map

compared to AMRR, AARR is used to represent the ARR in the HARTT model in the
current study. The corresponding regression coefficients, their significance levels and the
R2 are shown in Table 3.

13
 16280

13
Table 2  Results of the Mann–Kendall (Zc) test and Sen’s Slope test (bsen) for groundwater level in the Bhadra catchment
SL. no GW well Annual Winter Pre-monsoon Monsoon Post-monsoon
Zc bsen Zc bsen Zc bsen Zc bsen Zc bsen

1 NR Pura 1.109 0.039 1.902 0.094 1.004 0.057 0.792 0.049 1.743 0.074
2 Sangameshwarpet 1.162 0.018 1.743 0.032 1.004 0.031 −0.050 −0.003 1.268 0.025
3 Kalasa 2.007 0.019 2.582 0.027 0.696 0.008 1.340 0.024 2.218 0.03
4 Niduvale 1.761 0.035 1.067 0.025 1.315 0.031 1.513 0.038 2.235 0.037
5 Sunkasale −2.829 −0.02 −2.166 −0.009 −3.079 −0.022 −1.490 −0.015 −0.796 −0.008
*
Bold value indicates significant trend at the 95% confidence level
H. S. Nanditha et al.
Statistical analysis of rainfall and groundwater interaction… 16281

Fig. 8  Map showing Sen’s Slope (bsen) for groundwater for different seasons

K1 indicates impact of rainfall on the groundwater level fluctuation. p value less than
0.05 indicates the significance of the variable in the regression relation at 95% confi-
dence level. Average value of K1 during the study period for the catchment is found to
be −0.0012 m/mm which implies that 1 mm increase or decrease in the rainfall anomaly
causes 0.0012 m decrease or increase in depth to water table at monthly level. K2 indicates
the time trend present in the water table over and above the variations explained by the
rainfall anomaly. It is found that beyond the variation in the rainfall induced recharge, on
an average, the depth to water table increases at the rate 0.0119 m/month or 0.1428 m/year,
which is insignificant at 95% confidence level. Earlier studies also reported higher p value
for K2 in case of shallow bores (Ferdowsian et al., 2001). Thus, rainfall anomaly may be
considered as the major factor that explains the temporal variation in the groundwater level
in the catchment.
The R2 indicates the degree of fit between calculated and measured groundwater lev-
els. Since ARR is the major variable, good R2 values imply how much percentage of
the groundwater level fluctuation can be directly related to rainfall variations. From the

Table 3  Statistical parameters from the HARTT model for Bhadra catchment
WELL Lag (months) R2 (%) K0 (m) K1 (m/mm) p value K2 (m/month) p value

Narasimharajpur 0 49 4.603 −0.0009 < 0.001 0.0499 < 0.001

Sangameshwarpet 0 40 4.925 −0.0022 < 0.001 0.0122 < 0.001
Niduvale 0 39 10.388 −0.0027 < 0.001 0.0014 0.7300
Sunkasale 0 30 2.255 −0.0006 < 0.001 −0.0030 0.0217
Kalasa 5 23 2.020 0.0003 < 0.001 −0.0012 0.5273
AVERAGE 1 36 4.838 −0.0012 < 0.001 0.0119 0.2559

13
16282 H. S. Nanditha et al.

