Arsenic in groundwater of the Bengal Basin, Bangladesh: Distribution, field relations, and...

Arsenic in groundwater of the Bengal Basin, Bangladesh:Distribution, field relations, and hydrogeological setting

Peter Ravenscroft · William G. Burgess ·Kazi Matin Ahmed · Melanie Burren · Jerome Perrin

Abstract Arsenic contaminates groundwater across muchof southern, central and eastern Bangladesh. Groundwaterfrom the Holocene alluvium of the Ganges, Brahmaputraand Meghna Rivers locally exceeds 200 times the WorldHealth Organisation (WHO) guideline value for drinkingwater of 10 �g/l of arsenic. Approximately 25% of wells inBangladesh exceed the national standard of 50 �g/l,affecting at least 25 million people. Arsenic has enteredthe groundwater by reductive dissolution of ferric oxyhy-droxides, to which arsenic was adsorbed during fluvialtransport. Depth profiles of arsenic in pumped ground-water, porewater, and aquifer sediments show consistenttrends. Elevated concentrations are associated with fine-sands and organic-rich sediments. Concentrations are lownear the water table, rise to a maximum typically 20–40 mbelow ground, and fall to very low levels between about100 and 200 m. Arsenic occurs mainly in groundwater ofthe valley-fill sequence deposited during the Holocenemarine transgression. Groundwater from Pleistocene andolder aquifers is largely free of arsenic. Arsenic concen-trations in many shallow hand-tube wells are likely to

increase over a period of years, and regular monitoringwill be essential. Aquifers at more than 200 m below thefloodplains offer good prospects for long-term arsenic-freewater supplies, but may be limited by the threats of salineintrusion and downward leakage of arsenic.

R�sum� L’arsenic contamine les eaux souterraines dansla plus grande partie du sud, du centre et de l’est duBangladesh. Les eaux des nappes alluviales holoc�nes duGange, du Brahmapoutre et de la Meghna d�passentlocalement 200 fois la valeur guide donn�e par l’OMSpour l’eau de boisson, fix�e � 10 �g/l d’arsenic. Environ25% des puits du Bangladesh d�passent la valeur standardnationale de 50 �g/l, affectant au moins 25 millions depersonnes. L’arsenic a �t� introduit dans les nappes par ladissolution par r�duction d’oxy-hydroxydes ferriques surlesquels l’arsenic �tait adsorb� au cours du transportfluvial. Des profils verticaux d’arsenic dans l’eau souter-raine pomp�e, dans l’eau porale et dans les s�diments desaquif�res montrent des tendances convergentes. Lesconcentrations �lev�es sont associ�es � des s�diments �sable fin et riches en mati�res organiques. Les concen-trations sont faibles au voisinage de la surface de lanappe, atteignent un maximum typiquement entre 20 et 40m sous le sol, puis tombent � des niveaux tr�s bas entre100 et 200 m. L’arsenic est surtout pr�sent dans les eauxsouterraines de la s�quence de remplissage de vall�ed�pos�e au cours de la transgression marine holoc�ne.Les eaux souterraines des aquif�res pl�istoc�nes et plusanciens sont tr�s largement d�pourvus d’arsenic. Lesconcentrations en arsenic dans de nombreux puits creus�s� la main doivent probablement augmenter au cours desprochaines ann�es ; aussi un suivi r�gulier est essentiel.Les aquif�res � plus de 200 m sous les plaines alluvialesoffrent de bonnes perspectives pour des alimentations eneau sans arsenic � long terme, mais ils peuvent Þtrelimit�s par les risques d’intrusion saline et la drainancedescendante de l’arsenic.

Resumen El ars�nico ha contaminado gran parte de lasaguas subterr�neas en el Sur, centro y Este de BanglaDesh. Su concentraci�n en las aguas subterr�neas delaluvial Holoceno de los r�os Ganges, Brahmaputra yMeghna supera localmente en un factor 200 el valor gu�adel ars�nico en el agua potable, establecido por laOrganizaci�n Mundial de la Salud (OMS) en 10 �g/L.

