Analysis of arsenic-contaminated groundwater domain in the...

Hydrological Sciences-Journal-des Sciences Hydrologiques, 47(S) August 2002 S55 Special Issue: Towards Integrated Water Resources Management for Sustainable Development

Analysis of arsenic-contaminated groundwater domain in the Nadia district of West Bengal (India)

P. K. MAJUMDAR, N. C. GHOSH & B. CHAKRAVORTY National Institute of Hydrology, Jalvigyan Bhawan, Roorkee 247667, India ncgftfnih.ernet.in

Abstract An analysis of groundwater flow and transport processes of arsenic in the flow domain of the Yamuna sub-basin located in West Bengal (India) is presented. The objectives of the analysis are: to conceptualize the groundwater flow domain, to determine groundwater flow paths and groundwater velocities, to study arsenic transport in the flow domain, and to study the well captures zones. The first three objectives and issues are addressed by simulation of steady and transient groundwater flow and contaminant transport using the US Geological Survey three-dimensional finite difference code, MODFLOW, and the three-dimensional advective-dispersive transport code, MT3D. Simulated results of the calibrated model replicate the observed monthly water table conditions perfectly. Contaminant transport analysis indicates an in situ arsenic source. Using the particle tracking algorithm, MODPATH, the possibility of arsenic removal from a sample key location, and the design of wells for withdrawing arsenic-free groundwater are studied through analysis of the well capture zones.

Key words arsenic contamination; groundwater; contaminant transport; capture zone; model sensitivity; calibration; validation; Yamuna basin; Ganga-Brahmaputra Delta, West Bengal

Analyse du domaine hydrogéologique contaminé par l'arsenic dans le district de Nadia, au Bengale Ouest, en Inde Résumé Nous présentons une analyse des écoulements souterrains et des processus de transport d'arsenic dans le sous-bassin de Yamuna, dans le Bengale Ouest (Inde). Les objectifs de l'analyse sont de conceptualiser le domaine d'écoulement hydrogéologique, de déterminer les chemins et les vitesses d'écoulement souterrain, d'étudier le transport d'arsenic dans le domaine d'écoulement et d'étudier les zones de prélèvement des puits. Les trois premiers objectifs et questions sont abordés par simulation des écoulements souterrains permanents et transitoires et du transport de contaminants avec le code de calcul tri-dimensionnel à différences finies du service américain USOS, MODFLOW, et le code de calcul tri-dimensionnel de transport par advection-dispersion, MT3D. Les résultats simulés du modèle calé reproduit parfaitement les conditions mensuelles observées de la surface libre. L'analyse du transport du contaminant indique une source in-situ d'arsenic. Les possibilités d'éliminer l'arsenic en un point clé et de construire des puits prélevant une eau sans arsenic sont étudiées à travers l'analyse des zones d'influence des puits et grâce à l'algorithme de suivi particulate MODPATH.

Mots clefs contamination par l'arsenic; eau souterraine; transport de contaminant; zone de prélèvement; sensibilité de modèle; calage; validation; basin Yamuna; delta du Gange-Brahmapoutre, Bengale Ouest

INTRODUCTION

Epidemiological studies conducted in 1980 found contamination of the groundwater by arsenic above the permissible limit of 0.05 mg l"1. This was first detected in some localized pockets in a few district of West Bengal (India). Later, during 1983-1995,

Open for discussion until I February 2003

S56 P. K. Majumdar et al.

serious arsenic contamination was reported throughout a linear track of 470 km along the River Bhagirathi in the Ganga-Brahmaputra Delta covering an area of about 37 500 km2 and including 830 villages in eight districts, namely Malda, Murshidabad, Nadia, South and North 24-Paraganas, part of Calcutta (on the left bank of the river), Burdaman, and Hugh (on the right bank of the river) (Ghosh & Chakravorty, 1997). There were several opinions, but no definite conclusions about the cause of such large-scale presence of arsenic in the groundwater domain were available. The spread of arsenic in the groundwater is thought to be due to the hydrology and hydrogeological behaviour of the groundwater domain. Since the spread of arsenic had increased over time, doubts were raised as to whether the occurrence of arsenic in new pockets of the groundwater domain impacts on the transport of contaminants from other pockets. To quantify the migration pathways of arsenic contaminated groundwater and the transport phenomena, the groundwater domain of Yamuna sub-basin, covering the areas of North 24-Parganas and Nadia districts, in West Bengal were chosen for the modelling study.