analysis conducted for a period of 8 years, the average R2 value for the entire catchment is
found to be 36%, indicating that rainfall anomaly is responsible for 36% of the groundwa-
ter level variation observed during the study period. The spatial interpolation of R2 over
the catchment as observed in Fig. 9 shows that, the R2 value is not uniform throughout
the catchment. Whilst lowest value of R2 is obtained for Kalasa (23%), the highest value
is obtained for Narasimharajpura (49%), thereby showing an increasing pattern from the
upper to the lower catchment. Lower R2 values are observed in the upper catchment, at
places where depth to water table is very shallow (less than 2.5 m), and wells which are
either located along the steep slopes or near the groundwater discharge location. The pres-
ence of significant subsurface flow condition (DDMP, 2019), as may be due to the topog-
raphy and the steep hydraulic gradient, may be the additional factors causing the varia-
tions in groundwater level, over and above what can be explained from the rainfall anomaly
observed at that station. Subsurface flow conditions are also affected by the rainfall condi-
tions at the recharge sites, which may be different from the IMD grid under consideration.
The time lag (L1) between rainfall and its impact on groundwater is estimated by using
ARR at various lagged time (ARR t−L1) in the regression equations. The lag is found to
be zero for all the wells except for Kalasa. The regression relation shows 5 months lag for
Kalasa which is located near the discharge point as observed in the DEM in Fig. 2. A close
analysis of the local field condition indicates that due to the presence of high clay content
in the soil, groundwater flow downslope is slow resulting in the delay between the rainfall
at the recharge areas and the groundwater level near the discharge site.

4.5 Observations on the rainfall and groundwater interaction in Bhadra catchment

Pre-monsoon rainfall in the catchment shows an increasing trend. For all other season,
in the upper catchment, rainfall is decreasing, thereby showing a decreasing trend at the
annual level. However, at annual level, no significant trend has been observed in the lower
catchment. Groundwater trend analysis also shows that depth to water table is decreasing
in the upper catchment, except for Sunksale. The interaction between groundwater and
rainfall, analysed using HARTT model, shows that the groundwater variation is explained
partially by the rainfall induced recharge. From the analysis, no evidence of any temporal

Fig. 9  Map showing R2 values

for groundwater for various well
locations

13
Statistical analysis of rainfall and groundwater interaction… 16283

trend is observed, beyond that can be explained by the rainfall anomaly. However, the
overall R2 was found to be only 0.36, which indicates that only 36% of the groundwater
level variations are explained by the rainfall anomaly. In order to understand the poor
relation between the rainfall and the groundwater level variations, topography of the area,
rainfall characteristics, soil and aquifer characteristics have been carefully studied.
Lower R2 estimated for the catchment may be due to soil and aquifer type, recharge and
discharge conditions. The tropical catchment receives very high intensity spells of rainfall.
When occurs over the steep sloping terrains, much of this water contribute towards surface
runoff (Meenu et al., 2013; DDMP, 2019). The water that infiltrates into the soil and finally
percolate into the groundwater zone is also under the influence of steep hydraulic gradient
in the hilly terrain. These conditions favour significant subsurface flow in the area, finally
contributing to the rivers and streams. The same has been reported in the earlier studies
from the catchment (DDMP, 2019). These subsurface flows have impact on the aquifers
in the study area. Groundwater levels in wells located at higher elevations (Niduvale) are
declining, which may be attributed to the loss of percolated water to the subsurface flow.
Nevertheless, in the lower catchment (Narasimharajapura), the slope decreases and the
subsurface flow conditions become less prevalent, resulting in a better R2 between rainfall
and groundwater. Goodarzi (2020) has reported higher R ­ 2 for areas located at the recharge
site and for deep aquifers with zero lag time, whereas lower R2 for shallow aquifers near
the discharge sites. The same is observed in this study at Kalasa, where even though
rainfall is increasing, the groundwater level is decreasing. Only 23% of the groundwater
level variations are explained by the rainfall. Similarly, only 30% of the groundwater level
fluctuations in Sunksale is explained by the rainfall anomaly.
Thus, it is concluded that coefficient of determination, and hence the variation of
groundwater level with rainfall, is poor in areas where the water table is very shallow
and is located near the streams. This may be because of the presence of subsurface flow
conditions generated by the steep slope and hydraulic gradient. Similar results have been
observed in some of the earlier studies as well, where the HARTT model was reported to be
less effective in cases when the water table is shallow, and in cases, where the other factors
such as topography, hydraulic gradient, well defined flow pathways, evapotranspiration,
etc. are also affecting the water table dynamics (Ferdowsian et al., 2001; Goodarzi, 2020;
Yihdego et al., 2017). Since the HARTT model is attempting to establish a relationship
between groundwater and precipitation alone, the influence of other factors, as mentioned
above, cannot be effectively captured in the model. Any model that can capture the effect of
topography and larger spatial patterns of rainfall on groundwater levels can only simulate
groundwater dynamics in such catchments. As a result, numerical models would be a better
alternative to the statistical models.

