Journal of Hydrology (2007) 334, 151–161

ava i lab le a t www.sc iencedi rec t . com

journal homepage: www.elsevier .com/ locate / jhydrol

Regional-scale stable isotopic signatures ofrecharge and deep groundwater in thearsenic affected areas of West Bengal, India

Abhijit Mukherjee *, Alan E. Fryar, Harold D. Rowe

Department of Earth and Environmental Sciences, University of Kentucky, Lexington, KY 40506-0053, USA

Received 24 February 2006; received in revised form 2 October 2006; accepted 3 October 2006

00do

*

GeAuTe

je

KEYWORDSStable isotopes;Recharge;Groundwater;West Bengal;India

22-1694/$ - see front mattei:10.1016/j.jhydrol.2006.10

Corresponding author. Prology, Jackson School ofstin, University Station, Bl.: +1 512 232 1527; fax:E-mail address: abhijit.mu

e).

r ª 200.004

esent adGeoscienox X, A+1 512 4kherjee@

Summary Samples of deep groundwater, river water and rainwater were collected ford18O and d2H analyses from an area of �22,000 km2 in the arsenic-affected districts ofWest Bengal, India, in order to examine groundwater recharge. A plot of d18O versusd2H of groundwater falls subparallel to the constructed local meteoric water line(d2H = 7.2 d18O + 7.7), suggesting a predominance of meteoric recharge with some evap-oration. The stable isotopic signature of groundwater from the deeper part of the semi-confined main aquifer is similar to that of shallower groundwater, which suggests thatdeeper groundwater has been recharged in the present-day climatic regime, mostly frommonsoonal rainfall. Groundwater in deeper isolated aquifers falls within the isotopic rangeobserved for the main aquifer. A trend of isotopic depletion of groundwater that extendsnorthward and westward from the Bay of Bengal (the monsoonal moisture source area)indicates a rain-out process following Rayleigh-type distillation. Some recharge may havetaken place from the rivers Ganges and Bhagirathi-Hoogly, with minimal inflow from otherrivers (Jalangi and Ichamati).ª 2006 Elsevier B.V. All rights reserved.

6 Elsevier B.V. All rights reserved

dress: Bureau of Economicces, University of Texas atustin, TX 78713-8924 USA.71 0140.beg.utexas.edu (A. Mukher-

Introduction

The Bengal basin, located in eastern India and Bangladesh,is the largest fluvio-deltaic system in the world (Coleman,1981). The basin has been formed by the confluence oftwo of the world’s largest river systems, the Ganges andthe Brahmaputra, along with other important rivers likethe Meghna and Damodar. The physiography of the westernpart of the basin (eastern portion of the state of West

.

152 A. Mukherjee et al.

Bengal, India) is dominated by the River Ganges, its maindistributary the River Bhagirathi-Hoogly, and their tributar-ies and distributaries. The western part of the Bengal basin,with a human population of more than 40 million (one of thedensest in the world), is also referred to as Gangetic WestBengal (Fig. 1).

Most of the residents of the Bengal basin rely ongroundwater for domestic use and irrigation of staple crops.Unfortunately, widespread, non-point-source arsenic con-tamination in shallow groundwater (�20–80 m belowground level) poses a health risk to approximately 50 millionresidents of the basin (Smith et al., 2000). Moreover, sur-face water in the area, which is contaminated by sewage

Figure 1 Map of the study area showing groundwat

and industrial waste, is not potable without costly treat-ment. In West Bengal, government and municipal agencieshave drilled several hundred deep wells (generally around100 m or more, depending on the lithology) in order to pro-vide uncontaminated water for rural and urban populations.However, the suitability and sustainability of the deepergroundwater of the Bengal basin and its interaction withthe shallower groundwater has received little study.

During the last two decades, multiple research groupshave studied the groundwater dynamics and hydrogeochem-istry of the Bengal basin using a variety of techniques. Thesehave included stable isotopic studies of groundwater re-charge (Shivanna et al., 1999; Aggarwal et al., 2000; Basu

er, river-water and rainwater sampling locations.

Regional-scale stable isotopic signatures of recharge 153

et al., 2002; Stuben et al., 2003; Harvey et al., 2005; Zhenget al., 2005; Klump et al., 2006). However, no previous ba-sin-scale isotopic study has been done in West Bengal. Thispaper documents and discusses stable isotopic results froma multiyear study of deeper groundwater in Gangetic WestBengal. The objective is to delineate the source(s) ofgroundwater recharge by identifying regional trends in the18O and 2H compositions of groundwater, surface waterand rainwater.

Study area setting

The study area comprises �22,000 km2 of the arsenic-af-fected areas of five districts in southeastern West Bengal.These are the southernmost part of Maldah, eastern Mursh-idabad, Nadia, North 24 Parganas and northern and centralSouth 24 Parganas (Fig. 1). The area lies �200 km south ofthe eastern (Darjeeling) Himalayas and is bounded in thewest by the rivers Ganges and Bhagirathi-Hoogly, by the In-dia–Bangladesh border in the east, and the Bay of Bengal inthe south. The River Ganges bifurcates near Farakka to formthe River Padma (flowing toward Bangladesh) and the RiverBhagirathi-Hoogly (flowing southward through West Bengal)along the boundary between Maldah and Murshidabad. Sev-eral other important streams are also present in the studyarea, including the rivers Jalangi and Ichamati (see Fig. 1).

