Urban Water Journal

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/nurw20

A model-based estimate of the groundwater

budget and associated uncertainties in Bengaluru,
India

S. K. Tomer, M. Sekhar, K. Balakrishnan, Deepak Malghan, S. Thiyaku, M.

Gautam & Vishal K. Mehta

To cite this article: S. K. Tomer, M. Sekhar, K. Balakrishnan, Deepak Malghan, S. Thiyaku,

M. Gautam & Vishal K. Mehta (2021) A model-based estimate of the groundwater budget
and associated uncertainties in Bengaluru, India, Urban Water Journal, 18:1, 1-11, DOI:
10.1080/1573062X.2020.1836237

To link to this article: https://doi.org/10.1080/1573062X.2020.1836237

© 2020 The Author(s). Published by Informa View supplementary material

UK Limited, trading as Taylor & Francis
Group.

Published online: 16 Dec 2020. Submit your article to this journal

Article views: 1630 View related articles

View Crossmark data Citing articles: 4 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=nurw20
URBAN WATER JOURNAL
2021, VOL. 18, NO. 1, 1–11
https://doi.org/10.1080/1573062X.2020.1836237

RESEARCH ARTICLE

A model-based estimate of the groundwater budget and associated uncertainties in

Bengaluru, India
S. K. Tomera, M. Sekharb, K. Balakrishnan c
, Deepak Malghand, S. Thiyaku a
, M. Gautame and Vishal K. Mehtaf
a
Satyukt Analytics Pvt Ltd., Bangalore, India; bDepartment of Civil Engineering, Indian Institute of Science, Bangalore, India; cIndian Institute for
Human Settlements, Bangalore, India; dIndian Institute of Management, Bangalore, India; eArcadis Consulting India Pvt. Ltd., Bangalore, India;
f
Stockholm Environment Institute -US, Davis, CA, USA

ABSTRACT ARTICLE HISTORY

Understanding the water metabolism of developing country cities is challenging because of insufficient Received 25 April 2019
knowledge of how social, infrastructural and hydrological dimensions are coupled. Using Bengaluru City in Accepted 5 October 2020
India, we demonstrate how fine-resolution data can inform model-based estimates of urban groundwater KEYWORDS
budgets. Groundwater levels were measured at 154 locations in 2016 and used to estimate model parameters Groundwater; urban;
and uncertainty. Total water use was estimated at 1470 million litres per day (MLD). Groundwater pumping anthropogenic; modelling;
meets the majority of this use (827 MLD), followed by utility water supply (643 MLD). Total recharge was Bengaluru; metabolism
estimated at 973 MLD. Natural recharge is a much smaller portion (183 MLD) compared to anthropogenic
recharge from leaking water supply and wastewater systems (791 MLD). The city experienced a net negative
groundwater balance (40 MLD). Natural recharge and total water use estimates showed lower uncertainty.
Spatial variation in these fluxes is described and related to secondary information.

Introduction endangering this underground infrastructure (Hayashi et al. 2009).

Cities are tightly coupled socio-ecological systems (Schandl and Similar examples can be found in several other cities around the
Capon 2012) sustained by metabolic flows of various resources world including Milan, Italy (Gattinoni and Scesi 2017) and in the U.K.
(Wolman 1965; Kennedy, Cuddihy, and Engel-Yan 2007). Water is (Hurst and Wilkinson 1986; Brassington and Rushton 1987; Brassington
the largest metabolic flow and urbanization affects the water cycle 1990).
in both quantity and quality (Vázquez-Suñé, Sánchez-Vila, and Carrera Lerner (2002) and Vázquez-Suñé, Sánchez-Vila, and Carrera
2005; Decker et al. 2000). Several aspects of urban hydrology distin­ (2005), in reviewing the particular challenges of understanding
guish it from that of natural systems. Large amounts of water are urban groundwater budgets, included the following: (i) the90
imported, distributed and collected again in sewers and septic tanks. complexity of sources and pathways of recharge; (ii) the need
Leaking water supply lines, sewer mains and septic tanks comprise for large amounts of data, from various sources and (iii) prevalence
anthropogenic elements of groundwater recharge, that add to nat­ of large uncertainties and data gaps. Additionally, estimating
ural recharge from precipitation. Anthropogenic recharge was found groundwater budgets is especially difficult in cities of the devel­
to constitute 20% to 50% of total recharge in some cases (Lerner oping world given the many water sources and modes of supply
2002). On the negative side of the groundwater budget, pumping is that people rely on and the wide variation in supply across
the anthropogenic element which can dominate the urban ground­ different parts of these cities (Foster, Lawrence, and Morris 1998;
water balance. Jaglin 2014). India’s urban water supply scenario is illustrative of
Examples from some major cities illustrate the anthropogenic this. On the demand side, since no city receives 24 × 7 water
effects on water balances. In Seoul, Korea, groundwater pumping supply, consumers obtain water from multiple sources, including
and leaking water mains were dominant components of the city’s utility piped supply, tankers, private groundwater wells, bottled
groundwater budget (Kim, Lee, and Sung 2001). While the effect on water and untreated water bodies (Srinivasan, Gorelick, and
groundwater level was not substantial because leakages made up for Goulder 2010; Misra and Goldar 2008). As a result, ‘no city munici­
extraction, water quality was impaired because of recharge from pality knows what the real water demand is in the spaces they
leaking sewers and septic tanks. The dynamic nature of socio- govern’ (Narain and Pandey 2012).
hydrological coupling means that policies surrounding land and Although almost all of the above mentioned alternate sources
water management can have long-lasting and surprising conse­ rely on groundwater, the existing groundwater monitoring net­
quences. In Tokyo, in response to groundwater over-extraction in work is too coarse to understand the groundwater status of
the 1950s and 1960s, groundwater pumping was replaced with urban centres. Groundwater resources assessments in the country
imported surface water supply. Groundwater levels recovered over are typically conducted at river basin or state-wide scale, relying on
ensuing decades; however, during this time, a massive underground a combination of GIS, remote sensing, and a national groundwater
rail infrastructure was put in place. Groundwater levels are now monitoring framework that samples at a coarse density. For

CONTACT Vishal K. Mehta vishal.mehta@sei.org

Supplemental data for this article can be accessed here.
© 2020 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group.
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License (http://creativecommons.org/licenses/by-nc-nd/4.0/),
which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited, and is not altered, transformed, or built upon in any way.
2 S. K. TOMER ET AL.