To understand the transport processes affecting the areal spread of arsenic-contaminated water, the flow in the study area was modelled numerically using the US Geological Survey three-dimensional (3-D) finite difference code, MODFLOW, the three-dimensional advective-dispersive transport code, MT3D, for contaminant transport and the particle tracking algorithm, MODPATH, for capture zone analysis (Waterloo Hydrologie Inc., 1997). Key aspects of the modelling include: (a) a 3-D finite difference mesh allowing simulation of rivers and drains; (b) representation of vertical and horizontal stratigraphie formations; (c) steady-state groundwater conditions and calibration of model parameters; (d) simulation of transient-state validation of the model; (e) simulation of contaminant transport with various sources of arsenic; and (f) analysis of well captures zones.

STUDY AREA

The study area (Fig. 1) is a part of the upper Ganga-Brahmaputra Delta having a near surface succession of Quaternary sediments of varying thickness. Abandoned meander scrolls of various wavelengths and amplitudes, chute cutoffs, conspicuous levees along the Bhagirathi and other rivers, as well as back swamp lakes in between inter-distributary levees are the geomorphic formation which could easily be related to fluvial processes of sediment distribution (Acharya et al., 1999). The upper delta plain has a series of meander belts. In the central part of the delta an older, composite belt has been recognized and the study area is on its western side. These meander belts are not continuous stretches, but are preserved as festoons with the successive belts eroding and in setting into the older ones, so that most become consolidated through expulsion of pore water. Mostly sand, silt and clay deposits are underlain by gravel beds showing several fining upward cycles of various thicknesses. These sediments rest directly over the extensive clay beds of the tertiary period at a subsurface depth of about 150 m. The change in lithology in the horizontal plane is quite abrupt.

Rainfall data for two raingauge stations, located at Bongaon and Habra, in Ashoknagar, show the occurrence of effective precipitation results from the southwest (SW) monsoon during the months of May to October with peak rainfall occurring

Analysis of arsenic-contaminated groundwater S57

IN

4 wye "\ ~ip\

'2T> % ' •

SALT IvAKE ( C A L C B T l i )

B A N G L A D E

H

{(

54 "£• ~V

tzzxx S t u d v a r e a

Fig. 1 The study area located in the upper Ganga-Brahmaputra Delta.

typically during July and August. Average annual rainfall at Bongaon and Habra during the period 1986-1995 was 1850 and 1900 mm, respectively. Yearly average potential évapotranspiration is estimated to be of the order of 1389 mm.

The study area is bounded by the Bhagirathi River on the western side and the Ichhamati River to the east. The Yamuna River, which flows from northwest to southeast, remains mostly in a ponded condition. It has rare flow, but tidal effects cannot be ruled out. In addition to rainfall, irrigation return flows from the agricultural land are a source of groundwater recharge.

Groundwater occurs under unconfmed to semi-confined conditions. In some areas there are overlying top clay and sandy clay deposits that are as thick as 20-30 m. Water table fluctuations and lithologs do not indicate the presence of multiple or distinct aquifer systems up to a depth of 150 m from the normal ground level. However, based on the water extraction patterns, the alluvium deposits may be distinguished vertically as shallow, intermediate and deep aquifers. Water table fluctuation is between 1 and 2 m in the eastern part of the study area and between 2 and 4 m in the western part. The hydraulic gradient is steep (1:333) in the northern part and gentle (1:1666) in the central and southern parts. The gradient is in general southerly to southeasterly. Transmissivity values reported from pumping tests range from 671 to 4981 m2 day"1 and storativity values are from 1.3 x 10~2 to 6.2 x 10"4 (NIH & CGWB, 1999). Water yield from shallow tube wells (18-54 m below ground level) ranges from 23 to 43 m3 h"1, while for heavy-duty, deep (150 m b.g.l.) tube wells it ranges from 45 to 200 m3 h"1. The usual trend of groundwater table fluctuations in a calendar year is a declining trend from January to April, a rising trend from May to


July, and then declining again from August onwards. Moreover, analysis of monthly water level fluctuations for 62 wells within the study area during the year 1997/98 shows that the water levels in most of these wells follow an identical trend between September and April.