Received: 6 January 2003 / Accepted: 18 November 2003Published online: 9 March 2004

� Springer-Verlag 2004

P. Ravenscroft ())Arcadis Geraghty and Miller International,2 Craven Court, Newmarket, Suffolk, CB8 7FA, UKe-mail: [email protected].: +44-1638-674786Fax: +44-1638-668191

W. G. Burgess · M. Burren · J. PerrinDepartment of Earth Sciences,University College London,Gower Street, London, WC1E 6BT, UK

K. M. AhmedDepartment of Geology, Dhaka University,Dhaka, Bangladesh

Present address:M. Burren, 39 Durrants Lane, Berkhamstead, Herts, HP4 3PL, UK

Present address:J. Perrin, Centre of Hydrogeology,Neuchatel University,11 Rue E-Argand 2000, Neuchatel, Switzerland

Hydrogeology Journal (2005) 13:727–751 DOI 10.1007/s10040-003-0314-0

Aproximadamente, el 25% de los pozos de Bangla Deshsuperan el est�ndar nacional de 50 �g/L, afectando almenos a 25 millones de personas. El ars�nico ha llegado alas aguas subterr�neas por la disoluci�n reductora dehidr�xidos f�rricos a los que se adsorbe durante el trans-porte fluvial. Los perfiles del ars�nico en las aguas sub-terr�neas bombeadas, agua de poro y sedimentos delacu�fero muestran tendencias coherentes. Las concentra-ciones elevadas est�n asociadas a arenas finas y sedi-mentos ricos en materia org�nica. Las concentraciones dears�nico son bajas cerca del nivel fre�tico, se incrementanhasta un m�ximo que se localiza generalmente a entre 20y 40 m bajo la cota del terreno, y disminuyen a valoresmuy pequeos entre alrededor de 100 y 200 m. El ar-s�nico se encuentra sobretodo en las aguas subterr�neasexistentes en la secuencia de sedimentaci�n que tuvo lu-gar en el valle durante la transgresi�n marina del Holo-ceno. Las aguas subterr�neas del Pleistoceno y acu�ferosm�s antiguos est�n mayoritariamente libres de ars�nico.Es probable que las concentraciones de ars�nico aumen-ten en los pr�ximos aos en muchos pozos de tipo tuboperforados manualmente, por lo que ser� esencial efectuarun muestreo regular. Los acu�feros ubicados a m�s de 200m bajo las llanuras de inundaci�n ofrecen buenas pers-pectivas de abastecimiento a largo plazo sin problemas dears�nico, pero pueden estar limitados por las amenazas dela intrusi�n salina y de la precolaci�n de ars�nico desdeniveles superiores.

Keywords Arsenic · Bangladesh · Contamination ·General hydrogeology · Hydrochemistry

Introduction

Arsenic in groundwater in the alluvial and deltaic plainsof Bangladesh and West Bengal (India) has resulted in theworst case of mass chemical poisoning in the world(Smith et al. 2000). The occurrence of arsenic in theBengal Basin is unusual because most of the documentedcases of arsenic contamination of groundwater are frommining and industrial regions or areas of recent volcanicactivity (e.g. Welch et al. 1988; Nriagu 1994). Further, theBengal Basin case is exceptional in the great areal extentof arsenic occurrence and the number of people affected.It is estimated that more than 25 million people inBangladesh drink water containing more than 50 �g/l ofarsenic, and possibly an additional 25 million drink waterwith 10–50 �g/l arsenic (Department of Public HealthEngineering, Government of Bangladesh (DPHE) 1999).

In Bangladesh, more than 120 million people live in anarea of 144,000 km2. Land use is dominantly agriculturalbut urbanisation and industrialisation are proceedingrapidly. Until the 1970s, drinking water was drawn dom-inantly from surface-water sources, and water-borne dis-eases such as cholera and dysentery caused millions ofdeaths. During the last three decades, at least 3–4 millionhand-pumped tubewells (HTWs) have been installed. Thetypical HTW is manually drilled to between 20 and 70 m,

installed with 3 m of 38-mm diameter slotted PVC casing,and attached to a lever-action suction pump. Groundwaterprovides over 90% of drinking water and a large majorityof irrigation supplies. Before the arsenic problem wasrecognised, ready access to bacteriologically safe waterfrom HTWs was widely recognised as the principal factorin the dramatic reduction in waterborne disease and infantmortality in Bangladesh, and quoted as a developmentsuccess (UNICEF 1998).