instance, in Karnataka state, the monitoring well density is 1 per

100 km2 and measurements are taken 4 times in a year during
January, May, August and November (Central Groundwater Board
2016). Urban landscapes usually cover areas in the range of tens to
few hundreds of km2 and are highly heterogeneous. Therefore,
these require a much denser monitoring network than what cur­
rently exists. Both the demand and supply sides of the (ground)
water budget are ill-informed in India.
Consequently, the socio-hydrology of Indian cities remains
poorly understood, especially from a quantitative and spatially
disaggregated standpoint (Mehta et al. 2013). This poses
a serious challenge given India’s rapid urbanization (Planning
Commission 2013), lack of adequate water supply, wastewater
treatment and sanitation (Narain and Pandey 2012), and the
sheer fact of it being the largest user of freshwater and ground­
water in the world (Gilbert 2012). As of 2015, there were only two
quantitative socio-hydrological assessments: hydroeconomic
modelling of Chennai and its water supply system (Srinivasan,
Gorelick, and Goulder 2010) and the groundwater study of the
small town of Mulbagal (Sekhar et al. 2013).
In order to address these gaps, we began monitoring ground­
water levels in 154 wells across Bengaluru city at monthly fre­
quency from December 2015, with the overall objectives of
understanding the groundwater budget and spatio-temporal
behaviour with the help of data and models. A first set of results
was produced in 2017 (Sekhar et al. 2017), describing the spatio-
temporal behaviour of groundwater levels measured in 2016.
A simple, vertical, Water Table Fluctuation (WTF) method was
used to estimate groundwater storage change, natural (rainfall)
recharge and net groundwater outflow. In this paper, we use
a more refined and elaborate model to disaggregate the natural
and anthropogenic components of net outflow, while also incor­
porating other infrastructural and socio-economic data. The Figure 1. DEM and monthly rainfall during the study period (January – December,
2016). (a) Digital elevation model (DEM); (b) Monthly rainfall.
Generalized Likelihood Uncertainty Estimation (GLUE) approach
(Beven and Binley 1992) was used to compute uncertainties in
various components of the groundwater balance. Our objective
was to develop and spatially disaggregate the urban groundwater fractures. The city is underlain by Precambrian granite and gneiss,
budget for Bengaluru, with a specific interest in estimating draft weathered to a maximum depth of 60 m, covered by red loamy
(pumping) and recharge. and gravelly soils. The aquifer system in the larger Bengaluru
urban district is a combination of this shallow weathered zone
and the underlying hard rock system (CGWB 2011; Sekhar and
Materials and methods Kumar 2009). Hard-rock aquifers of this kind are characterized by
low hydraulic conductivity (10–65 m2/d) and specific yield (0.005–­
Study area
0.01) and high lateral heterogeneity. Bore-well yields are typically
Climate and physiography very low, from 0.8 to 5 litre/s. A typical weathering profile is
The Greater Bangalore Metropolitan Corporation called Bruhat presented in Maréchal, Dewandel, and Subrahmanyam (2004).
Bengaluru Mahanagara Palike (BBMP) has an area of approxi­ The saprolite or regolith is a clay-rich material derived from pro­
mately 741 km2 and lies between 12°48ʹ-13°9ʹN latitude and 77° longed in situ decomposition of bedrock and is few tens of meters
27ʹ-77°47ʹ longitude. The city core is above mean sea level, on thick. The saprolite can reach a quite high porosity, depending on
a divide with a roughly North-South axis, with the Arkavathi river lithology of parent rock, and generally has low conductivity.
drainage westward, and the Ponnaiyar drainage to the East Recharge to the system will be through the shallow weathered
(Figure 1). The climate is classified as ’topical savannah (Aw)’ in zone, and from there some water will drain into the deeper frac­
the Köppen–Geiger classification system (Beck et al. 2018), with tured rock aquifer. Given this difference in character, they may
a long-term mean annual rainfall of 820 mm (Sekhar et al. 2017). appear to be independent aquifers, but in the longer term and
Monthly rainfall during the study period (January 2016– on a broad scale they will be interconnected. The saprolite, when
December 2016) is shown in Figure 1. 764 mm rainfall occurred saturated, provides the capacitive role while the fissured layer
in 2016, of which 75% was received from May–August. The provides the transmissivity role of the global composite aquifer.
Arkavati river basin has NNE-SSW and ENE-WSW lineaments, Several studies that are at a non-local, larger scale, have used
while the Ponnaiyar drainage is along the NW-SE and WSW-ENE simplifying assumptions regarding transmissivity and specific
 URBAN WATER JOURNAL 3

yield that can be interpreted as effective parameters of this global fastest population growth occurring in outer areas, a negative
composite aquifer (Sekhar et al. 2017, 2013; Sekhar and Kumar correlation between population growth and per capita water
2009). More details on the hydrogeological conceptual model are supplied has been observed, testifying to the challenges in
provided in supplementary material S2. keeping up with the city’s rapid growth (Mehta et al. 2013).
Moreover, less than half the water that reaches the city is
Population growth and status of water infrastructure metered and billed for at the consumption end. In 2015,
Bengaluru, with its origins in the 12th century, has undergone while actual water reaching city borders was 1354 million
a rapid transformation in recent decades that is continuing litres per day (MLD) on average, only 643 MLD (47.49%) was
today. Figure 2 illustrates some of the key characteristics of metered and billed. Bulk of the remaining water is lost to
the city. Bengaluru is the capital of Karnataka state and is one of physical leakages. BWSSB estimates this leakage loss in
the fastest growing urban agglomeration for 2001–2010 September 2013 to be 37.7% of the total water reaching the
(Census 2011). It grew from 1.65 million people in 1971 to city. Piped water supply availability remains highly variable
8.5 million people in 2011, during which time its built-up area across the city. Data on population and public water supply
increased from 20% to approximately 70% of its corresponding from 2011 to 2013 shows that a large population receives less
municipal boundaries (Mehta et al. 2013). Population growth than the World Health Organization (WHO) minimum stan­
between 2001 and 2011 has been especially rapid with some dard (WHO 1993) of 70 litres per capita per day (lpcd) from
city wards growing 150%-300%. The public water supply from the public water supply system. Consumers rely on a variety of
the utility – called the Bengaluru Water Supply and Sewerage other sources, like bottled water, tankers, private bore-wells,
Board (BWSSB) – is largely dependent on imported water from and open water sources.
the Cauvery river which is approximately 100 km away, against On the wastewater side, BWSSB had 778 MLD of total
a total head of about 500 m. installed capacity for wastewater treatment. Sewage treatment
While BWSSB supply has also grown in several phases, it plants with a cumulative capacity of 520 MLD are at various
has not been able to keep up (Mehta et al. 2014). With the stages of project approvals and construction and a further 175

Figure 2. Spatial distribution of relevant variables in Bengaluru. (a) Land Use-Land Cover (LULC) is based on Balakrishnan (2016), (b) population, (c) percentage of
households using borewells and (d) relative density of water supply pipeline. (a) LULC; (b) Population; (c) Percent of people using borewell; (d) Relative density of water
supply pipeline.
4 S. K. TOMER ET AL.