Arsenic in the study area

The extent and concentrations of arsenic was first reported in the study area in 1985 (Chakraborty, 1995). It has been observed that, over the years, levels and concentrations of arsenic have increased. Abstracted water is found to contain more arsenic wherever transmissivity decreases and storativity increases (Fig. 2). Arsenic concentrations observed during pumping tests show fluctuations in some of the wells. This indicates that there is a possibility of a rise in arsenic concentration due to local dissolution of precipitated arsenic compound in the zone of influence of the pumping well. Natural systems like rivers, drains and other water bodies, which are interacting with the aquifer, have low arsenic concentrations. Areal discontinuity of the arsenic plumes in the flow domain has also been noticed. This may be due to the existence of clay patches. In general, arsenic concentration in the region appears to correlate with the local lithology (Mallick & Rajagopal, 1996).

A three-dimensional representation of recent scenario of arsenic concentration in groundwater in 1997 (Fig. 3) shows that the aquifer has maximum arsenic concentrations near Kundalia, Mandalhat, Khusbasi, Devipur, Ashok Nagar and Maslandpur; low concentrations are found around Nagarukra, Napitsatbaria, Balidanga and Digra. Clay patches influence the concentration plumes invariably, even if its influence on the flow domain is not visible in some of those locations of high concentration. Vadose zone sampling shows that, in general, arsenic is retained by the surface soil and the water underneath is found potable. In the absence of any history of

260o 0 0010

# <y

1

o o CO C*5

- ARSENIC CONCENTRA 1 ION (tng/1)

TRANSMISSIVITY (m2 /day)

' % • ° ^

X X ^ 2

°-040

% 0.00^ %

0?° S

y?

%

' •STORATIVITY (m/m)

' i *

1 o ^

9 B »

«a-0.002S

o Je0o%

500»-51)00 10000 15000 20000 "'5000 30000 Î5000 40000 45000

Fig. 2 Superimposed plots of storativity, transmissivity and arsenic concentration.


%

°10

^

KHILSBASI KllNDAIJA

KHNDAIHU ^

•V-'

-̂ V- *'

I \

• - • " "

-e>

DEVIPUR

".'' MASLAWIR

"•" . , ASMHi NAGAR " .

* _ /'

^ X •- . .

.'" ^ ^

N , ^

• 0 2 2

"20

,18

• ™ 14 12

,10

CO

CO

C

D

CD

Uo.05

l o , 0 4

H e 02

•o.OO

Fig. 3 Three-dimensional plot of arsenic concentration in the flow domain.

water quality studies in the study area, it is difficult to draw conclusions about the cause and source of arsenic in the groundwater. Monthly arsenic concentrations monitored at different key wells of the region during 1997 also failed to reveal any specific trend. There were significant fluctuations observed in the arsenic concentration values during the year. Wells having higher concentration of arsenic during the pre-monsoon period were observed to have arsenic concentrations below detection level and vice versa during the post-monsoon period. In arsenic affected zones, the presence of iron pyrites was detected. Oxidation of arsenopyrite has been reported as one of the reasons for activation of arsenic in the groundwater (Bhattacharya et al, 1997). The validity of such a conclusion is yet to be ascertained in the study area, as the affected aquifers lie between 13 and 18 m b.g.l., and water table fluctuations have not shown such a fall in the seasonal water level over the last 10 years to generate conditions for oxidation and reduction processes.

The likelihood of chemical reaction in the aqueous phase (Welch et al., 1988) is low, as the pH is more or less stable areally inside the aquifer domain. There is always a possibility of solid-liquid phase interaction through adsorption and precipitation of contaminants on clay lenses (Carrillo & Drever, 1998) and subsequent dissolution due to pumping. The areal distribution of arsenic concentration ranges is summarized in Table 1.

CONCEPTUAL MODEL

The conceptual model is postulated primarily to suit the groundwater quality scenario. Arsenic contamination is detected more in the intermediate zone (15-40 m), less in the shallow zone (above 15 m) and hardly at all in the deeper abstraction


Table 1 Range and percentage area affected by arsenic in the study domain.

Arsenic concentration range (mg 1"') Percentage area 0.011-0.051 50% 0.051-0.091 10% 0.091-0.131 20% 0.131-0.172 10% 0.172-0.212 00% 0.212-0.252 10%

Table 2 Zones of transmissivity and storativity used as initial model parameters.