Arsenic was first identified in the groundwater of WestBengal in 1983, following a medical diagnosis of arsenicpoisoning (Saha 1984, 1995; Mazumder et al. 1988).Investigations in India in the 1980s and early 1990sprogressively identified the extent of pollution there(Public Health Engineering Department, Government ofWest Bengal, (PHED) 1991; Das et al. 1994, 1996).Unfortunately, this information was effectively unknownin Bangladesh until the early 1990s. The earliest knownanalyses of arsenic in groundwater in Bangladesh (re-ported by Dhaka Water and Sewerage Authority (DWA-SA) 1991) were from three municipal supply wells inDhaka City. All were below the analytical methoddetection limit (10 �g/l) and therefore attracted noattention. Arsenic was first detected in groundwater inBangladesh by the DPHE in 1993. Between 1995 and1998, a series of surveys revealed the extent of thecatastrophe (NRECA 1997; Jakariya et al. 1998 andDPHE 1999).

Chronic exposure to arsenic in drinking water resultsin skin ailments such as hyperpigmentation and keratosis,and leads progressively to cancers of the skin, to damageto internal organs, cancer and ultimately death (WHO1993; National Academy Press 2001). Symptoms maytake five to fifteen years or longer to develop. The currentstandard for arsenic in drinking water in both Bangladeshand India is 50 �g/l. In 1993 the WHO recommended aprovisional guideline level of 10 �g/l, based on thepractical limit of detection at the time. In 2001 theEnvironmental Protection Agency (EPA) in the UnitedStates adopted a reduced standard in the USA of 10 �g/lfor public water supplies. Even this lower limit is notexpected to be protective at the one excess cancer in 106

lifetime exposures (National Academy Press 2001). TheWHO guideline has not been adopted in either India orBangladesh. The treatment for arsenic poisoning requiresthe removal of exposure to arsenic in drinking water.Installing and maintaining safe water supplies in themagnitude now required severely challenges the capacityof the people and governments of Bangladesh and India.The scale of the clinical and social effects of arse-nic poisoning can be appreciated by reference to websites maintained by the West Bengal and BangladeshArsenic Crisis Information Centre (http://www.bicn.com/acic/) and Harvard University (http://phys4.harvard.edu/%7Ewilson/ arsenic_project_main.html). The full humandimension of the tragedy is still unclear and will dependin part on the rate at which mitigation programmes can beimplemented.

728


Secondary impacts on health may result from agricul-tural activities whereby arsenic in soil or irrigation wateris taken up by crops, and thereby enters the human foodchain. Preliminary data on this subject are reviewed byHuq et al. (2001) who conclude that it is a matter ofserious concern that requires immediate attention.

In this paper the principal observations of arsenicoccurrence at a regional scale (104–105 km2, DPHE 1999)are combined with results from sub-regional scale studies(103 km2, DPHE 1999) and localised studies (15 km2,Burren 1998; Perrin 1998) to establish a hydrogeologicalinterpretation of arsenic in groundwater of the BengalBasin. In this paper the occurrence of arsenic in ground-water is considered in relation to the geological history ofthe Bengal Basin, the groundwater chemical evolution,and possible anthropogenic influences. A detailed de-scription of mitigation options is beyond the scope of thispaper but, in conclusion, the aspects of mitigation that aredirectly related to groundwater resources management arediscussed.