MLD are at a proposal stage (BWSSB 2017b). Although at least that while groundwater depletion occurs during drought years,
1400 MLD of sewage is generated in the city, on average only extreme rainfall events have high potential to recharge
about 520 MLD of this is treated (BWSSB 2017a). This means groundwater.
only 67% of the installed wastewater treatment capacity of 778
MLD is utilized on a regular basis. The reason for this is the lack
of sewerage network in parts of the city, and the age of the Groundwater measurements: current study
existing sewerage network. More than 75% of the main sewer
In order to understand groundwater levels and dynamics at
lines are over 30 years old and in many places these sewer lines
finer scale than afforded by existing monitoring, a grid-
are damaged or blocked, resulting in 50% of the sewage gen­
sampling framework was implemented, with the city divided
erated flowing into storm water drains (BWSSB 2012). Leaking
into 28, 5 km x 5 km grids. Details are provided in Sekhar et al.
sewer lines could therefore contribute a significant portion of
(2017); a summary is presented here. In November 2015, the
the anthropogenic groundwater recharge in the city. Since the
research team visited each grid, and with the help of local
sewerage network coverage exhibits a spatial pattern similar to
residents and shopkeepers, identified between five and seven
that of the piped water supply network, some of the central
existing, unused wells as potential candidates for monthly
areas and most of the peripheral areas of the city have to make
measurement of piezometric levels. Well locations were noted
their own arrangements for sewage treatment and disposal.
using GPS and pictures were taken of each well. A total of 154
While larger apartment buildings and commercial establish­
wells were finally selected based on access and use considera­
ments are required to set up their own sewage treatment
tions. Wells chosen had to be accessible at any time, and they
plants (KSPCB 2013). Individual households and smaller resi­
had to be unused in order to measure static levels. Most of the
dential and commercial buildings may rely on septic tanks,
selected wells are abandoned municipal bore-wells, which tend
which represent an additional source of anthropogenic
to be close to roads and are not pumped. Figure 1 shows the
groundwater recharge.
sampling grid and sample locations.
From January 2016, piezometric levels of each monitoring
Groundwater
well were recorded using Skinny Dipper static levels instrument
Most of the non-utility options for water depend on ground­
by Heron Instruments Inc. The number of measurements on
water. That said, there are no precise estimates of the total
a single day varied from eight to fifteen, the main constraint
number of private borewells in the city. The BWSSB requires
being heavy traffic and the related (un)willingness of taxi dri­
properties owning borewells to register their borewells at
vers to drive the field assistants to multiple locations in a day.
a nominal cost if they receive BWSSB water supply. In areas
Field notes were transferred to electronic form each day. In any
without BWSSB supply, little is known about the number of
given month, all monitoring wells were covered within ten
borewells. The only census on of borewells was conducted in
days.
2004 for only one ward, where 873 borewells were counted in
a 2.9 km2 area (Raju, Manasi, and Latha 2008). In January 2010,
approximately 100,000 wells were reported to be registered. By
Groundwater model
March 2013, 175,000 borewells were registered (Subramanyam
2011). In February 2019, 370,000 wells appeared to be regis­ Modelling approaches of different complexity have been used,
tered (Alva 2019). The total number of borewells, including depending upon the availability of experimental data, size of
unregistered wells, could be two to three times more. the study area and study goals. In smaller watersheds with
At a coarse scale, for the larger Bengaluru urban district, the intensive field hydro-geological experiments, double porosity
CGWB has estimated that the groundwater draft is higher than net models have been employed (e.g. Sekhar, Kumar, and
groundwater availability in 2009 (Table 2 in CGWB 2011). Using Sridharan 1994; Maréchal, Dewandel, and Subrahmanyam
sparse piezometric time series data several researchers have also 2004). In other studies, as explained in the Study Area section,
pointed out that while the city as a whole is in groundwater over­ a composite unconfined aquifer has been assumed, that is
draft, there are some areas in which groundwater levels are rising homogeneous and isotropic (e.g. Sekhar and Kumar 2009;
(the older, core areas with low population growth, and piped, old Sekhar et al. 2013, 2017). In Bengaluru, the latter approach is
water supply and sewer infrastructure); whereas in others (newer used since detailed information about the aquifer is not avail­
rapidly growing outer areas with less water supply and greater able. The governing equation of groundwater flow in two
pumping), the overdraft is more severe (Hegde and Subhash dimensions for unconfined, homogeneous and isotropic con­
Chandra 2012; CGWB 2011; Mehta et al. 2014). From a water quality ditions can be written as (Todd and Mays 2005),
standpoint, almost 30% of the 2137 wells that were sampled across � � � �
the city had nitrate levels in excess of the permissible limit in 2011 @ @h @ @h Sy @h
h þ h ¼ þ Q; (1)
(45 mg/l) (DMG 2011). The Department of Mines and Geology @x @x @y @y K @t
which conducted the study noted that anthropogenic sources
like sewage and industrial effluents were the dominant cause of where h is the hydraulic head [L], Sy is the specific yield [-], K is the
groundwater pollution in Bengaluru. hydraulic conductivity [L/T], Q is the source/sink term [-], x and y are
The overall status of groundwater overdraft in Bengaluru the coordinates [L], and t is time [T].
has been reiterated, for the city as a whole, using lumped If the drawdown in the aquifer is very small compared to the
models (Hegde and Subhash Chandra 2012; Mehta et al. saturated thickness, h can be replaced with an average thick­
2014). Data from our earlier paper (Sekhar et al. 2017) indicates ness, b assumed to be constant over the aquifer, and replacing
 URBAN WATER JOURNAL 5