Zone

A B C D E F

Transmissivity range (nr day"1) 538-1000 1001-1500 1501-2000 2001-2500 2501-3000 3001-4001

Average value

769 1250 1750 2250 2750 3250

Percentage area covered 14.3% 4.8% 28.6% 23.8% 14.3% 14.3%

Storativity range

3.30 x 10"6-2.20xl0"4

2.20xl0"4-4.00x 10"4

4.00 x 104-1.09x 10"3

1.09 x 10"3-1.30xl0"3

1.30 x l0~'-3.30x 10"-' 3.30 x 10~M.30x 10~2

Average value

1.12 x 104

3.10 x 10~4

7.45 x 10"4

1.19 x 10° 2.30 x 10° 8.15 x 10"-'

Percentage area covered 12.5% 12.5% 18.8% 18.8% 18.8% 18.8%

zones (40-80 m). The regional groundwater flow system is unilateral with deposition of clay minerals in abundance up to 150 m of depth and the geological formation consists mainly of fluvial deposits. The flow domain was conceptualized as a set of uniformly thick layers of sediment sheets, deposited one above another with identical slope, to form a three-dimensional grid comprising 100 blocks horizontally and vertically. Each block measures 500 x 300 m. The total depth was considered as 80 m and was divided into 10 layers, each parallel to the surface layer, to accommodate horizontal and vertical anisotropy and variation of local lithologs. Ground level data at 62 locations were kriged to the centre of gravity of 10 000 blocks with the help of the geostatistical package, GEOPACK, for generating ground levels at each block in the flow domain. For the sake of simplicity and to avoid negative values, all the reduced levels below mean sea level were transformed to a hypothetical reference plane by adding a value of 100 m.

The overall hydraulic properties were enumerated on the basis of pumping test results in different locations. The grid values of transmissivity and storativity were generated through SURFER and divided horizontally into six zones (A-F), which were used as initial hydraulic parameters for the calibration of the conceptual model. These zones are given in Table 2. All ten layers were assumed to have similar non-homogeneity in the horizontal direction. It was also assumed that the porosity value for all the zones is 0.38.

Recharge through rainfall and irrigation return flow were taken as 20% of the annual rainfall and 20% of the net draft, respectively, for each block. Withdrawal from the aquifer and évapotranspiration were considered as the discharge components. The évapotranspiration value was taken as 1389 mm year"1 from a root zone depth of 2.0 m, while withdrawals from the aquifer were assessed from the projection of block-wise groundwater withdrawal data of 1991.


In the absence of complete natural hydrological boundary conditions, calibration of the model parameters was carried out through step-by-step sensitivity analysis for a hydrostatic condition by trial-and-error method. Available river/drain hydraulic characteristics for April 1997 were considered as the initial boundary inputs. The calibrated model was validated for additional transient boundary conditions such as recharge from the top and heads at the boundaries. The water levels of April 1997 were considered for processing the initial conditions for the transient validation runs. The validated model was used for simulation of monthly water level conditions for the period October 1997-March 1998 using the water levels of September 1997 (Fig. 4) as the initial piezometric heads.

The following boundary conditions were used for the transient run: ~ River stages in Bhagirathi, Ichhamati and Jamuna rivers. - Head values at the northern and southern boundaries based on monthly observation

of well water levels. - Net recharge/discharge component based on the rainfall, évapotranspiration and

pumpage data for the respective months of the simulation period. The arsenic pollution problem was conceptualized considering the following

probable sources: (a) the Bhagirathi River, (b) the head at the northern side, and (c) point sources at localized pockets. No sink term was introduced. The initial parameters considered in MT3D runs were as follows: - longitudinal dispersivity: 150 m; - transverse horizontal dispersivity: 10% of longitudinal dispersivity; - transverse vertical dispersivity: 1 % of longitudinal dispersivity; and - molecular diffusion: 1.0 x 10"6 n r s"'. Arsenic concentrations measured during September 1997 were taken as the initial concentrations.

The calibrated and validated flow model was applied to examine the possibility of cleaning arsenic pollutants from the aquifer domain and designing wells for pumping

^ . / / , 0 10000 20000 30000 40000 50000

Fig. 4 Initial water table considered for transient simulation (September 1997).


of arsenic-free groundwater. These remedial measures were tested for possible improvement of the concentration scenario of September 1997. Analysis of capture zones was carried out using backward track of 10 particles in a circle of radius 1000 m to MODPATH. The capture zone is the region within which all flow lines converge to the extraction well. Water within the capture zone eventually finds its way to the well, from where it is pumped out of the aquifer. Flow lines outside of the capture zone may be bent toward the well, but the regional flow associated with the natural hydraulic gradient is strong enough to carry that groundwater past the well. The width of the capture zone is directly proportional to the pumping rate, Q, and inversely proportional to the product of the regional Darcy flow velocity and the aquifer thickness. Pumping rates of clean-up wells are considered as 250, 500 and 1000 m3 day-1, while for the design wells the values of 100, 200 and 500 nr' day"1 were used.