Due to its proximity to the Bay of Bengal and its low lat-itude (the Tropic of Cancer passes through the study area),Gangetic West Bengal experiences a warm and humid cli-mate with distinct dry and wet seasons. These may be

Figure 2 Simplified hydrostratigraphic diagram of the study area (this paper (see Mukherjee, 2006 for details). Note that no data we300 m from mean sea level, so those areas have been marked as un

termed as pre-monsoon (January–May), monsoon (June–October) and post-monsoon (November–December). Thebulk of the precipitation takes place as a result of mois-ture-laden monsoon winds from the Bay of Bengal. Localconvectional storms contribute to some pre-monsoonalrainfall. On average (based on Global Historic ClimatologyNetwork [GHCN] version 2 data from 1901 to 1970, GHCN,2005), the pre-monsoon period contributes 16.23% of theannual rainfall, monsoon 82.21% and post-monsoon 1.57%.It is hypothesized that most of the groundwater in the studyarea is recharged from monsoonal precipitation.

Mukherjee (2006) delineated the hydrostratigraphy ofthe study area up to a depth of 300 m below mean sea level,excluding the part of the study area in Maldah (Fig. 2). Themain aquifer is a continuous, semiconfined sand body under-lain by a thick clay aquitard. This aquifer deepens from amaximum of �50–80 m below land surface in the north to�180 to >200 m in the south. In the southern half of thestudy area, discontinuous clay layers locally divide the mainaquifer to form several deeper, confined aquifers. Severalsmaller, isolated aquifers (termed hereafter as isolated)at greater depths than the main aquifer (�200–300 m) con-tain groundwater with considerably different solute compo-sitions (Mukherjee, 2006). Finally, some deep parts of themain aquifer are not isolated but appear to demonstrate re-stricted conditions (intermediate between the main and iso-lated aquifers). The hydrostratigraphic positions of the foursampled wells in Maldah and the two sampled wells that arescreened below 300 m are not known, but the latter twowells are grouped with the isolated aquifers for the pur-poses of this paper.

not to scale) showing the different type of aquifers discussed inre available for the part of the study area in Maldah and belowdefined (U).

154 A. Mukherjee et al.

Methods

Sampling

Sixty-four groundwater, seven river-water and 14 rainwatersamples were collected during the months of May to Augustof 2003–2005 (Fig. 1, Table 1). During 2003, unfilteredgroundwater samples were collected after �20–30 min ofpurging, and field parameters were not measured. During2004 and 2005, groundwater was sampled according to stan-dard protocols (Wood, 1981). The wells were purged for 40–75 min until EH, pH, temperature and electrical conductivitystabilized. Each sample was then collected through a dis-posable 0.45 lm plastic capsule filter. River water was sam-pled using a hand-filtration unit with the same kind of filter.As much as possible, each rainwater sample was collectedduring an entire storm to capture an average composition.

Groundwater samples were collected from deep, com-munity water supply wells with single intake zones (strain-ers). These wells (except the one in Calcutta) are ownedand operated by the Public Health Engineering Directorate(PHED), Government of West Bengal. Detailed informationabout the well completion and tapped aquifer was obtainedfrom the PHED database and a local drilling agency (B.M.Engineering, KrishnaNagar). Similar information was notavailable for the four wells in the district of Maldah, butthey were ascertained to have single strainers (personalcommunication with PHED officers). The wells are generallyscreened at depths of 50–409 m below ground level (bgl),with the overall trend of increasing well depth toward thesouth and east of the study area.

Analyses

Isotope analyses for 2003 and 2004 samples were performedat the University of Kentucky Environmental Research andTraining Laboratory (UK ERTL) on a ThermoFinnigan DeltaPlus XP isotope ratio mass spectrometer (IRMS) using a GasBench II peripheral unit equipped with a GC-autosampler.The analyses were standardized with the international ref-erences Vienna-Standard Mean Ocean Water (VSMOW),Greenland Ice Sheet Project (GISP), and Standard Light Ant-arctic Precipitation (SLAP), as well as internal standards.For d18O, 0.5 mL of each standard and sample were flushedfor 1 min with 0.5% CO2 in He, then left to equilibrate atroom temperature on the gas bench for 48 h before beinganalyzed. For d2H, 0.5 mL of each standard and sample weretransferred into individual exetainers containing platinumrods and flushed for three minutes with 2% H2 in He. Thesamples were then left to equilibrate in a hot block in thegas bench for 16 h at 22 �C before analysis. Precision was±0.1& (VSMOW) for d18O and ±1.0& (VSMOW) for d2H.

Analyses of 2005 isotope samples were performed atthe University of Arizona Isotope Geochemistry Laboratoryon a gas-source IRMS (Finnigan Delta S) standardized withVSMOW, SLAP, and internal standards. The d18O sampleswere equilibrated with CO2 at �15 �C in an automated equil-ibration device coupled to the IRMS. For d2H, samples werereacted with Cr at 750 �C using a Finnigan H device coupledto the spectrometer. Precision was ±0.08& (VSMOW) ford18O and ±0.9& (VSMOW) for d2H.