the term Kb by the transmissivity (T), Equation (1) can be re- observed that since the values of T are quite low, the lateral
written as, (Darcy) flux was insignificant compared to the vertical fluxes
and hence a constant value of T was used in the analysis.
@ 2 h @ 2 h Sy @h
þ ¼ þ Q=b: (2)
@x2 @y2 T @t
The 2-dimensional transient groundwater flow equation Estimation of model parameters
(Equation (2)) is solved using split operator approach. First, We used the GLUE approach for parameter estimation with
Equation (2) is solved vertically for the source/sink term (Q/b), uncertainties (Beven and Binley 1992). GLUE uses Monte Carlo
then it is solved horizontally for the lateral flow. A detailed simulations to identify a set of behavioural (optimal) para­
description of the numerical solution is available in Subash meters which can be used to simulate a corresponding ensem­
et al. (2017). The vertical flux (q) is computed as, ble of the variable of interest. GLUE has been successfully
applied to estimate parameters and their associated uncertain­
qi;t ¼ qi;t
r qid qi;t
b; (3)
ties in hydrological models (Kumar et al. 2010; Ruiz et al. 2010;
where qr is the recharge from all sources [L], qd is groundwater Hassan, Bekhit, and Chapman 2008). A detailed description of
draft (i.e. pumping) [L] and qb is baseflow [L]. Superscript GLUE can be found in (Sreelash et al. 2017).
i represents a spatial location (grid) and t represents time. The In the current study, Root Mean Squared Error (RMSE) is
qr is computed as, taken as the likelihood estimate which is computed as,
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
qi;t 1 X X � i;t �2
i;t i
r ¼ q r r þ qr l (4)
RMSE ¼ hsim hi;t obs ; (10)
where qr−r is the recharge due to rainfall, qr−l is the recharge n i t
due to leakage in water supply and wastewater drainage and
where n is the number of observations across grid and time.
are computed as,
The observed groundwater level in January 2016 was used
qi;t i t as the initial condition for the model. Then, the model was run
r r ¼ βr � R ; (5)
for one year at monthly time step for the year 2016. Initially,
a wide and plausible range of all parameter values was
qir l ¼ βil � WUi ; (6)
adopted, as shown in Table 1. A total of 100,000 model simula­
where R is rainfall, β is recharge factor for respective compo­ tions were performed with these ranges of parameters.
nent denoted by subscript, WU is the total water used from all The set of parameters comprising the behavioural ensemble
sources. Note that only recharge from rainfall varies temporally, was iteratively arrived at using the Containing Ratio (CR). The
due to the lack of temporal data about the other variables. WU CR represents the percentage of observed data bracketed by
is computed as the sum of water billed by BWSBB (WS) and qd, selected prediction limits (Xiong and O’Connor 2008). We
selected a CR that covered the 10 to 90 percentile spread.
WUi ¼ WSi þ qid (7)
Theoretically a value of 1 for CR represents the best prediction
The qd is computed as the difference between total water use of the uncertainty. In the first iteration, we assumed that the
and the water supplied by BWSBB, behavioural ensemble consists of only 10 best (ranked based
on the objective function) set and CR was computed. The size
qid ¼ /i � Pi WSi (8) of the parameter set was increased until the CR approached the
where P is the population and α is the per capita water use
Table 1. Parameters used for GLUE calibration.
across all sectors. Boundary conditions are flow-based where
Posterior range
flow is computed as in Park and Parker (2008), (behavioural
� � Parameter Definition Prior range ensemble mean)
qi;t
b ¼ max ð1 λi Þðhi;t himin Þ; 0 ; (9) α Per capita water use 100–200 lpcd 110–170 lpcd
βr Rainfall recharge 0.0–0.3 0.07–0.21
where λ is the parameter for baseflow computation and hmin is coefficient
the groundwater level at which baseflow ceases. Baseflow is βl Recharge coefficient for 0.3–0.7 0.43–0.67
assumed to be water lost from the system (modelling grid) to leakage
hmin The groundwater level 0 to 100 m below 750–885 m
stream or deeper aquifer. at which baseflow ground surface
ceases
λ Parameter for baseflow 0.8–1 0.9–0.95
computation
Input data
Data were collected from various agencies. Population (P),
metered BWSSB supply (WS) and monthly rainfall from 2016 Table 2. Mean and uncertainty in city aggregated fluxes.
(R) formed the major input data. Effective aquifer Sy and trans­ Flux Description Mean Std. CV Range (25–75 percentile)
missivity (T) were set constant at 0.005 and 1500 m2/month, WU Total water use 1470 48.0 0.033 1431–1499
respectively. The value of Sy was used in the earlier study by the qd Groundwater draft 827 48.0 0.058 789–857
qr−l Recharge from leaks 791 34.8 0.044 771–812
authors (Sekhar et al. 2017), who also provide a justification qr−r Rainfall recharge 183 2.2 0.012 181–184
from the literature. A sensitivity analysis was performed to qb Baseflow 186 4.2 0.023 183–189
understand the role of changes in T on the fluxes. It was Groundwater balance −40 49 −1.22 −69 – −13
6 S. K. TOMER ET AL.