5 Conclusions

In the present study, data from a tropical hilly catchment are analysed to find the
relationship between rainfall anomaly and groundwater level fluctuations using the
HARTT model. Rainfall analysis over the Bhadra catchment reveals a statistically
significant increasing trend in pre-monsoon rainfall. In a substantial part of the upper
catchment, rainfall is observed to exhibit a decreasing tendency at the 95% confidence
level. A corresponding increase in the depth to water table is observed generally in the
upper catchment. However, the relation between rainfall and groundwater level fluctuations

13
16284 H. S. Nanditha et al.

are not uniform throughout the catchment. Though rainfall is the major contributor for the
groundwater level fluctuations, it is observed to have a partial influence on the groundwater
level and percentage dependency increases towards the lower catchment. The direct effect
of rainfall on the groundwater level is observed to be less in case of shallow aquifers located
near the discharge site. Further, in this study, the HARTT model uses a regression relation
between the groundwater level observation and rainfall at that well location, and this is
true for wells located near the recharge sites. Wells which are at the discharge site may be
influenced by the rainfall at the recharge areas as well. Similar conditions can be observed
on a large number of tropical hilly catchments. Incorporation of the additional information
about the topography, subsurface flow conditions and rainfall at the recharge areas in the
statistical model may help to improve the results further. The study thus shows that though
the HARTT model performs better in moderate catchments, they are less effective in cases
of shallow aquifers and in areas where the steep topography causes significant subsurface
flow. Numerical models would be the better alternative to incorporate these specific
topographical and hydrogeological conditions into the analysis in such areas.
Acknowledgements Authors are thankful to ISRO for providing Cartosat-1-based Digital Elevation Model
used in study. We also acknowledge Indian Meteorological Department (IMD) for providing rainfall datasets
and Central Groundwater Board (CGWB) for providing groundwater data used in the study.

Data availability All the data used are publicly available and can be shared upon request. Whereas
groundwater data are collected from the Central groundwater board and so are not publicly available.
Groundwater data that support the findings of this study are however available from the authors upon
request and with permission of central groundwater board.

Declarations
Conflict of interest The authors declare no conflict of interest.

References
Almedeij, J., & Al-Ruwaih, F. (2006). Periodic behavior of groundwater level fluctuations in residential
areas. Journal of Hydrology, 328(3–4), 677–684. https://​doi.​org/​10.​1016/j.​jhydr​ol.​2006.​01.​013
Anamika Shalini, T. (2012). Groundwater level and rainfall variability trend analysis using GIS in parts of
Jharkhand state (India) for Sustainable Management of Water Resources. Int. Res. J. Environment Sci.
International Science Congress Association, 1(4), 24–31. www.​isca.​in
Aslam, L. B. M. A. M. (2018). Long - term trend analysis of water - level response to rainfall of Gulbarga
watershed, Karnataka, India, in basaltic terrain : Hydrogeological environmental appraisal in arid
region. Applied Water Science, 8(4), 1–9. https://​doi.​org/​10.​1007/​s13201-​018-​0723-9
Cai, Z., Wang, W., Zhao, M., Ma, Z., Lu, C., & Li, Y. (2020). Interaction between surface water and ground-
water in yinchuan plain. Water (switzerland), 12(9), 1–23. https://​doi.​org/​10.​3390/​w1209​2635
Chanapathi, T., Thatikonda, S., & Raghavan, S. (2018). Analysis of rainfall extremes and water yield of
Krishna river basin under future climate scenarios. Journal of Hydrology: Regional Studies, 19(June),
287–306. https://​doi.​org/​10.​1016/j.​ejrh.​2018.​10.​004
Crosbie, R. S., Binning, P., & Kalma, J. D. (2005). A time series approach to inferring groundwater recharge
using the water table fluctuation method. Water Resources Research, 41(1), 1–9. https://​doi.​org/​10.​
1029/​2004W​R0030​77
Cuthbert, M. O., Taylor, R. G., Favreau, G., Todd, M. C., Shamsudduha, M., Villholth, K. G., MacDonald,
A. M., Scanlon, B. R., Kotchoni, D. O. V., Vouillamoz, J. M., Lawson, F. M. A., Adjomayi, P. A.,
Kashaigili, J., Seddon, D., Sorensen, J. P. R., Ebrahim, G. Y., Owor, M., Nyenje, P. M., Nazoumou, Y.,
& Kukuric, N. (2019). Observed controls on resilience of groundwater to climate variability in sub-
Saharan Africa. Nature, 572(7768), 230–234. https://​doi.​org/​10.​1038/​s41586-​019-​1441-7