Geomorphology

Bangladesh has a tropical monsoonal climate. Meanannual rainfall (Rashid 1991) is lowest in the west (e.g.Rajshahi: 1435 mm) and increases both to the northeast(Sylhet: 4177 mm) and the southeast (Chittagong:2740 mm) (Fig. 1). Long-term average maximum andminimum monthly temperatures at Dhaka range from25.5 and 11.7 C in January to 35.1 and 23.4 C in April.Despite the high rainfall, around 90% of river flows inBangladesh originate in India, Nepal and China. TheBengal Basin (Morgan and McIntire 1959), which con-stitutes the major part of Bangladesh and the adjoiningstate of West Bengal in India, is effectively the delta ofthe Ganges – Brahmaputra – Meghna (GBM) Riversystem (Fig. 1). These rivers show a broad transition frombraided plains through meander belts to tidal and estua-rine plains as they approach the sea, accompanied by ageneral decrease in the median grain-size of the bed load.They flood large parts of their alluvial plains each yearduring the monsoon. The Holocene floodplains arecharacterised by immature soil development over a thicksequence of sedimentary deposits, and the formation of aploughpan beneath agricultural land (Brammer 1996).The rivers converge to become the Lower Meghna Riverto the south of Dhaka, and their alluvial plains combine toform the largest delta in the world. The discharge of theLower Meghna (1.1�106 Mm3/yr) makes it the thirdlargest river in the world, but in terms of sediment transfer(c. 1�109 t/yr) it is by far the largest (Friedman andSanders 1978). The active delta has advanced south byabout 100 km in the last 1000 years (Bakr 1977). The tidalrange in the Meghna Estuary is mostly between two andfour metres.

Geology

Regional GeologyThe Bengal Basin is bounded by the Himalayas and theShillong Plateau to the north, the Indian platform to thewest, and the Indo-Burman ranges to the east (Morganand McIntire 1959). The alluvial plains of the GBM riversystem slope from north to south, smooth on a regionalscale but interrupted locally by ridges and basins. Pleis-tocene terraces – the Madhupur and Barind Tracts –locally interrupt the flat topography of central Bangla-desh, rising by up to 20 m above the adjacent floodplains.These tracts have an incised dendritic drainage, withchannels filled by organic-rich muds of Holocene age(Monsur 1995). It is convenient to consider the regionalgeology in terms of these three major subdivisions – theTertiary hills, Pleistocene terraces and the Holocenefloodplains. Arsenic contamination in the Bengal Basinoccurs predominantly beneath the Holocene floodplains.

Stratigraphic correlation of the Bengal Basin has beendifficult (Brunnschweiler and Khan 1978), and the Qua-ternary (Monsur 1995) is particularly poorly definedowing to the absence of well-exposed sections and thedifficulty of establishing absolute ages for the litholo-

Fig. 1 Location map, showing the names of places referred to inthe text. Solid shading indicates areas elevated above the surround-ing alluvial floodplains; MT, Madhupur Tract, BT, Barind Tract,and CHT, Chittagong Hill Tracts. The line ABC shows the line ofsection in Fig. 2

729


gies represented. A simplified stratigraphic sequence forBangladesh is shown in Table 1. The nomenclature differsfrom that used in West Bengal (Wadia 1975; PHED1991), where arsenic occurs mostly beneath the UpperDeltaic Plains, approximate time-equivalents of the Chan-dina and Dhamrai Formations.

Considerable thicknesses of Holocene sediment, in-cluding the Chandina and Dhamrai Formations, underliethe floodplains of the Ganges, Brahmaputra and Meghnarivers [Bangladesh Agricultural Development Corpora-tion (BADC) 1992]. The Dupi Tila Formation and theBarind and Madhupur Clays of the Pleistocene terraceswere deposited in the Lower Pleistocene or earlier(Monsur 1995). Whitney et al. (1999) estimated that thegeomorphic surfaces of the Barind and Madhupur Tractsdated from about 25,000 years BP and more than 110,000years BP respectively. The underlying Tertiary strata arealluvial or shallow marine clastic sediments and havelittle direct influence on the exploited groundwaterresources (Ravenscroft 2003). Only in northern Bangla-desh does the Holocene alluvium directly overlie theIndian continental crust [Master Plan Organisation (MPO)1987]. Due to the extreme incision of the GBM system,there are few remnants of sedimentation from the Middleor Upper Pleistocene, resulting in an age contrast of twoorders of magnitude between the two major alluvialaquifer systems, comprising Holocene and Lower Pleis-tocene sediments.