Results

Because of the small number of samples, rainwater datawere not used to calculate weighted annual mean composi-tions, but were used to generate a local meteoric water line(LMWL) for Gangetic West Bengal. As noted in Table 1, 11 ofthe 14 samples were collected during 2004; the remainingthree were collected in 2005. The samples represent Mayto August; 10 samples were from the monsoon season andfour from the pre-monsoon. In 2004, the monsoon in WestBengal began on 12 June (news release from Regional Mete-orological Office, Alipur, Calcutta; Anandabazar Patrika,Calcutta, 12 June 2004, p. 1). Calcutta received 156 mmof rainfall (9.39% of 1661 mm total for 2004) during thepre-monsoon (January to May) and 1505 mm (90.61%) fromthe monsoon (June to October). No rainfall occurred inNovember and December. Therefore, assuming that the restof the study area had a similar pattern and proportion ofprecipitation (as evident from the seasonal means statedin ‘‘Study area setting’’), our rainfall dataset should fairlyrepresent the isotopic composition of precipitation in Gan-getic West Bengal.

As shown in Fig. 3, the LMWL plotted from our data(d2H = 7.2 d18O + 7.7 [r2 = 0.93]) compares well to the globalmeteoric water line (GMWL) of Craig (1961) (d2H = 8d18O + 10). The LMWL is quite similar to the line developedby Bhattacharya et al. (1985) for precipitation samples fromNew Delhi, Ahmedabad and Shillong, India (d2H = [7.2 ± 0.1]d18O + [5.1 ± 0.1]). Our LMWL slope agrees with the value of�7.2 suggested by Krishnamurthy and Bhattacharya (1991)for average Indian monsoonal precipitation. Fig. 3 also dis-tinguishes between the monsoon and pre-monsoon rains.Excluding sample R8, the monsoonal d18O values (�4.7&to �9.3&) are at least 5& lighter than those of the pre-monsoon (4.4–0.3&) (Table 1). This suggests that thepre-monsoon rains may originate from isotopically enrichedlocal moisture sources (such as evaporated surface waterbodies), whereas the monsoon rain clouds have traveled agreater distance and had multiple fractions before precipi-tation. Similarly, for Global Network of Isotopes in Precipi-tation (GNIP) sites at New Delhi and Yangon, Myanmar,mean d18O values are lighter for monsoonal precipitationthan for dry-season precipitation (Araguas-Araguas et al.,1998).

All the river-water samples plot below the LMWL (Fig. 4).The Bhagirathi-Hoogly samples, with d18O values of �5.6&to �5.7&, are more depleted than the Jalangi and Ichamatisamples, with d18O of �3.2& to �0.4& (Table 1). The Bha-girathi-Hoogly values are more enriched than other reportedvalues for that river (Shivanna et al., 1999; Stuben et al.,2003). No data from other studies appear to exist for theJalangi and Ichamati. One of the Ichamati samples (river-water location [RW] # 7, Table 1) was much more enrichedthan the other river-water samples.

With the exception of location # 64, the d18O values ofgroundwater samples range from �3.1& to �6.1& (Table1 1). Of the 64 samples, 40 have d18O values more depletedthan the mean (�4.8&) of all samples. Five of six isolatedaquifer samples and 19 of the 54 main aquifer samples weremore enriched than the mean, which indicates that the iso-lated groundwaters tend to be isotopically heavier than the

Table 1 Sample locations for groundwater, river water and rainwater from West Bengal

Location Collection date Village/Location Administrativeblock/Town

Districta Latitude Longitude Depthb Aquiferc d18O (&) d2H (&)

1 17/5/03 Banipur Habra town NP 22.82 88.69 150 M �4.5 �242 27/5/03 Nabipur RaniNagar-II MD 24.25 88.60 243 M �5.9 �363 27/5/03 Islampur RaniNagar-I MD 24.15 88.47 88 M �5.9 �374 18/6/03 Kalighat Calcutta SP 22.52 88.34 120 M �5.6 �365 3/7/03 Dattapara Shantipur town ND 23.25 88.43 128 M �5.2 �356 16/5/03 Berachampa

(Deulia)Deganga NP 22.70 88.68 M �4.9 �31

7 23/5/03 Sujapur(Bakharpur)

Kaliachak-I MA 24.92 88.09 U �6.1 �38

8 23/5/03 Baishnabnagar(Dariapur)

Kaliachak-III MA 24.82 87.99 U �5.7 �34

9 23/5/03 Mothabari Kaliachak-II MA 24.94 88.06 U �5.8 �3710 23/5/03 Sahapur Old Maldah town MA 25.01 88.15 U �5.6 �3311 24/5/03 Aurangabad Suti-II MD 24.62 88.03 M �5.6 �3412 20/5/04 Lalgola Lalgola MD 24.42 88.25 66 M �5.7 �3613 21/5/04 Lalbag Morgue Murshidabad town MD 24.23 88.27 32 M �4.6 �3014 21/5/04 Kantanagar Bhagwangola-I MD 24.34 88.31 66 M �5.7 �3515 22/5/04 Edrakpur Nawda MD 23.91 88.46 82 M �5.2 �3116 22/5/04 Takipur Beldanga-I MD 23.87 88.25 55 M �5.1 �3117 23/5/04 Chaighari Behrampur MD 24.16 88.41 228 I �3.9 �2418 24/5/04 Sekendra Raghunathganj-I MD 24.50 88.08 37 M �5.2 �3619 25/5/04 Behrampur Girls’