theoretical value of 1. It was observed that out of the 100,000 wide aggregated fluxes. Total water use was estimated
simulations, a behavioural ensemble of 100 sufficiently exhib­ as 1470 (1431–1499) MLD. The range describes the
ited similar uncertainty in the simulated groundwater levels. 25–75 percentile estimates computed using the behavioural
ensemble, as described in the Methodology section. Spatial
distribution of this total water use is shown in Figure 3.
Results and discussion
Groundwater pumping meets the majority of this use, at 827
Groundwater level data, and interpolated water table maps for (789–857) MLD, followed by utility water supply, at 643 MLD.
each month, are publicly available at http://bangalore.urbanme Total recharge from all sources was estimated to be 973 (­
tabolism.asia/2017/10/17/groundwater-levels-from-2015-to 953–993) MLD, with natural recharge accounting for a much
-2017/and http://bangalore.urbanmetabolism.asia/2017/10/23/ smaller portion, 183 (181–184) MLD, compared to anthropo­
groundwater-table-maps/. Our earlier paper based on this data genic recharge from leaking water and wastewater systems of
discusses the groundwater level behavior in detail (Sekhar et al. 791 (771–812) MLD. After accounting for base-flow, the city as
2017). We focus here on the model-estimates of fluxes and a whole experienced a net negative groundwater balance at
uncertainties. −40 MLD (−69 – −13). Significantly, the total anthropogenic
recharge for the city as a whole is quite close in magnitude to
total groundwater extraction.
Model fit and parameter estimates
At the city-wide scale, all the fluxes have relatively low
A good fit was observed between the simulated and the grid uncertainties except net groundwater balance, using
averaged measured groundwater level data with RMSE of Coefficient of Variation (CV) as a measure of uncertainty
1.25 m and correlation coefficient of 0.91 (See Figure S1 in (Table 2). Groundwater draft has Coefficient of Variation (CV)
supplementary materials). of 5.8%. Rainfall recharge had lower uncertainty than ground­
Estimated parameters and fluxes are presented below, and water draft. Possible under – or over estimation in the uncer­
compared against secondary data and our earlier estimates tainty of the fluxes is described in Section 3.3. Net groundwater
from the WTF method (Sekhar et al. 2017). balance has a standard deviation very similar to that of total
Figure 3 shows the spatial distribution of the mean of beha­ water use and groundwater draft; however, its CV is high
vioral parameters for α and βr. The per capita water use (α) because its mean value is low.
varied from 110 to 170 LPCD (see Table 1). These estimates are The estimated total water use (1470 MLD) is consistent with
compared against an independent estimate from a residential the independent survey estimate of residential-only demand,
water use survey conducted in 2013 (Malghan et al. 2018; which at 2013 population estimates was 977 MLD (Malghan
Goswami 2017). The survey’s statistically weighted estimate of et al. 2018), and scaled to 2016 would be 1033 MLD. A corollary
residential per capita water use was 111 lpcd; ranging from 77 estimate of non-residential demand can thus be made
to 156 lpcd across five regions of the city. Our estimate of total (although only at a city-wide scale) from the difference, at
(i.e. both residential and non-residential) per capita water use 437 MLD.
should be – and is – higher than residential-only water use Natural recharge estimates are very close to our earlier WTF
estimated independently, and appears reasonable both at city- method-based estimate of 186 MLD (Sekhar et al. 2017). The
wide and spatially disaggregated scales. finding of anthropogenic recharge being much higher than
The rainfall recharge factor (βr) showed significant spatial natural recharge from rain is consistent with earlier lumped
variability with values ranging from 0.07 to 0.21 with a mean of model estimates (e.g. Mehta et al. 2013, 2014), and reflects
0.13 (Table 1). These estimates are very close to our earlier the urban water balances of many cities around the world as
estimate (βr = 0.135) using the simpler WTF method (Sekhar described in the introduction. A net negative groundwater
et al. 2017). Higher values of βr were observed in grids with less storage change of 2016 is consistent with our earlier estimate
built-up area, which is also consistent with Sekhar et al. (2017). based on the WTF method, which found a larger deficit at 51.6
βl showed a range of 0.43 to 0.67 with a mean value of 0.53. This MLD (Sekhar et al. 2017).
factor combines leakage from both water and wastewater systems.
For comparison, BWSSB water accounts from September 2013 esti­ Spatial variation of model estimates
mate leakage losses from water supply alone at 37.7%. The spatial variation in rainfall recharge qr−r is presented in
λ varied from 0.9 to 0.95 with a mean value of 0.94. A value of 1 Figure 3. Natural, rainfall recharge fluxes follow the pattern of
for λ means that the entire groundwater column (above hmin) will rainfall recharge coefficients (βr), as expected. A visual compar­
become base-flow in unit time step and a value of 0 represents zero ison with the land-use map (see Figure 2) confirms that ordinal
baseflow for any groundwater level. hmin showed a range of 750 to ordering of the grids in terms of rainfall recharge follows the
885 m with a mean value of 841 m. A significant correlation was land use patterns. Grid-27 and Grid-28 at southern edge of the
observed between hmin and elevation of the grid with a Pearson city that abuts a reserved forest are also predicted by our model
correlation coefficient of 0.78. to produce the highest level of rainfall recharge. The central
parts of the city, which are most densely built, show the lowest
levels of recharge. We see a factor of two variation in recharge
Estimated fluxes
rates between built parts of the city and places that are domi­
City aggregate nated by vegetation.
Figure 4 presents the mean city-wide estimated water fluxes Figure 3c shows the spatial variation in the estimated
in scaled context. Table 2 presents the uncertainty in the city- WU. A relatively higher WU is observed in the most
 URBAN WATER JOURNAL 7

Figure 3. Spatial distribution of ensemble means for (a) per capita water use, (b) rainfall recharge coefficient, (c) total water use, (d) recharge from rainfall (qr−r), (e)
groundwater draft (qd) and (f) net groundwater balance. (a) α(LPCD); (b) βr; (c) WU; (MLD); (d) qr−r (MLD); (e) qd (LPCD); (f) qnet (MLD).

densely populated area situated close to Rajajinagar and earlier estimates of net groundwater balance based on the
Chamarajpet and relatively lower WU is observed in the WTF method Sekhar et al. (2017).
periphery of the city. The observed variation is consistent
with spatial distribution of population centres (see Figure 2
Limitations
(b)), and spatial variation in surface water availability (see
Figure 2(d)). Pumping rates per capita (Figure 3(e) as LPCD) This study’s limitations stem mainly from the lack of sufficient
are higher in peripheral areas where the utility piped net­ granular understanding of many aspects of Bengaluru’s coupled
work is sparse (Figure 2(d)). social-hydrological system. We used simplifying assumptions which
Table 3 lists various components of groundwater balance are described below.
for each grid. qr−l is observed to be higher where population
density is higher and water supply is provided by BWSSB (see Water supply data
Figure 2). Figure 3(f) shows the spatial distribution of the net We have assumed that water used over and above BWSSB
groundwater balance at the end of the year 2016. Grids lying ‘metered and billed’ supply in a 25 km2 grid is met through
in the center of the city showed a positive groundwater groundwater pumping from within the grid. However, some
balance and those lying in the periphery showed a negative additional BWSSB water supply exists, that is not captured in
groundwater balance. This result is also consistent with our the ’metered and billed’ data. Some of this is ‘Non-Revenue
8 S. K. TOMER ET AL.

Figure 4. Schematic of the urban water budget. Mean city-wide fluxes are displayed to scale.