Quaternary History of the Bengal BasinGlobal climatic changes, uplift of the Himalayas andsubsidence in the Bengal Basin interacted to controlQuaternary alluvial sedimentation in Bangladesh (Umitsu1993; Ravenscroft 2003). Himalayan glaciation sup-pressed monsoonal circulation, thus reducing rainfalland river flows (Dawson 1992). For much of the last120,000 years, global sea level stood about 50 m below itspresent level, falling to a minimum of 130 m at 18 Ka BP.During low-stands, the GBM system occupied deeply

incised channels within a series of terraces now largelyburied beneath the Holocene floodplains. During the post-glacial transgression, maximum sedimentation initiallyoccurred on the submarine fan between 12.8 and 9.7 KaBP (Kudrass et al. 1999) but shifted nearshore after about11 Ka BP when sea level intercepted the coastal plainabout 50 m below the present surface (Goodbred andKuehl 2000). Kudrass et al. (1999) and Goodbred andKuehl (2000) concluded that during the mid-Holocene sealevel was slightly higher, the climate was warmer, and therivers discharged up to two and a half times more than inpresent times. Goodbred and Kuehl (2000) estimatesubsidence rates of 1 to 4 mm a year during the Holocene,the maximum being in the Sylhet Basin.

Hydrogeology

Highly productive aquifers occur beneath both theHolocene floodplains and Pleistocene terraces, the shal-lowest generally encountered within 10 to 60 m of thesurface. Most water wells are less than 100 m deep.Typically, 50–75% of the sequence can be successfullydeveloped to this depth (MPO 1987). Saline groundwateris present in parts of the coastal area (Ravenscroft 2003)and some inland areas (Hoque et al. 2003). In the coastalregion, fresh-water aquifers are encountered either withinthe first 25 m or below about 150–200 m depth. Anequivalent deeper aquifer is also exploited at some townsnorth of the coastal area (DPHE 1996), below an aquitardwhich is present in places but appears not to be regionallyextensive.

HydrostratigraphyIn the absence of reliable and extensive dating, the term‘Holocene aquifers’ is used in this paper to describe theyoung aquifers that underlie the alluvial, estuarine anddeltaic floodplains of Holocene age. The actual age of theaquifer sediments is inferred generally to be younger than

Table 1 Simplified stratigraphic succession of Bangladesh

Age Stratigraphic units Lithology Notes

Late Pleistocene–Holocene

Chandina Formation Upward fining, grey micaceous, medium andcoarse sand to silt with organic mud andpeat.

Forms major aquifers beneath recentfloodplains. Probably <150 m thick.Dhamrai Formation

Unclassified depositsLower Pleistocene Madhupur Clay Tough, red-brown to grey, silty-clay; residual

deposits; kaolinite and iron oxide.Often absent beneath Holocene floodplains.Thickness 6 to 60 m.Barind Clay

Plio-Pleistocene Dupi Tila Formation Yellowish-brown to light grey, medium andcoarse sand to clay; very weakly consoli-dated; depleted in mica and organic matter.

Forms major aquifers beneath the terracesand hills, and deeper aquifers beneath theHolocene floodplains. Hundreds to thou-sands of metres thick

Dihing Formation

Tertiary Tipam Group Yellowish-brown, weakly consolidated sand-stone and mudstone.

Minor aquifers in hills

Surma Series Consolidated sandstone and shale No significant aquifersBarail Series Consolidated sandstone and shaleJaintia Group Shale, limestone and sandstone

Mesozoic Sylhet Traps Basalt, shale and sandstone

After Alam et al. (1990), Khan (1991) and DPHE (1999)

730


about 11,000 years BP, coincident with a marine floodingsurface at about 50 m below modern sea level (Goodbredand Kuehl 2000). Slightly older, but younger than the lastglacial maximum (LGM), sediments are inferred to bepresent along the axes of major valleys such as theJamuna (Japanese International Cooperation Agency(JICA) 1976).