13
Statistical analysis of rainfall and groundwater interaction… 16285

District Disaster Management Plan (DDMP) for Chikkamagaluru District (2019), Preparerd by: District dis-
aster management authority Chikkamagaluru District, Karnataka.
Dong, L., Chen, J., Fu, C., & Jiang, H. (2012). Analysis of groundwater-level fluctuation in a coastal con-
fined aquifer induced by sea-level variation. Hydrogeology Journal, 20(4), 719–726. https://​doi.​org/​10.​
1007/​s10040-​012-​0838-2
Dragoni, W., & Sukhija, B. S. (2008). Climate change and groundwater : A short review. Geological Society
Special Publication, 288, 1–12. https://​doi.​org/​10.​1144/​SP288.1
Ferdowsian, R., Pannell, D. J., McCarron, C., Ryder, A., & Crossing, L. (2001). Explaining groundwater
hydrographs: Separating atypical rainfall events from time trends. Australian Journal of Soil Research,
39(4), 861–875. https://​doi.​org/​10.​1071/​SR000​37
Ferdowsian, R., Ryder, A., George, R., Bee, G., & Smart, R. (2002). Groundwater level reductions under
lucerne depend on the landform and groundwater flow systems (local or intermediate). Australian
Journal of Soil Research, 40(3), 381–396. https://​doi.​org/​10.​1071/​SR010​14
Ferdowsian, R., & Pannell, D. J. (2009). Explaining long-term trends in groundwater hydrographs. In 18th
World IMACS Congress and MODSIM 2009 - International Congress on Modelling and Simulation:
Interfacing Modelling and Simulation with Mathematical and Computational Sciences, Proceedings,
July, pp. 3060–3066.
Fu, G., Crosbie, R. S., Barron, O., Charles, S. P., Dawes, W., Shi, X., Van Niel, T., & Li, C. (2019). Attribut-
ing variations of temporal and spatial groundwater recharge: A statistical analysis of climatic and non-
climatic factors. Journal of Hydrology, 568, 816–834. https://​doi.​org/​10.​1016/j.​jhydr​ol.​2018.​11.​022
Gibrilla, A., Anornu, G., & Adomako, D. (2018). Trend analysis and ARIMA modelling of recent ground-
water levels in the White Volta River basin of Ghana. Groundwater for Sustainable Development, 6,
150–163. https://​doi.​org/​10.​1016/j.​gsd.​2017.​12.​006
Goodarzi, M. (2020). Application and performance evaluation of time series, neural networks and HARTT
models in predicting groundwater level changes, Najafabad Plain. Iran. Sustainable Water Resources
Management, 6(4), 1–10. https://​doi.​org/​10.​1007/​s40899-​020-​00427-2
Green, T. R., Taniguchi, M., Kooi, H., Gurdak, J. J., Allen, D. M., Hiscock, K. M., Treidel, H., & Aureli, A.
(2011). Beneath the surface of global change: Impacts of climate change on groundwater. Journal of