CollegeBehrampur town MD 24.09 88.26 74 M �5.1 �32

20 25/5/04 Debagram Kaliganj ND 23.69 88.33 78 M �4.2 �2721 26/5/04 Juranpur Kaliganj ND 23.72 88.22 155 M �5.4 �3522 26/5/04 Birpur Nakashipara ND 23.63 88.48 185 M �5.8 �3723 27/5/04 Bethuadahari Nakashipara ND 23.61 88.43 89 M �5.6 �3324 27/5/04 Jumpukur Kaliganj ND 23.72 88.38 269 M �3.6 �2025 28/5/04 Hanspukuria Tehatta-II ND 23.74 88.48 230 I �5.1 �3226 29/5/04 Gagnapur Ranaghat-II ND 23.14 88.64 176 M �4.9 �3127 1/6/04 Berachampa

(Aminpur)Deganga NP 22.69 88.67 298 M �3.4 �20

28 2/6/04 Janapoll Habra-I NP 22.87 88.69 151 M �3.8 �2629 2/6/04 Chikanpara Gaighata NP 22.93 88.77 409 >300 �3.7 �2230 3/6/04 Dalalpara Bashirhat town NP 22.69 88.85 230 M �3.4 �1931 3/6/04 Bansjhari

MullickpurBashirhat-I NP 22.58 88.88 273 I �3.1 �17

32 4/6/04 Berachampa Deganga NP 22.70 88.68 155 M �5.2 �3333 4/6/04 Eojhnagar Bashirhat-II NP 22.71 88.72 151 M �4.5 �2834 5/6/04 Ghasiara Sonarpur SP 22.44 88.46 240 M �3.5 �2035 5/6/04 Kalikapore Sonarpur SP 22.40 88.48 240 M �4.8 �3036 7/6/04 Canning Canning-I SP 22.32 88.65 189 M �3.6 �1937 16/5/05 Habaspur Bhagawangola-I MD 24.32 88.34 76 M �6.1 �4038 17/5/05 Chua Hariharpara MD 24.03 88.45 84 M �4.7 �3639 17/5/05 Surangapur Nawda MD 23.93 88.46 96 M �5.2 �3340 18/5/05 Mahimapur Palace Murshidabad MD 24.21 88.26 32 M �5.0 �3141 18/5/05 Kismat-Imadpur Hariharpara MD 24.03 88.35 71 M �4.7 �3142 19/5/05 Nazirpur Tehatta-I ND 23.86 88.54 110 M �5.1 �3443 19/5/05 Narayanpur Karimpur-II ND 23.92 88.55 108 M �5.1 �3444 20/5/05 Palashipara Tehatta-II ND 23.79 88.46 95 M �5.1 �3545 20/5/05 Iswarchandrapur Tehatta-I ND 23.76 88.46 217 I �4.5 �3046 21/5/05 Dhubulia KrishnaNagar-II ND 23.50 88.45 132 M �5.4 �3647 21/5/05 Sonatala KrishnaNagar-II ND 23.52 88.51 140 M �5.1 �3448 22/5/05 Gobrapota KrishnaNagar-I ND 23.42 88.55 120 M �5.2 �3049 22/5/05 Bhaluka KrishnaNagar-I ND 23.35 88.41 279 M �4.3 �2950 23/5/05 Bhimpur Krishnanaga-I ND 23.43 88.62 211 M �5.6 �3651 26/5/05 Newtown Rajarhat NP 22.58 88.45 132 M �5.1 �3352 26/5/05 Patharghata Rajarhat NP 22.58 88.51 120 M �4.4 �3053 27/5/05 Ghentugachia Chakdah ND 23.03 88.59 129 M �3.6 �2954 27/5/05 Kamagachi Ranaghat-I ND 23.22 88.56 152 M �5.8 �4355 28/5/05 Panchita Bongaon NP 23.11 88.80 193 M �5.4 �3456 28/5/05 Bongaon Hospital Bongaon town NP 23.05 88.82 202 M �4.9 �3057 29/5/05 Rampur Gaighata NP 22.95 88.75 200 M �4.9 �3158 29/5/05 Kankphul Habra-II NP 22.86 88.62 190 M �3.6 �2359 30/5/05 Bamangachi Barasat-I NP 22.75 88.51 158 M �5.0 �3260 30/5/05 Humaipur Barasat-II NP 22.68 88.47 159 M �4.5 �3061 31/5/05 Haroa Haroa NP 22.60 88.68 167 M �4.9 �3262 31/5/05 Arsula Baduria NP 22.74 88.77 148 M �5.1 �3363 1/6/05 Murarisha Hasnabad NP 22.59 88.84 362 >300 �3.6 �2064 1/6/05 Kalinagar Sandeshkhali-I NP 22.45 88.87 165 M �0.5 �15

(continued on next page)

Regional-scale stable isotopic signatures of recharge 155

Table 1 (continued)

Location Collection date Village/Location Administrativeblock/Town

Districta Latitude Longitude Depthb Aquiferc d18O (&) d2H (&)

RiverRivers1 28/5/04 Mayapur Bhagirathi-Hoogly ND 23.42 88.39 �5.7 �392 23/5/04 Behrampur town Bhagirathi-Hoogly MD 24.09 88.26 �5.7 �403 11/6/04 Belur town Bhagirathi-Hoogly HW 22.63 88.36 �5.6 �364 17/5/05 Amtala Jalangi ND 23.93 88.46 �2.9 �305 21/5/05 Sonatala Jalangi ND 23.52 88.51 �3.2 �246 28/5/05 Bongaon town Ichamati NP 23.05 88.83 �3.0 �217 31/5/05 Taki Ichamati NP 22.59 88.94 �0.4 �7