The Holocene aquifers, which include the Chandinaand Dhamrai Formations, reach a maximum thickness ofabout 100 m (JICA 1976; BADC 1992). Grain sizes fineupwards, from coarse sands and gravels at the base, tofine and very fine sands towards the top of the aquifer(MPO 1987; Davies 1989; BADC 1992). Hydraulic con-ductivity values span at least four orders of magnitude(Burgess et al. 2002), resulting in a highly transmissivemulti-layered aquifer (MPO 1987; Herbert et al. 1989).Silts and clays predominate in the upper few metres,forming a surficial aquitard, generally less than 10 mthick, with typical specific yield values of 2–3%, andvertical permeability values in the range 3–8�10-3 m/d(BADC 1992). This aquitard is extensive, but may not becontinuous across active and recently abandoned river-beds. The contact between the upper aquitard and theexploited aquifers is gradational. The aquifers are mostlymedium-to-fine and medium-to-coarse sands, with per-meabilities of 40–80 m/d. Short-term pumping tests onthe Holocene aquifers indicate a leaky response, but forlonger pumping periods the aquifer is best described asregionally unconfined (MPO 1987).

The Holocene sands are grey, highly micaceous, oftencontaining abundant organic matter (Davies 1989, 1995),and show relatively few signs of weathering, in contrast tothe thoroughly oxidised and weathered Dupi Tila sands ofPleistocene age (BADC 1992). The principal mineralog-ical components of the Holocene sands are quartz, plagio-clase feldspar, potassium feldspars, micas (muscovite,biotite and chlorite), and clays (smectite, kaolinite, illite),[Perrin 1998; Asian Arsenic Network (AAN) 1999].Organic matter is present at up to 6% by weight and ironoxyhydroxides occur as grain coatings and fine particulatematter. Pyrite is rare; where observed it is framboidal andapparently authigenic (Perrin 1998; Nickson et al. 2000).

The Pleistocene aquifer system is formed of Madhupuror Barind Clay overlying Dupi Tila sands. The Pleis-tocene clays are thicker (up to 60 m) and more consol-idated than the Holocene aquitards, with lower verticalpermeability and lower specific yield (BADC 1992). Theyellowish-brown Dupi Tila sand aquifer is tens of metresto more than a hundred metres thick. The sands containless mica and less organic matter than the Holocenesands. Permeabilities of the Dupi Tila sands are typically20–30 m/d, about half that of Holocene sediments withthe same median grain-size, an effect attributed to thepresence of secondary clays and iron oxides whichpartially clog pore throats (BADC 1982; Ahmed 1994).Despite leaky or confined responses during pumping tests,over periods of a few months the aquifer response is alsobest characterised as regionally unconfined (MPO 1987;BADC 1992).

In the coastal regions, at Khulna, Barisal and Noakhali,shallow fresh-water aquifers overlie saline groundwater atdepths of 20–30 m. Deeper sands, below about 150 m,form productive fresh-water aquifers which are apparent-ly protected from saline intrusion by intermediate claylayers (e.g. Rus 1985). North of the coastal area, clayeyaquitards are present in some places and at varyingdepths, e.g. at Meherpur, where there is an aquitard 30–65 m thick at a depth of 160 m (Burgess et al. 2002), buttheir lateral extent is only locally defined. Sands deeperthan about 150 m beneath the Holocene floodplains maybe equivalent to the Dupi Tila sands, but their identifi-cation is ambiguous because of the removal of the Barindand Madhupur Clays at times of lower base levels and thepaucity of absolute dates. Where deep clayey aquitardsexist, the sands below are commonly referred to as the‘deep aquifer’, although there is no generally agreeddefinition. Where the aquitards are absent, the deepersands may be Pleistocene in age, but they effectivelyconstitute deeper regions of the same multi-layered aqui-fer that at shallower levels is formed of Holocene sands.Across the Holocene floodplain in southern Bangladesh,the deeper levels of the aquifer are exploited for potablewater supply to depths of up to 350 m at individual towns(DPHE 1996).