Hydrology, 405(3–4), 532–560. https://​doi.​org/​10.​1016/j.​jhydr​ol.​2011.​05.​002
Hamed, K. H. (2009). Enhancing the effectiveness of prewhitening in trend analysis of hydrologic data.
Journal of Hydrology, 368(1–4), 143–155. https://​doi.​org/​10.​1016/j.​jhydr​ol.​2009.​01.​040
Hanson, R. T., Newhouse, M. W., & Dettinger, M. D. (2004). A methodology to asess relations between
climatic variability and variations in hydrologic time series in the southwestern United States. Journal
of Hydrology, 287(1–4), 252–269. https://​doi.​org/​10.​1016/j.​jhydr​ol.​2003.​10.​006
Jain, S. K., & Kumar, V. (2012). Trend analysis of rainfall and temperature data for India. Current Science,
102(1), 37–49.
Jan, C. D., Chen, T. H., & Lo, W. C. (2007). Effect of rainfall intensity and distribution on groundwater level
fluctuations. Journal of Hydrology, 332(3–4), 348–360. https://​doi.​org/​10.​1016/j.​jhydr​ol.​2006.​07.​010
Jayasree, V. (2008). Spatial distribution of daily precipitation in the Bhadra River, Karnataka. Journal of
Applied Meteorology, 21, 1–10.
Kulkarni, A., & Storch, H. V. (1995a). Monte Carlo experiments on the effect of serial correlation on the
Mann-Kendall test of trend. Meteorologische Zeitschrift, 4(JANUARY), 82–85.
Kulkarni, A., & von Storch, H. (1995b). Monte Carlo experiments on the effect of serial correlation on the
Mann-Kendall test of trend. Meteorologische Zeitschrift, 4(2), 82–85. https://​doi.​org/​10.​1127/​metz/4/​
1992/​82
Lasagna, M., Mancini, S., Antonio De Luca, D., & Cravero, M. (2019). Piezometric levels in the Piedmont
plain (NW Italy): Trend and hydrodynamic behaviour of the shallow aquifer. Rendiconti Online Soci-
eta Geologica Italiana, 48(July), 2–9. https://​doi.​org/​10.​3301/​ROL.​2019.​30
Leong, C., & Yokoo, Y. (2019). An interpretation of the relationship between dominant rainfall-runoff pro-
cesses and the shape of flow duration curve by using data-based modeling approach. Hydrological
Research Letters, 13(4), 62–68. https://​doi.​org/​10.​3178/​hrl.​13.​62
Lutz, A., Minyila, S., Saga, B., Diarra, S., Apambire, B., & Thomas, J. (2015). Fluctuation of groundwater
levels and recharge patterns in Northern Ghana. Climate, 3(1), 1–15. https://​doi.​org/​10.​3390/​cli30​10001
Mack, T. J., Chornack, M. P., & Taher, M. R. (2013). Groundwater-level trends and implications for sustain-
able water use in the Kabul Basin. Afghanistan. Environment Systems and Decisions, 33(3), 457–467.
https://​doi.​org/​10.​1007/​s10669-​013-​9455-4
Mann, H. B. (1945). Non-parametric test against trend. Econometrica, 13(3), 245–259. http://​www.​jstor.​org/​
stable/​19071​87.