Table 3. Grid-wise mean water budget in MLD. Water’ (NRW), while the second could be called ’Used But Not
Grid Metered’ (UBNM) water. NRW is water that is ‘metered but not
ID Grid Name WU qr r qr l qr qb qd qnet billed‘. NRW is very small – 0.11% of water pumped from the
1 Ganganahalli 9.9 6 4.3 10.3 2.5 9.2 −1.4 Cauvery river in 2013, and 0.2% of billed water. This difference
2 Vidyaranyapura 24.5 8.7 11.5 20.1 9.1 16.2 −5.2 is too small to impact our results substantially.
3 Yelahanka New 28.9 4.1 17 21.1 4 15.2 1.9
Town The UBNM component consists of water used through
4 Agrahara 15.3 7.7 7.1 14.8 2 12.8 0 unauthorized connections, unmetered public taps and non-
5 Peenya 42.1 7.6 25.2 32.8 3.5 33.9 −4.5 functional meters. BWSSB estimated this UBNM component to
6 Jalahalli 61.8 6.6 35.3 41.9 6.5 35.4 0.1
7 Hebbal 65.8 7.6 34.9 42.5 0.6 44.4 −2.5 be 12.38% of the total water supplied within the city in 2013.
8 Hennur 48.7 7.6 28.4 36 0 40.6 −4.6 No spatially disaggregated estimates are available and after
9 Channasandra 17.1 7.2 8.6 15.7 4.1 17 −5.4 September 2013, BWSSB water accounts do not report UBNM.
10 Herohalli 71.9 7.6 45.2 52.9 0.9 53.5 −1.6
11 Rajajinagar 127.9 4 58.3 62.4 6.6 57.9 −2.1 Therefore, we do not include UBNM in our model for 2016.
12 Bangalore Palace 84.1 4 39.9 43.9 9.8 27.5 6.6 Although the supply deficit in a particular grid is met through
13 Cooke Town 74.4 3.5 35.8 39.3 5.8 29.2 4.4 groundwater pumping, this pumping need not be from within the
14 Krishnarajapuram 56.6 4.9 33.2 38.1 9.4 30 −1.3
15 Kodigehalli 25 6.1 12.9 19 5.2 17.8 −4 same grid. Water tankers, which depend on groundwater extraction,
16 Bangalore 43.3 6.1 29 35.1 17.2 18.9 −1 may be moving water across the grids. There is no systematic infor­
University mation available to address this issue. Similarly, bottled water used
17 Chamarajpet 136.4 6.1 82.9 89 18.7 73.7 −3.4
18 Shantinagar 79.3 6 40 46 13.6 25.9 6.6 within a grid is assumed to be largely from groundwater, which is
19 Domlur 51.4 4.5 24.7 29.2 5.7 20.6 2.9 likely a safe assumption; however, not all bottled water is sourced
20 HAL Airport 37 6 16.5 22.5 6.1 19.1 −2.7 from the city. The impact of this assumption is likely to be minimal
21 Whiteeld 20.6 6.1 10.2 16.3 4.4 15.8 −3.9
22 Kengeri 22.8 5 10.3 15.4 8 16.1 −8.6 since the residential water use survey from 2013 (Malghan et al. 2018)
23 Chikkasandra 71.9 6.5 36.8 43.4 11.2 38.8 −6.6 estimates that the volume of bottled water use in the city is minor
24 JP Nagar 89.9 6.6 54.2 60.9 11 43.9 6 compared to other sources.
25 HSR Layout 75.3 7.6 40.4 48.1 0 49.8 −1.7
26 Doddakannelli 21 7.6 10.4 17.9 8.7 16.1 −6.8
27 Kothnur 39.2 10.6 21.1 31.8 6 25.5 0.2 Effects of lakes
28 Begur 27.6 10.5 16.3 26.9 5.3 22.4 −0.8 Our modelling framework does not explicitly account for the
Totals 1470 183 791 973 186 827 −40
effects of lakes. Our objective was to sample the city at
 URBAN WATER JOURNAL 9

a sufficient resolution to estimate the groundwater budget at the rainwater harvesting and new water supply sources. We are
25 km2 scale, and not to understand the impact of lakes specifi­ currently working on developing and applying such scenarios.
cally. However, grid level estimates of groundwater fluxes should
theoretically reflect the influence of lakes – at least at the resolu­
tion of 25 km2 – to the extent that the groundwater measure­ Recommendations
ment network is able to capture these effects. Variogram analysis (1) It is unlikely that in the near future, large knowledge gaps in
presented in our earlier paper (Sekhar et al. 2017) suggests that the social hydrology of Indian cities can be filled by govern­
an investigation concerning influence of lakes would need an ment agencies alone. A citizen-science – crowdsourcing
even denser network of observation wells than we implemented. experiment placed within a participatory groundwater man­
agement context should be explored. A recent example on
Rainfall and recharge groundwater quality monitoring from Lebanon provides
Monthly rainfall data used as model input was based on an average useful insights, albeit at the scale of a village (Baalbaki
of several stations distributed across the city (described in Sekhar et al. 2019).
et al. (2017)). We used an average because of the lack of data quality (2) Model verification will need water quality and chemistry
flags in the dataset received, and the temporal inconsistency in investigation. We recommend that government agen­
data availability among different observation stations. Any errors in cies conduct regular sampling. Local universities should
this assumption could be artificially compensated for by the rainfall be encouraged to include such sampling and analysis in
recharge parameterization. However, the qualitative agreement of their curriculum and student theses.
the spatial rainfall recharge factor with the built-up area lends (3) Simultaneously, more effort must go into mapping the aqui­
a measure of confidence that the effect of spatially varying rainfall fer and aquifer properties. A 2013 report from the CGWB
may not be substantial at the grid and city scale on a monthly time mentioned only 12 exploratory wells, and 9 observation
step, especially given that rainfall recharge is estimated to be wells, some of which were used to derive hydrogeological
a small component of total groundwater recharge. parameters (Central Groundwater Board 2013). Challenges
Another limitation of our study is that we have lumped together include the dense urban environment, and the inability to
recharge from leaking wastewater and water supply pipelines. This control when static levels might occur, since pumping times
decision was made to prevent over-parameterization of the model. are determined at household/building level. To start with,
If a large database of fine scale groundwater level measurements is aquifer tests could be attempted in public parks, other public
generated over time, the wastewater and water leakages could be lands, university campuses and industrial estates.
estimated separately using the model and GLUE framework applied (4) Overall, an institutional framework must be created that
in this study. can manage these different efforts. Such an institutional
framework must include mechanisms for data curation
Conclusions and recommendations and analysis, converting that to sound policy, and for
enforcing regulations stemming from that policy.
Quantification of the groundwater budget is necessary for effective
groundwater management. As in most developing country cities,
this is a challenge in urban India given serious data gaps on supply Acknowledgements
and demand, and inadequate groundwater monitoring. We have Research support from the Cities Alliance Catalytic Fund and the Stockholm
demonstrated how groundwater level monitoring at a fine resolu­ Environment Institute’s City Health and Well-Being Initiative is greatly
tion can be used with models to provide useful estimates of the appreciated. K. Balakrishnan was supported by the PEAK Urban program,
urban (ground) water budget. Given the data gaps, any such funded by UKRI’s Global Challenge Research Fund. We are grateful to field
estimate of the budget should be accompanied by an assessment assistants Sanjeeva Murthy and P. Giriraj, who measured the groundwater
levels. Malghan was supported by the Ministry of Housing and Urban
of its uncertainty. In this paper, we have developed and applied Affairs, Government of India, through an IIMB-CoE grant.
a novel framework which achieves this objective.
We implemented a dense sampling framework adequate for
analysis at a 25 km2 grid. For devising neighbourhood scale Disclosure statement
groundwater management, in a city as heterogeneous as
No potential conflict of interest was reported by the author(s).
Bengaluru, an even denser sampling frame would be needed.
Simultaneously, several types of social and infrastructural data
would be needed at the same scale. Funding
An important finding of this study is the large proportion of
This work was supported by the City Alliance Catalytic Fund [CA/CATF/356/
recharge from leaking water and wastewater systems, reminiscent
01]; Stockholm Environment Institute City Health and WellBeing Initiative
of findings from several other large cities as noted in the [CHeW]; UKRI Global Challenge Research Fund [PEAK Urban Program/ES/
Introduction. Discrimination between water and wastewater lea­ P011055/1]; and Ministry of Housing and Urban Affairs, Government of
kages will require more detailed investigation, involving additional India Grant ID: IIMB-CoE.
secondary datasets and primary investigations regarding chemical
species that are known to be associated with those sources
(Vázquez-Suñé, Sánchez-Vila, and Carrera 2005). ORCID
Models are useful for testing future development scenarios, K. Balakrishnan http://orcid.org/0000-0003-1753-1353
e.g. involving demand growth, treated wastewater reuse, S. Thiyaku http://orcid.org/0000-0002-4523-8907
10 S. K. TOMER ET AL.