Groundwater CirculationIn the north and centre of the country, the aquifer systembeneath the Holocene floodplains behaves essentially as asingle layer, but to the south and east layering becomesincreasingly important. Annual potential recharge is from400 to >1000 mm (MPO 1987), but actual recharge ismuch less because aquifer-full conditions develop duringthe monsoon. In the absence of pumping, the water tablefluctuates seasonally by around 4–6 m (UNICEF 1994).With the advent of pumping for irrigation, water tablefluctuations have increased. The effect is greatest in theDupi Tila aquifer beneath the Pleistocene terraces, whereseasonal depression of the water table to 15 m belowground level (bgl) is common (BADC 1992; Hasan et al.1998). Beneath the floodplains, the additional water tablelowering due to irrigation pumping is typically 2–3 m(UNICEF 1994). Only at Dhaka City has continuouspumping from the Dupi Tila aquifer for water supplyalmost completely suppressed seasonal fluctuations andcaused long-term decline of the water table (Ahmed et al.1999). Due to low topographic gradients on both thePleistocene terraces and the Holocene floodplains, hy-draulic gradients are very small, commonly 0.0001 (e.g.Burgess et al. 2002) and lateral groundwater flow in theshallow aquifer is very slow, the Darcy velocity beingabout 2 m per year. Three groundwater flow systems arepostulated to operate simultaneously on different scales:

– A local-scale flow system, between local topographicfeatures (floodplains, levees, flooded depressions,minor rivers), to a depth of about 10 m over distancesof a few kilometres.

731


– An intermediate-scale flow system, between regionallyextensive topographic features (hills, terraces and themajor rivers), with flow paths up to tens of kilometreslong and residence times of hundreds to thousands ofyears.

– A basinal-scale flow system, between the boundariesof the basin and the Bay of Bengal, with flow pathshundreds of kilometres in extent and residence times ofthe order of tens of thousand years. Radiocarbon agesfor groundwater in deep coastal aquifers (e.g. Rus1985; Aggarwal et al. 2000) relate to the closure of alow-stand flow system of this scale, buried during theHolocene transgression.

Where groundwater is pumped, the natural flowsystems are considerably disrupted and vertical compo-nents of flow dominate the groundwater flow system.

Water QualityBefore the discovery of arsenic contamination, the chem-ical quality of groundwater beneath the Holocene flood-plains was thought to be generally good (MPO 1987;Davies and Exley 1992), although the shallow ground-water is vulnerable to contamination by bacteria (Hoque1998). Iron was known to be a widespread nuisance, andsalinity a constraint in the shallow aquifers of the coastalarea. Subsequently, in addition to arsenic, the DPHE(1999) has identified manganese and boron as common,naturally occurring constituents, present in places abovethe WHO health-related guidelines for drinking water,0.5 mg/l in both cases.

Groundwater beneath the Holocene floodplains ismainly of the Ca-Mg-HCO3 type with relatively highmineralization (EC 500–1000 �S/cm), tending towards aNa-Cl type water near the coast. This contrasts withgroundwater from the Dupi Tila sands aquifer beneath thePleistocene terraces, which is typically of Na-HCO3 typeand less mineralised, EC 200–300 �S/cm (Davies andExley 1992; DPHE 1999). Groundwater beneath theHolocene floodplains is characterised by high bicarbon-ate, with HCO3

- commonly present at several hundredmg/l. It is predominantly anoxic, and mostly stronglyreducing, locally to the extent of methanogenesis (Ahmedet al. 1998; Hoque et al. 2003). Dissolved iron is typicallypresent at around 5–10 mg/l. Manganese commonlyexceeds 0.5 mg/l. Sulphate concentration is generally lowbeneath the Holocene floodplains, mostly less than about5 mg/l, although, as with nitrate, it is higher beneath areasof settlement (e.g. Burgess et al. 2002).