13
16286 H. S. Nanditha et al.

Marofi, S., Soleymani, S., Salarijazi, M., & Marofi, H. (2012). Watershed-wide trend analysis of tem-
perature characteristics in Karun-Dez watershed, southwestern Iran. Theoretical and Applied Cli-
matology, 110(1–2), 311–320. https://​doi.​org/​10.​1007/​s00704-​012-​0662-6
Meals, D. W., Dressing, S. A., & Davenport, T. E. (2010). Lag time in water quality response to best
management practices: A review. Journal of Environmental Quality, 39(1), 85–96. https://​doi.​org/​
10.​2134/​jeq20​09.​0108
Meenu, R., Rehana, S., & Mujumdar, P. P. (2013). Assessment of hydrologic impacts of climate change
in Tunga-Bhadra river basin, India with HEC-HMS and SDSM. Hydrological Processes, 27(11),
1572–1589. https://​doi.​org/​10.​1002/​hyp.​9220
Ngongondo, C. S. (2006). An analysis of long-term rainfall variability, trends and groundwater availabil-
ity in the Mulunguzi river catchment area, Zomba mountain. Southern Malawi. Quaternary Inter-
national, 148(1), 45–50. https://​doi.​org/​10.​1016/j.​quaint.​2005.​11.​006
Overpeck, J., & Udall, B. (2010). Dry times ahead. Science, 328(5986), 1642–1643. https://​doi.​org/​10.​
1126/​scien​ce.​11865​91
Panwar, S., & Chakrapani, G. J. (2013). Climate change and its influence on groundwater resources,
International Journal of Engineering and Science ISSN: 2278–4721, Vol. 1, Issue 5 (October
2012), pp. 43–60 www.​resea​rchin​venty.​com
Patakamuri, S. K., Muthiah, K., & Sridhar, V. (2020). Long-Term homogeneity, trend, and change-point
analysis of rainfall in the arid district of ananthapuramu, Andhra Pradesh State, India. Water (Swit-
zerland), 12(1). https://​doi.​org/​10.​3390/​w1201​0211
Pathak, A. A., & Dodamani, B. M. (2019). Trend analysis of groundwater levels and assessment of
regional groundwater drought: Ghataprabha river basin. India. Natural Resources Research, 28(3),
631–643. https://​doi.​org/​10.​1007/​s11053-​018-​9417-0
Patil, N. S., & Chetan, N. L. (2017). Groundwater modeling of Hiranyakeshi watershed of Ghataprabha
sub-basin. Journal of the Geological Society of India, 90(3), 357–361. https://​doi.​org/​10.​1007/​
s12594-​017-​0724-6
Petpongpan, C., Ekkawatpanit, C., & Kositgittiwong, D. (2020). Climate change impact on surface water
and groundwater recharge in northern Thailand. Water (Switzerland), 12(4). https://​doi.​org/​10.​
3390/​W1204​1029
Reghunath, R., Murthy, T. R. S., & Raghavan, B. R. (2005). Time series analysis to monitor and assess
water resources: A moving average approach. Environmental Monitoring and Assessment, 109(1–
3), 65–72. https://​doi.​org/​10.​1007/​s10661-​005-​5838-4
Reshmidevi, T. V., Nagesh Kumar, D., Mehrotra, R., & Sharma, A. (2018). Estimation of the climate
change impact on a catchment water balance using an ensemble of GCMs. Journal of Hydrology,
556, 1192–1204. https://​doi.​org/​10.​1016/j.​jhydr​ol.​2017.​02.​016
Russo, T. A., & Lall, U. (2017). Depletion and response of deep groundwater to climate-induced pump-
ing variability. Nature Geoscience, 10(2), 105–108. https://​doi.​org/​10.​1038/​ngeo2​883
Salehi, S., Dehghani, M., Mortazavi, S. M., & Singh, V. P. (2020). Trend analysis and change point
detection of seasonal and annual precipitation in Iran. International Journal of Climatology, 40(1),
308–323. https://​doi.​org/​10.​1002/​joc.​6211
Sayemuzzaman, M., & Jha, M. K. (2014). Seasonal and annual precipitation time series trend analysis
in North Carolina, United States. Atmospheric Research, 137, 183–194. https://​doi.​org/​10.​1016/j.​
atmos​res.​2013.​10.​012
Seager, R., Ting, M., Li, C., Naik, N., Cook, B., Nakamura, J., & Liu, H. (2013). Projections of declin-
ing surface-water availability for the southwestern United States. Nature Climate Change, 3(5),
482–486. https://​doi.​org/​10.​1038/​nclim​ate17​87
Shamsudduha, M., Chandler, R. E., Taylor, R. G., & Ahmed, K. M. (2009). Recent trends in groundwater
levels in a highly seasonal hydrological system: The Ganges-Brahmaputra-Meghna Delta. Hydrol-
ogy and Earth System Sciences, 13(12), 2373–2385. https://​doi.​org/​10.​5194/​hess-​13-​2373-​2009
Sonali, P., & Nagesh Kumar, D. (2013). Review of trend detection methods and their application to
detect temperature changes in India. Journal of Hydrology, 476, 212–227. https://​doi.​org/​10.​1016/j.​
jhydr​ol.​2012.​10.​034
Storch, H. (1995). Misuses of Statistical Analysis in Climate Research. H. von Storch et al. (eds.), Analysis
of Climate Variability. Springer, Berlin, Heidelberg.
Taminskas, J., Linkevičienė, R., Šimanauskienė, R., Jukna, L., Kibirkštis, G., & Tamkevičiūtė, M. (2018).
Climate change and water table fluctuation: Implications for raised bog surface variability. Geomor-
phology, 304, 40–49. https://​doi.​org/​10.​1016/j.​geomo​rph.​2017.​12.​026
Tian, D., & Liu, D. (2011). A new integrated surface and subsurface flows model and its verification.
Applied Mathematical Modelling, 35(7), 3574–3586. https://​doi.​org/​10.​1016/j.​apm.​2011.​01.​035