References Hurst, C. W., and W. B. Wilkinson. 1986. “Rising Groundwater Levels in

Cities.” Geological Society, London, Engineering Geology Special
Alva, N. 2019. “Every Day, at Least 6 Seek Nod to Dig Borewells in Publications 3 (1): 75–80. doi:10.1144/GSL.ENG.1986.003.01.07.
Bengaluru.” The Times of India, 14 February. Jaglin, S. 2014. “Rethinking Urban Heterogeneity.” In The Routledge
Baalbaki, R., S. H. Ahmad, W. Kays, S. N. Talhouk, N. A. Saliba, and A.- Handbook on Cities of the Global South, edited by Susan Parnell and
H. Mahmoud. 2019. “Citizen Science in Lebanon—a Case Study for Sophie Oldfield, 434–446. Abingdon: Routledge.
Groundwater Quality Monitoring.” Royal Society Open Science 6 (2): Kennedy, C., J. Cuddihy, and J. Engel-Yan. 2007. “The Changing Metabolism of
181871. doi:10.1098/rsos.181871. Cities.” Journal of Industrial Ecology 11 (2): 43–59. doi:10.1162/jie.2007.1107.
Balakrishnan, K. 2016. “Heterogeneity within Indian Cities: Methods for Kim, Y.-Y., K.-K. Lee, and I. Sung. 2001. “Urbanization and the Groundwater
Empirical Analysis.” PhD diss., University of California, Berkeley. Budget, Metropolitan Seoul Area, Korea.” Hydrogeology Journal 9 (4):
Beck, H. E., N. E. Zimmermann, T. R. McVicar, N. Vergopolan, A. Berg, and E. F. Wood. 401–412. doi:10.1007/s100400100139.
2018. “Present and Future Köppen-Geiger Climate Classification Maps at 1-km KSPCB. 2013. A Report of Monitoring Committee on Performance of
Resolution.” Scientific Data 5 (1): 180214. doi:10.1038/sdata.2018.214. Wastewater Treatment Plants in the Apartments and Other
Beven, K., and A. Binley. 1992. “The Future of Distributed Models: Model Establishments in BBMP Limits of Bangalore. Bangalore: Karnataka State
Calibration and Uncertainty Prediction.” Hydrological Processes 6 (3): Pollution Control Board.
279–298. doi:10.1002/hyp.3360060305.
Kumar, S., M. Sekhar, D. V. Reddy, and M. Kumar. 2010. “Estimation of Soil
Brassington, F. C. 1990. “Rising Groundwater Levels in the United Kingdom.”
Hydraulic Properties and Their Uncertainty: Comparison between
Proceedings of Institution of Civil Engineers. Part 1 88 (6): 1037–1057.
Laboratory and Field Experiment.” Hydrological Processes 24 (23):
doi:10.1680/iicep.1990.11616.
3426–3435. doi:10.1002/hyp.7775.
Brassington, F. C., and K. R. Rushton. 1987. “A Rising Water Table in Central
Lerner, D. N. 2002. “Identifying and Quantifying Urban Recharge: A Review.”
Liverpool.” Quarterly Journal of Engineering Geology and Hydrogeology 20
Hydrogeology Journal 10 (1): 143–152. doi:10.1007/s10040-001-0177-1.
(2): 151–158. doi:10.1144/GSL.QJEG.1987.020.02.05.
Malghan, D., S. Lele, V. K. Mehta, G. Taneja, D. Sharma, A. Nehra, and
BWSSB. 2012. “Annual Performance Report 2011–12.” [Online]. Accessed 26
R. Goswami. 2018. Source-disaggregated Domestic Water Consumption
April 2018. https://bwssb.gov.in/content/about-bwssb-2
Patterns in Bengaluru, India. Bengaluru: Ashoka Trust for Ecology and
BWSSB. 2017a. “About BWSSB: Overview of the Sewerage System.” [Online].
Environment (ATREE).
Accessed 26 April 2018]. https://bwssb.gov.in/content/about-bwssb-2
Maréchal, J.-C., B. Dewandel, and K. Subrahmanyam. 2004. “Use of Hydraulic
BWSSB. 2017b. “Plans, Roles, Duties and Responsibilities of Chief Engineer
Tests at Different Scales to Characterize Fracture Network Properties in
(Waste Water Management).” [Online]. Accessed 26 April 2018. https://
the Weatheredfractured Layer of a Hard Rock Aquifer.” Water Resources
bwssb.gov.in/sites/default/files/ss/CE\%20(WWM)\%20English.docx
Research 40 (11). doi:10.1029/2004WR003137.
Central Groundwater Board. 2013. Ground Water Information Booklet:
Mehta, E. K.-B., R. Goswami, S. Muddu, and D. Malghan. 2013. “Social
Bangalore Urban District, Karnataka. Faridabad: Ministry of water
Ecology of Domestic Water Use in Bangalore.” Economic and Political
Resources, Government of India.
Weekly 48 (15): 20–30.
Central Groundwater Board. 2016. Ground Water Year Book of Karnataka State
Mehta, R. G., E. Kemp-Benedict, S. Muddu, and D. Malghan. 2014. “Metabolic
2015–2016. Faridabad: Ministry of Water Resources, Government of India.
Urbanism and Environmental Justice: The Water Conundrum in Bangalore,
CGWB. 2011. “Ground Water Scenario in Major Cities of India.” Technical
India.” Environmental Justice 7 (5): 130–137. doi:10.1089/env.2014.0021.
Report. Central Ground Water Board of India.
Decker, E. H., S. Elliott, F. A. Smith, D. R. Blake, and F. Sherwood Rowland. Misra, S., and B. Goldar. 2008. “Likely Impact of Reforming Water Supply and
2000. “Energy and Material Flow through the Urban Ecosystem.” Annual Sewerage Services in Delhi.” Economic and Political Weekly 43 (41): 57–66.
Review of Energy and the Environment 25 (1): 685–740. doi:10.1146/ Narain, S., and P. Pandey. 2012. Excreta Matters: How Urban India Is Soaking
annurev.energy.25.1.685. up Water, Polluting Rivers and Drowning in Its Own Waste. New Delhi:
DMG. 2011. “Groundwater Hydrology and Groundwater Quality in and Centre for Science and Environment.
around Bangalore City.” Technical Report. Bangalore: Department of Park, E., and J. C. Parker. 2008. “A Simple Model for Water Table Fluctuations
Mines and Geology. in Response to Precipitation.” Journal of Hydrology 356 (3): 344–349.
Foster, S., A. Lawrence, and B. Morris. 1998. Groundwater in Urban doi:10.1016/j.jhydrol.2008.04.022.
Development: Assessing Management Needs and Formulating Policy Planning Commission. 2013. Twelfth Five Year Plan (2012–2017). New Delhi:
Strategies. Washington, D.C.: World Bank. Govt of India.
Gattinoni, P., and L. Scesi. 2017. “The Groundwater Rise in the Urban Area of Raju, K. V., S. Manasi, and N. Latha. 2008. “Emerging Groundwater Crisis in Urban
Milan (Italy) and Its Interactions with Underground Structures and Areas – A Case Study of Ward No.39, Bangalore City.” Technical Report.
Infrastructures.” Tunnelling and Underground Space Technology 62: Ruiz, L., M. R. R. Varma, M. S. Mohan Kumar, M. Sekhar, J.-C. Maréchal,
103–114. doi:10.1016/j.tust.2016.12.001. M. Descloitres, J. Riotte, S. Kumar, C. Kumar, and J.-J. Braun. 2010.
Government of India. 2011. “Provisional Population Totals, Census of India, 2011.” “Water Balance Modelling in a Tropical Watershed under Deciduous
Chapter 3:27-60. Office of the Registrar General, Ministry of Home Affairs. New Forest (Mule Hole, India): Regolith Matric Storage Buffers the
Delhi. Groundwater Recharge Process.” Journal of Hydrology 380 (3): 460–472.
Gilbert, N. 2012. “Water under Pressure: A UN Analysis Sets Out Global doi:10.1016/j.jhydrol.2009.11.020.
Water Management Concerns Ahead of Earth Summit.” Nature 483 Schandl, H., and A. Capon. 2012. “Cities as Social-ecological Systems: Linking
(7389): 256–258. doi:10.1038/483256a. Metabolism, Wellbeing and Human Health.”Current Opinion in Environmental
Goswami, R. 2017. “Hydrosocial Metabolism of Bangalore City: A Comprehensive Sustainability 4 (4): 375–377. doi:10.1016/j.cosust.2012.09.001.
Study of Urban Water Consumption.” PhD diss., Indian Institute of Sekhar, M., M. Shindekar, S. K. Tomer, and P. Goswami. 2013. “Modeling the
Management Bangalore. Vulnerability of an Urban Groundwater System Due to the Combined
Hassan, A. E., H. M. Bekhit, and J. B. Chapman. 2008. “Uncertainty Assessment of Impacts of Climate Change and Management Scenarios.” Earth
a Stochastic Groundwater Flow Model Using GLUE Analysis.” Journal of Interactions 17 (10): 1–25. doi:10.1175/2012EI000499.1.
Hydrology 362 (1): 89–109. doi:10.1016/j.jhydrol.2008.08.017. Sekhar, M., and M. S. M. Kumar. 2009. “Geo-hydrological Studies along the Metro
Hayashi, T., T. Tokunaga, M. Aichi, J. Shimada, and M. Taniguchi. 2009. Rail Alignment in Bangalore.” Technical Report. Technical report, Department
“Effects of Human Activities and Urbanization on Groundwater of Civil Engineering, Indian Institute of Science, Bangalore, India.
Environments: An Example from the Aquifer System of Tokyo and the Sekhar, M., M. S. M. Kumar, and K. Sridharan. 1994. “A Leaky Aquifer Model for Hard
Surrounding Area.” Science of the Total Environment 407 (9): 3165–3172. Rock Aquifers.” Applied Hydrogeology 2 (3): 32–39. doi:10.1007/s100400050039.
doi:10.1016/j.scitotenv.2008.07.012. Sekhar, M., S. K. Tomer, S. Thiyaku, P. Giriraj, S. Murthy, and V. K. Mehta.
Hegde, G. V., and K. C. Subhash Chandra. 2012. “Resource Availability for Water 2017. “Groundwater Level Dynamics in Bengaluru City, India.”
Supply to Bangalore City, Karnataka.” Current Science 102 (8): 1102–1104. Sustainability 10 (1): 26. doi:10.3390/su10010026.
 URBAN WATER JOURNAL 11