SeasonRains1 21/5/04 Behrampur town Pre-monsoon MD 24.09 88.26 4.4 242 25/5/04 Debagram Pre-monsoon ND 23.69 88.33 1.4 223 6/6/04 Barasat town Pre-monsoon NP 22.76 88.59 0.3 144 12/6/04 Calcutta Monsoon SP 22.52 88.34 �7.3 �495 15/6/04 Behrampur town Monsoon MD 24.09 88.26 �9.3 �646 19/6/04 Calcutta Monsoon SP 22.52 88.34 �8.1 �527 4/7/04 Behrampur town Monsoon MD 24.09 88.26 �6.7 �468 12/7/04 Calcutta Monsoon SP 22.52 88.34 �0.5 89 15/7/04 Shantipur town Monsoon ND 23.25 88.43 �6.4 �37

10 14/7/04 Barasat town Monsoon NP 22.76 88.59 �4.7 �1411 2/8/04 Calcutta Monsoon SP 22.52 88.34 �5.5 �4112 19/5/05 Betai Pre-monsoon ND 23.75 88.38 1.2 1913 1/7/05 Calcutta Monsoon SP 22.52 88.34 �4.9 �914 2/8/05 Calcutta Monsoon SP 22.52 88.34 �7.1 �52

a District name: MD: Murshidabad, ND: Nadia, NP: North 24 Parganas, SP: South 24 Parganas, HW: Howrah.b Depth of bottom of strainer in m below ground level.c Aquifer type: M: main, I: isolated, U: undefined, >300: greater than 300 m.

156 A. Mukherjee et al.

main aquifer samples. Most of the groundwater samples plotslightly below the LMWL (Fig. 4), and the linear fit for allgroundwater samples (d2H = 5.8 d18O � 3.0 [n = 64, r2 =0.82]) has a lower slope than both the LMWL and GMWL.The slope and intercept of this groundwater regression areless than those determined by Bhattacharya et al. (1985)(d2H = [6.8 ± 0.1] d18O + [2.2 ± 0.4]) and Krishnamurthy andBhattacharya (1991) (d2H = 6.4 d18O � 1.0 [n = 55, r2 =0.95]) for shallow groundwater along a transect from Cal-cutta to Delhi. Our relationship is likewise different fromthat of Aggarwal et al. (2000) for Bangladesh (d2H = 6.9d18O + 1.0 [n = 53, r2 = 0.91]). That regression was gener-ated by combining reported values for both shallow(<100 m depth) and deep (>100 m) samples. However, whenonly our main aquifer samples other than #64 are consid-ered, the resulting relationship (d2H = 6.7 d18O + 1.4[n = 53, r2 = 0.84]) is very similar to those of Bhattacharyaet al. (1985) and Aggarwal et al. (2000).

Discussion

Precipitation and recharge

Our groundwater d18O and d2H values fall within theranges of our monsoonal precipitation samples. Moreover,the mean d18O value of groundwater (�4.8&) is similar tothe monsoonal d18O mean for Yangon, Myanmar (�4.5&),which is the nearest comparable GNIP station (Araguas-Araguas et al., 1998). Consequently, deep groundwaterin our study area appears to have been recharged primar-ily by precipitation in a climatic regime like the present.

In contrast, Aggarwal et al. (2000) reported a broaderrange for the stable isotopic composition of groundwaterfrom depths >100 m in Bangladesh, although their valuesstill fell along the GMWL. They concluded that the stableisotopic signature of deeper water (at depths of 100–150 m and �300 m) is different (d18O values >�3& or<�6&) from that of the shallower aquifers, suggestingpaleorecharge during a different climatic regime between3 and 30 ka ago. It is possible that because of hydrostrati-graphic differences, the deeper samples of Aggarwalet al. (2000) are older than ours, but without radioisotopedates for our samples, we cannot evaluate this possibility.The slight deviation of our samples from the LMWL sug-gests that some evaporation occurs prior to or during infil-tration, or that recharge represents a mixture ofisotopically enriched and depleted waters. Enrichedsources of recharge could include pre-monsoon rains or,as inferred by Zheng et al. (2005), partly evaporated sur-face waters.

Evidence of the continental effect in groundwater

The d18O composition of groundwater varies significantlyboth with latitude (r2 = 0.31 and significance = 0.0000015for the analysis of variance F-statistic) and longitude(r2 = 0.28 and significance = 0.0000080). The trendline de-crease from ��4& to ��6& both south to north (Fig. 5)and east to west across the study area is consistent with re-charge from the southwest monsoon, which moves inlandnorthward and west-northwestward from the Bay of Bengal.A Rayleigh distillation effect results, whereby the isotopic

Figure 3 d18O–d2H bivariate plot of the rainwater samples collected during the present study. The linear regression fit for thesedata indicates the local meteoric water line (LMWL) for the study area. The global meteoric water line (GMWL) of Craig (1961) andthe expected isotopic composition of the Bay of Bengal (0, 0) are also plotted for reference.

Figure 4 d18O–d2H bivariate plot of deeper groundwater and river water from the study area. The groundwater samples have beengrouped by aquifer types (main and isolated; see Table 1). The LMWL represents the linear-fit line defined in Fig. 3. The dashed linerepresents the linear fit of all the groundwater samples. Location no. 64 represents the marine influence on the groundwater in thesouthern part of the study area. The expected isotopic composition of the Bay of Bengal (0, 0) is also plotted for reference.