13
Statistical analysis of rainfall and groundwater interaction… 16287

Vano, J. A., Nijssen, B., & Lettenmaier, D. P. (2015). Seasonal hydrologic responses to climate change in
the Pacific Northwest. Water Resources Research, 51(4), 1959–1976. https://​doi.​org/​10.​1002/​2014W​
R0159​09
Vousoughi, D. F., Dinpashoh, Y., Aalami, M. T., & Jhajharia, D. (2013). Trend analysis of groundwater
using non-parametric methods (case study: Ardabil plain). Stochastic Environmental Research and
Risk Assessment, 27(2), 547–559. https://​doi.​org/​10.​1007/​s00477-​012-​0599-4
Woldeamlak, S. T., Batelaan, O., & De Smedt, F. (2007). Effects of climate change on the groundwater sys-
tem in the Grote-Nete catchment. Belgium. Hydrogeology Journal, 15(5), 891–901. https://​doi.​org/​10.​
1007/​s10040-​006-​0145-x
Yihdego, Y., Danis, C., & Paffard, A. (2017). Why is the groundwater level rising? A case study using hartt
to simulate groundwater level dynamic. Water Environment Research, 89(12), 2142–2152. https://​doi.​
org/​10.​2175/​10614​3017x​14839​99452​3785
Yue, S., Pilon, P., Phinney, B., & Cavadias, G. (2002). The influence of autocorrelation on the ability to
detect trend in hydrological series. Hydrological Processes, 16(9), 1807–1829. https://​doi.​org/​10.​1002/​
hyp.​1095
Zeru, G., Alamirew, T., Shishaye, H. A., Olmana, M., Tadesse, N., & Reading, M. J. (2020). Groundwater
level trend analysis using the statistical auto-regressive hartt method. Hydrological Research Letters,
14(1), 17–22. https://​doi.​org/​10.​3178/​hrl.​14.​17
Zhang, W., Yan, Y., Zheng, J., Li, L., Dong, X., & Cai, H. (2009). Temporal and spatial variability of annual
extreme water level in the Pearl River Delta region. China. Global and Planetary Change, 69(1–2),
35–47. https://​doi.​org/​10.​1016/j.​glopl​acha.​2009.​07.​003
Zhang, Y., He, B. I. N., Guo, L., & Liu, D. (2019). Differences in response of terrestrial water storage com-
ponents to precipitation over 168 global river basins. Journal of Hydrometeorology, 20(9), 1981–1999.
https://​doi.​org/​10.​1175/​JHM-D-​18-​0253.1

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under
a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted
manuscript version of this article is solely governed by the terms of such publishing agreement and applicable
law.

Authors and Affiliations

H. S. Nanditha1,2 · T. V. Reshmidevi1,2 · L. Udaya Simha1,2 · Parthan Kunhikrishnan1,2

T. V. Reshmidevi
[email protected]
L. Udaya Simha
[email protected]
Parthan Kunhikrishnan
[email protected]
1
Department of Civil Engineering, B.M.S. College of Engineering, Bengaluru 560019, India
2
Visvesvaraya Technological University, Belagavi, Karnataka, India

13