Sreelash, K., S. Buis, M. Sekhar, L. Ruiz, S. K. Tomer, and M. Guerif. 2017. Todd, D. K., and L. W. Mays. 2005. Groundwater Hydrology Edition. Vol. 1625.
“Estimation of Available Water Capacity Components of Two-layered Soils New Jersey: Wiley.
Using Crop Model Inversion: Effect of Crop Type and Water Regime.” Journal Vázquez-Suñé, E., X. Sánchez-Vila, and J. Carrera. 2005. “Introductory Review of
of Hydrology 546: 166–178. doi:10.1016/j.jhydrol.2016.12.049. Specific Factors Influencing Urban Groundwater, an Emerging Branch of
Srinivasan, V., S. M. Gorelick, and L. Goulder. 2010. “A Hydrologic-economic Hydrogeology, with Reference to Barcelona, Spain.” Hydrogeology Journal
Modeling Approach for Analysis of Urban Water Supply Dynamics in 13 (3): 522–533. doi:10.1007/s10040-004-0360-2.
Chennai, India.” Water Resources Research 46 (7). doi:10.1029/2009WR008693. WHO. 1993. Guidelines for Drinking Water Quality. 2nd ed. Geneva:
Subash, Y., M. Sekhar, S. K. Tomer, and A. K. Sharma. 2017. “A Framework for Switzerland.
Assessment of Climate Change Impacts on the Groundwater System.” Wolman, A. 1965. “The Metabolism of Cities.” Scientific American 213 (3):
Chap. in Sustainable Water Resources Management, edited by T. Zhang, 178–193. doi:10.1038/scientificamerican0965-178.
C. S. P. Ojha, S. Rao, and A. Bardossy, 14. Virginia: ASCE. Xiong, L., and K. M. O’Connor. 2008. “An Empirical Method to Improve the
Subramanyam, S. 2011. “BWSSB Trying to Register All the Borewells in Prediction Limits of the GLUE Methodology in Rainfall–runoff Modeling.”
Bangalore.” Citizen Matters, 3 November. Journal of Hydrology 349 (12): 115–124. doi:10.1016/j.jhydrol.2007.10.029.