Regional-scale stable isotopic signatures of recharge 157

158 A. Mukherjee et al.

composition of atmospheric moisture becomes progressivelydepleted with successive precipitation events (Dansgaard,1964). Krishnamurthy and Bhattacharya (1991) and Bhat-tacharya et al. (2003) documented similar subcontinent-scale depletion from the southwest monsoon. However,Krishnamurthy and Bhattacharya (1991) also concluded thatthe continental effect from Calcutta to Delhi cannot be so-lely explained by Rayleigh distillation, but must also ac-count for re-entry of enriched vapor into the air fromevapotranspiration (with an efficiency of �40%).

The latitudinal gradient of d18O depletion for our samples(��0.8& per 100 km) is higher than the reported inlandgradients of �2& per 1000 km by Krishnamurthy and Bhat-tacharya (1991) for Calcutta to Delhi and �6& per1000 km by Bhattacharya et al. (1985) for Ahmedabad (wes-tern India) to Delhi. A decrease in the inland gradient can becaused by significant recycling of water vapor within the ba-sin by evapotranspiration, as noted by Salati et al. (1979) forthe Amazon basin and by Krishnamurthy and Bhattacharya(1991). However, estimates for Calcutta (Mukherjee, 2006)indicate up to 50% evapotranspiration occurs during thesummer, suggesting sufficient vapor mass recycling. There-fore, the cause of the high gradient in our study area isunresolved.

River–groundwater interactions

In order to examine how major streams in the study areamay influence the isotopic composition of groundwater,we have plotted d18O versus latitude for groundwater sam-ples grouped by proximity to three rivers. These samplegroups are (a) Rivers Ganges–Bhagirathi-Hoogly (n = 32),(b) River Jalangi (n = 14), and (c) River Ichamati (n = 11).Some of the samples co-exist in both the Ganges and Jalangigroups. Distinct differences can be seen among the threegroups.

(a) Rivers Ganges–Bhagirathi-Hoogly (Fig. 6a): The com-position of the river water samples from the Bhagirathi-Hoo-gly is relatively uniform over a distance of �300 km,indicating a minimal latitudinal effect on the river watercomposition. The river water is generally more depleted

Figure 5 d18O of groundwater plotted versus latitude. The linear-fiBay of Bengal (0, 0), which is plotted for reference.

than groundwater, perhaps because of the highly depletedsnow melt from the Himalayas (source of the Ganges) (Ra-mesh and Sarin, 1992; Stuben et al., 2003). In contrast tothe river water, the groundwater isotope values trend posi-tively toward the south. In the northern to central portionof the study area (from Maldah district to central Nadia dis-trict), the river water and groundwater values are similar,indicating the possibility of recharge from the river, as in-ferred by Shivanna et al. (1999) for Murshidabad district.Some of the samples deviating from the Ganges–Bhagira-thi-Hoogly river water are also in proximity to the River Jal-angi, indicating the possibility of influence from both therivers. The maximum deviation from the linear trend in thisarea is Jumpukur (#24), which is suspected to have re-stricted recharge because of hydrostratigraphic heteroge-neity. In the south, the groundwater-river waterdissimilarity increases. The two most dissimilar groundwa-ter samples, from Ghasiara (#34) and Canning (#36), arethought to be affected by restricted recharge (because ofdepth) and seawater intrusion, respectively (Mukherjee,2006). The value from Calcutta (#4) is very similar to riverwater of Belur (RW #3), suggesting that recharge of ground-water in Calcutta is influenced by the Bhagirathi-Hoogly.This agrees with the prediction of a groundwater flow modeldeveloped for the area that high urban pumpage of ground-water induces inflow from the river (Mukherjee, 2006).

(b) Jalangi (Fig. 6b): The river is highly meandering (incontrast to the Ganges–Bhagirathi-Hoogly and the Ichamati)and covers relatively little latitudinal distance. The two iso-topic values of the river water are similar to each other, butthey are �2& more enriched than groundwater in thisgroup. The enrichment can be attributed to evaporationfrom the relatively narrow stream channel and runoff. Thegroundwater samples in this group do not show any latitudi-nal trend in d18O. Outliers include the isolated aquifer sam-ple from Iswarchandrapur (#45) and the restricted mainaquifer sample from Bhaluka (#49). Isotopic data thus sug-gest that the Jalangi has little influence on groundwater re-charge. Conversely, arsenic detections in river water andreaction-path modeling indicate groundwater discharge tothe Jalangi (Mukherjee, 2006).

t line does not include the expected isotopic composition of the

Figure 6 d18O composition of river water and groundwater plotted versus latitude. The groundwater data have been groupedbased on proximity to the rivers (a) Ganges–Bhagirathi-Hoogly, (b) Jalangi, and (c) Ichamati. The solid line represents the trend-lineobtained from the linear regression fit of the data. The approximate latitude and d18O value of the Bay of Bengal (21.4�N, 0) are alsoplotted for reference.

Regional-scale stable isotopic signatures of recharge 159

(c) Ichamati (Fig. 6c): Both groundwater and riverwater become enriched toward the south. The trendsare sub-parallel to each other, and the isotopic valuesof the river water are more enriched than the groundwa-ter. Enrichment of the Ichamati at Bongaon (RW #6) prob-ably reflects evaporation and runoff influence.Downstream at Taki (RW #7), where the channel is wider,

the enrichment is consistent with tidal inflow from theBay of Bengal. The maximum deviation from the ground-water trend is demonstrated by Kalinagar (#64), nearthe delta front. Solute concentrations for this sampleindicate seawater intrusion, which masks any evidenceof groundwater-river interactions in this area (Mukherjee,2006).