RESULTS AND DISCUSSION

To start with, the model was calibrated assuming hydrostatic conditions of the flow domain using April 1997 as the base. The initial water table conditions were assigned at the highest altitude of the study domain. The river and drainage systems were considered for calibration of the model parameters. The calibrated model parameters are given in Table 3.

The calibrated heads for layer 1 are shown in Fig. 5. It can be observed that the computed water table in the northern half of the study area is governed by the river/drain and the topography, as the aquifer is in an unconfmed to semiconfmed state. The results match well with the observations. In the southern half of the study area, the rivers and drains appear to be dominating the calibrated scenario, which differs from the observed trends. Additional natural and manmade disturbances appear to influence the water table condition of this part of the study area. These are taken care of in the subsequent validation runs, so that an agreement between the computed and observed static water table is reached. This is the case for all other layers due to unilateral conditions prevailing in the flow system. The validated run showed a perfect matching with the observed static water table conditions.

The transient simulations for the period from October 1997 to March 1998 show a good match to the measured values. Error maps developed for each month reflect scattered contours of errors. A comparison of the observed and computed water levels for October 1997 is shown as a sample result of the simulation (Fig. 6). Results for other stress periods have a similar degree of fitness between the computed and observed water levels.

Table 3 Calibrated hydraulic conductivity values.

Zone Kx or Kv (m day"1) Kz (m day"1)

_ 4.83840 0.48384 B 8.29440 0.82944 C 11.6640 1.16640 D 21.6864 2.16864 E 18.3168 1.83168 F 15.0336 1.50336


25000

15000

\0&

% 108 109

10$ 105

108

%

104

2? $3

o

loooo

o 0 10000 20000 30000 40000 5000

Fig. 5 Model outputs corresponding to the response of calibrated model parameters.

25000i

5000-

105.5

A0*-(

N # > - 6

Joe.0

\ . 105,5 " 1965)0 104fy

106.0 N^W-

* • $

'a*,

103.5 103.5

-\0A 8*5 *

C5

<&•

* # 3 ,

\0 f c '

o o •t»C>3

' o b i

O ( o 103.0

<o

W CM

<g> o

M $̂

5000 10000 15000 20000

Fig. 6 Comparison of observed I 25000 30000 35000 40000 45000 5000'

—) and computed ( ) water levels (October 1997).

The results of the transient transport runs, considering Bhagirathi River and the northern boundary as a source of arsenic, showed negligible contaminant transport for five years of transient transport runs. Study of the transport of contaminants considering in situ sources at six localized points (Fig. 7) resulted in a trend of spreading that was similar to the observed spatial distribution although with different magnitudes. In situ sources are assumed to be point loads of arsenic to the flow regime and their input patterns are conceptualized as:


- propagation of arsenic-bearing material by the oxidation process and then entry of arsenic to the flow domain through the rainfall recharge during monsoon months (May-September), and

- no input of arsenic from the source during non-monsoon months. This artificial input is only to incorporate the effect of the oxidation-reduction

process actually occurring in the subsurface. From Fig. 7, it may be seen that the occurrence of arsenic at one location has no bearing on the transport of arsenic from other pockets, but it is due to the in situ activation. The zone of influence of localized sources for the given input condition is found to be of the order of 2 km from the point of release. The temporal variation of arsenic concentrations was analysed for five years of simulation at the source points. The results of this analysis suggest that, for the same concentration of arsenic during the recharge period, the concentration during the deactivation period increases in the subsequent years, whereas away from the source the concentration increases over time.

30000r

Ebisbasi

E'eYlTH

Ifcaliiiip^ kùir]

Source Spreading after 1 ye-j 23 spreading after 5 years Fig. 7 Computed extent of spreading of point sources at the end of 5 years.