160 A. Mukherjee et al.

Possible effects of pumping on isotopicdistributions

Based on similarities between isotopic data from 1979 and1999 (the latter reported by Aggarwal et al. (2000)), Basuet al. (2002) concluded that extensive irrigational pumpingsince 1979 had not disturbed regional groundwater flow inthe Bengal basin. Those authors inferred that such pumpingwould have resulted in mixing of isotopically enriched irriga-tion return flow with more depleted groundwater. Subse-quent local studies in Bangladesh have shown differingresults. At Araihazar, 20 km east of Dhaka, Zheng et al.(2005) found that isotopic values for piezometer nests fellbetween the GMWL and ‘‘moderately evaporated surfacewaters’’; highly evaporated return flow did not appear tocontribute significantly to recharge. At Munshiganj, 30 kmsouth of Dhaka, Harvey et al. (2005) observed that d18O val-ues decreased from ��1.5& to ��4& between �5 and25 m depth, increased abruptly to ��1& at �30 m depth,then gradually decreased again to ��3& at �170 m depth.Using environmental tracers and flow modeling, Klumpet al. (2006) concluded that the chemocline at �30 m depthresulted from mixing induced by pumping.

Our groundwater isotopic data fall within the range ofthe data of Aggarwal et al. (2000) and, as noted above,linear fits to their data and ours are similar. However,numerical modeling by Mukherjee (2006) indicates thatpumping has caused widespread disruption of regionalgroundwater flow within our study area. Because we didnot collect samples from multiple depths at a single loca-tion (in contrast to Harvey et al., 2005; Zheng et al.,2005), and because the depth of the main aquifer tendsto vary with distance from the coast, it is difficult to as-sess isotopic variations with depth separately from latitu-dinal variations in our data set. Nonetheless, the relativelynarrow ranges of our isotopic values (excluding the seawa-ter-affected #64), along with the fact that our d18O valuesfor main aquifer samples fall within ranges reported forshallow groundwater in Murshidabad and South 24 Parg-anas districts (Shivanna et al., 1999; Stuben et al.,2003), suggest limited depth-dependent isotopic variabilityin the main aquifer. It is possible that pumping has verti-cally homogenized the isotopic composition of main aqui-fer water in our study area, and that lateral differencespersist because of the continental effect. In any case,conclusions regarding groundwater mixing cannot be solelybased on isotopic signatures.

Conclusions

Stable oxygen and hydrogen isotopic compositions of rain-water samples collected during this study show thatmonsoonal precipitation is generally more depleted thanpre-monsoon rainfall in Gangetic West Bengal, India. Theisotopic composition of deeper water from the main aqui-fer of the study area suggests recharge under current cli-matic conditions, with the prevailing southwest monsoonoriginating from the Bay of Bengal and moving to the north(toward the eastern Himalayas) and west-northwest(toward north-central India). Groundwater isotopes showevidence of a continental effect associated with precipita-

tion following a Rayleigh-type distillation, as well as slightevaporation during recharge. The continental depletiongradient determined in the present study is more thanany previous values reported from India. Comparison withprevious studies suggests that the isotopic compositions ofthe deeper and shallower parts of the main aquifer aresimilar, and inferences about groundwater mixing fromvarious depths cannot be based exclusively upon stableisotopic data at the regional scale. Moreover, the lack ofprominent differences between the isolated and mainaquifer samples makes it impossible to distinguishbetween the climatic regimes during which they were re-charged. The sample from the active delta front indicatesseawater intrusion, while those from locations near thenorth-central part of the study area and from Calcuttaindicate recharge influenced by the River Ganges and itsdistributary the River Bhagirathi-Hoogly. The snow-fedGanges–Bhagirathi-Hoogly river samples are distinctlymore depleted than those of the other major rivers ofthe area (Jalangi and Ichamati), which are affected byevaporation, runoff and tidal influences.

Acknowledgements

This project could not have been executed without thecooperation of the Public Health Engineering Directorate,Government of West Bengal. However, the ideas presentedin this paper are those of the authors and have not been offi-cially endorsed by the Government of West Bengal or anyother person or organization. Special thanks to P.K. De, A.Banerjee, A. Bhattacharya, and G. RoyChowdhury (PHED)for helping with selection of sampling sites. The authorsgratefully acknowledge the immense assistance providedby T. Datta and B. Hazra (PHED), Barindra Lal Mukherjeeand Murali Singh (Calcutta), Sudipta Rakshit (Plant and SoilScience, University of Kentucky), and Basanamoy Moitra(Shantipur) in sample collection; B.M. Engineering (Krish-naNagar) in supplying lithologs; Bob King (UK ERTL) andDr. Chris Eastoe (University of Arizona) with sample analy-ses; Dr. Helena Truszczynska (UK) for helping with statisti-cal analyses; and Dr. Steve Fisher (Kentucky GeologicalSurvey) and an anonymous reviewer for comments on themanuscript. The project was financially supported by theUniversity of Kentucky (Department of Geological Sciences,College of Arts & Sciences, and the Graduate School), theKentucky NSF-EPSCoR program, and the Geological Societyof America (Grant No. 7751-04). Last but not least, theauthors are sincerely grateful to the inhabitants of WestBengal for the hospitality and cooperation provided duringthe fieldwork.

References

Aggarwal, P.K., Basu, A.R., Poreda, R.J., Kulkarni, K.M., Froehlich,K., Tarafdar, S.A., Ali, M., Ahmed, N., Hussain, A., Rahman, M.,Ahmed, S.R., 2000. A report on isotope hydrology of groundwa-ter in Bangladesh: implications for characterization and mitiga-tion of arsenic in groundwater. International Atomic EnergyAgency-TC Project BGD/8/016, 64p.

Araguas-Araguas, L., Froehlich, K., Rozanski, K., 1998. Stableisotope composition of precipitation over Southeast Asia. Jour-nal of Geophysical Research 103, 721–742.

Regional-scale stable isotopic signatures of recharge 161

Basu, A.R., Jacobsen, S.B., Poreda, R.J., Dowling, C.B., Aggarwal,P.K., 2002. Response to Harvey, 2002: groundwater flow in theGanges delta. Science 296, 1563.

Bhattacharya, S.K., Gupta, S.K., Krishnamurthy, R.V., 1985. Oxygenand hydrogen isotope ratios in groundwaters and rivers fromIndia. Proceedings of Indian Academy of Science (Earth andPlanetary Science) 94, 283–295.

Bhattacharya, S.K., Froehlich, K., Aggarwal, P.K., Kulkarni, K.M.,2003. Isotopic variation in Indian monsoon precipitation: recordsfrom Bombay and New Delhi. Geophysical Research Letters 30,2285–2288.

Coleman, J.M., 1981. Deltas: Processes of Deposition and Models ofExploration, second ed. Burgess, Minneapolis, MN.

Craig, H., 1961. Isotopic variations in meteoric water. Science 133,1702–1703.

Dansgaard, W., 1964. Stable isotopes in precipitation. Tellus 16,436–467.

GHCN, 2005. The Global Historical Climatology Network version 2.<http://www.ncdc.noaa.gov/oa/climate/research/ghcn/ghcn.html> (accessed December 2005).

Harvey, C.F., Swartz, C.H., Badruzzaman, A.B.M., Keon-Blute, N.,Yu, W., Ali, M.A., Jay, J., Beckie, R., Niedan, V., Brabander, D.,Oates, P.M., Ashfaque, K.N., Islam, S., Hemond, H.F., Ahmed,M.F., 2005. Groundwater arsenic contamination on the GangesDelta: biogeochemistry, hydrology, human perturbations, andhuman suffering on a large scale. Comptes Rendus Geosciences337 (1–2), 285–296.

Klump, S., Kipfer, R., Olaf, A.C., Harvey, C.F., Brennwald, M.S.,Khandkar, N.A., Badruzzaman, A.B.M., Hug, S., Imboden, D.M.,2006. Groundwater dynamics and arsenic mobilization in Ban-gladesh assessed using noble gases and tritium. EnvironmentalScience & Technology 40, 243–250.

Krishnamurthy, R.V., Bhattacharya, S.K., 1991. Stable oxygen andhydrogen isotope ratios in shallow groundwater from India and astudy of the role of evapotranspiration in the Indian monsoon.

In: Taylor, H.P., O’Neil, J.R., Kaplan, I.R. (Eds.), Stable IsotopeGeochemistry. A Tribute to Samuel Epstein, Special PublicationNo. 3. The Geochemical Society, San Antonio, TX, pp. 187–194.

Mukherjee, A., 2006. Deeper groundwater flow and chemistry in thearsenic affected western Bengal basin, West Bengal, India. Ph.D.thesis. University of Kentucky, Lexington, USA, 248p.

Ramesh, R., Sarin, M.M., 1992. Stable isotope study of the Ganga(Ganges) river system. Journal of Hydrology 139 (1–4), 49–62.

Salati, E., Oall’Olio, A., Matsui, E., Gat, J.R., 1979. Recycling ofwater in the Amazon basin, an isotope study. Water ResourcesResearch 15, 1250–1258.

Shivanna, K., Sharma, S., Sinha, U.K., Nair, A.R., Navada, S.V., Ray,A., Talukdar, T., Mehta, B.C., Ghosh, A.K., 1999. Arsenicpollution in ground water of West Bengal. In: Proceedings ofWorkshop on Groundwater Pollution and Its Protection withSpecial Reference to Arsenic Contamination. Central GroundWater Board, Calcutta.

Smith, A.H., Lingas, E.O., Rahman, M., 2000. Contamination ofdrinking-water by arsenic in Bangladesh: a public health emer-gency. Bulletin of the World Health Organization 78 (9), 1093–1103.

Stuben, D., Berner, Z., Chandrashekharam, D., Karmakar, J., 2003.Arsenic enrichment in groundwater of West Bengal, India:geochemical evidence for mobilization of As under reducingconditions. Applied Geochemistry 18 (9), 1417–1434.

Wood, W.W., 1981. Guidelines for collection and field analysis ofgroundwater samples for selected unstable constituents. Tech-niques of Water-Resources Investigations of the United StatesGeological Survey, Book 1, Chapter D2, 11 p.

Zheng, Y., Datta, S., Stute, M., Dhar, R., Hoque, M.A., Rahman,M.W., Ahmed, K.M., Schlosser, P., van Geen, A., 2005. Stableisotopes (18O, 2H) and arsenic distribution in the shallowaquifers in Araihazar, Bangladesh. Eos Transactions AmericanGeophysical Union, 86 (52), Fall Meeting Supplement, AbstractH31B-1305.