A sample result of capture zone analysis for a specified pumping rate is shown in Fig. 8. Reclamation of arsenic through clean-up wells is not effective even with a set of three wells, as there is no capture zone which can take out the arsenic plume without disturbing the clean water zones. An increase in the pumping rate results in flow lines becoming localized. However, particle-tracking scenarios could be useful for designing clean water wells. Well locations could be ascertained such that a higher rate of safe water pumping is possible. For example, it is evident from analysis that if a specific, higher rate of pumping is assumed in all wells, wells CZ1, CZ2, CZ4 and CZ5 are drawing water from arsenic contaminated zones, whereas well no. CZ3 still draws clean water. All these locations are, however, selected hypothetically, but the technique could be used for deciding locations of wells which would draft arsenic-free groundwater.


30000.00,

25000.00,

20000.00

15000 .Off

0.00

Cz4

* C z 3 0<b 10000.00 y '

o

5000.00„ <=> ' b •a -""

Cz1

<?. Cz2

Cz5

0 0 0 . 6 0 7™10000.00 um° 2000rjorJ0 ' "fÔOOO.OO 35OT 400004Ôff ^SÔto.OO

Fig. 8 Particle tracking with pumping rates for a clean-up well (750 m' day"') and a design well (400 m' day"').

CONCLUSIONS

Arsenic-contaminated groundwater in West Bengal has been analysed. A three-dimensional finite difference mesh comprising ten layers, each having 10 000 blocks, was conceptualized. The model was calibrated, validated and simulated using MODFLOW and coupled with MT3D for contaminant transport and MODPATH for capture zone analysis. The flow model identified the field flow paths and velocities with the expected degree of precision. Simulations of a number of alternatives of source conceptualization established that arsenic contamination is the result of localized in situ sources. The study indicates that the occurrence of arsenic at a location has no correlation with the transport of arsenic from other sources; rather it indicates that there is spreading at a localized scale due to transport of in situ activation. Analysis of the geochemical behaviour of the localized pockets is recommended to assess the influence of the reactive component. This would also help in ascertaining the effects of clay lenses on the transport of arsenic. The model has the capability of accommodating such effects in the future. Capture zone analysis does not show encouraging results for removing arsenic through a specific set of clean-up wells. However, capture zone analysis may be applied for fixing the location of wells and their pumping rate for drawing arsenic-free groundwater.

Acknowledgement The authors duly acknowledge the data support given the by Central Ground Water Board, Calcutta Region, India.

S66 P. K. Majumdar et al

REFERENCES

Acharya, S. K., Chakraborly, P., Lahiri, S. & Mukherjee, P. K. (1999) Ganga-Bhagirathi Delta Development and Arsenic Toxicity in Groundwater, Silver Jubilee Celebrations Commemoration volume. (Int. Conf. on Man and Environment, Calcutta, 25-27 November 1999), 178-192. Centre for Study of Man and Environment, Calcutta, India.

Bhattacharya, P., Chatterjee, D. & Jacks, G. (1997) Occurrence of arsenic contaminated groundwater in alluvial aquifers. In: Safeguarding Groundwater from Arseniferous Aquifers (22nd WEDC Conference, New Delhi, India), 220-223.

Carrillo, A. & Drever, J. I. (1998) Adsorption of arsenic by natural aquifer material in the San Antonio-El Triunfo Mining Area, Baja California, Mexico. Environ. Geo!. 35(4), 251-257.

Chakraborly, D. (1995) Arsenic in groundwater in seven districts of West Bengal. In: Proc. National meeting on S&T Issues for Water Resources Management (11-12 September 1995, Calcutta), 145-152.

Ghosh, N. C. & Chakravorty, B. (1997) Arsenic pollution in groundwater—a status report. Report CS(AR) 15/96-97, National Institute of Hydrology, Roorkee, India.

Mallick, S. & Rajagopal, N. R. (1996) Scientific correspondence: Groundwater development in arsenic-affected alluvial belt of West Bengal—some questions. Current Sci. 70( 11 ), 956-958.

Nil I & CGWB (National Institute of Hydrology and Central Ground Water Board) (1999) Report on arsenic pollution study in Jamuna sub-basin, Nadia and 24 Paraganas districts, West Bengal.

Waterloo Hydrologie Inc. ( 1997) Visual MODI-LOW User Manuals. US Govt Printing Office, Washington DC, USA. Welch, A. H., Lico, M. S. & Hughes, J. L. (1988) Arsenic in groundwater of western United States. Ground Water 26(3),

333-347.

Analysis of arsenic-contaminated groundwater domain in the...

Documents

Transcript of Analysis of arsenic-contaminated groundwater domain in the...