These chemical characteristics reflect the conditionsunder which groundwater beneath the Holocene flood-plain has evolved. Groundwater gradients in the Holocenesediments are likely to have been low since their de-position, and the aquifer would not have undergone theflushing experienced by the Pleistocene and older sedi-ments. The elevated bicarbonate concentrations, togetherwith the high dissolved iron and other indications ofreducing conditions, suggest that oxidation of organic

matter (Lovley 1987), combined with hydrolysis of feld-spar and weathering of mica (Breit 2001) are the dom-inant processes in the evolution of the groundwaterchemistry.

Hydrogeological SynthesisThe simplified hydrogeological section (after Ravenscroft2003) through northeast Bangladesh in Fig. 2 generalisesand contrasts the aquifer conditions where elevatedarsenic concentrations occur with those where arsenic isabsent. The principal differences are between thosesediments that pre- and post-date the 18 Ka BP sea levellow-stand. In the central area, the thick Madhupur Clayconfines brown Dupi Tila sands with a relatively sharptransition in grain size. The sands are weathered andoxidised, and contain less mineralised, Na – HCO3 type,water and lower concentrations of trace elements. To boththe east and west are the grey Holocene channel-fillsediments that belong to the Dhamrai Formation in theJamuna valley and the Chandina Formation along the OldBrahmaputra. The upper aquitards are thin, and they areseparated from the main aquifer horizons by thick finesands (marginal aquifers). The sands show few signs ofweathering, and the waters are more mineralised andstrongly reducing with high bicarbonate, iron and man-ganese contents. Brown clays at about 40 m depth inthe Jamuna valley are probably remnants of an UpperPleistocene terrace, although the underlying sands signif-icantly post-date the adjacent Dupi Tila sands. The OldBrahmaputra channel is apparently less deeply incisedthan the Jamuna, and some wells may penetrate the DupiTila. Thick clays with lenticular sand bodies characterisethe piedmont deposits at the foot of the Shillong Plateau.

Methods of Investigation and Sources of Data

The principal data sources for this paper are derived fromtwo survey programmes. The first is the GroundwaterStudies for Arsenic Contamination in Bangladesh (DPHE1999), a project carried out by Mott MacDonald Ltd andthe British Geological Survey (BGS). The second is theLondon-Dhaka Arsenic in Groundwater Programme con-ducted by University College London (UCL) and DhakaUniversity (Nickson 1997; Burren 1998; Perrin 1998;Burgess et al. 2001). The DPHE studies comprised thecompilation of existing data on arsenic in groundwaterand regional (104–105 km2 scale) and sub-regional(103 km2 scale) surveys of arsenic and its hydrochemicaland hydrogeological context (DPHE 1999). The UCL andDhaka University studies focussed on detailed local scale(10 km2) mapping of tubewell water quality, porewaterchemistry and hydrogeological controls on arsenic occur-rence (Burgess et al. 2002; Cuthbert et al. 2002). Heredata are drawn principally from the DPHE (1999)surveys, which covered two-thirds of the country (theRegional Survey), mainly central and southern Bangla-desh, the associated compilation of available data (the

732


Pre-existing Surveys), and the detailed survey of Meher-pur town in western Bangladesh (the Meherpur Survey ofthe London-Dhaka Arsenic in Groundwater Programme).

The Regional Survey was based on an average ofeight evenly-spaced groundwater samples in each of 250upazilas (sub-district), stratified to take account of mul-tilayered aquifers, but unbiased with respect to medicalreports, surface geology or the age of wells. The wellssampled were predominantly hand-pumped tubewells.Water samples were analysed for arsenic by hydridegeneration – atomic absorption spectrometry (HG-AAS)by the BGS in the UK, and spectrophotometry by DPHEin Bangladesh. One sample per upazila was analysed bythe BGS for cations by inductively coupled plasma atomicemission spectrometry (ICP-AES), and for anions by ionchromatography (IC), to build up a geochemical baseline.Quality control checks showed that the BGS results werethe more reliable, and these data are used in this paper.Full analytical results are given in DPHE (1999).

The Pre-existing Surveys were evaluated according totheir sampling and analytical methodology, documenta-tion, geographic referencing, and by comparison with theunbiased sampling by others in the same region whereavailable (e.g. NRECA 1997). The medically-orientedsurveys aiming to confirm the cause of suspected arseni-cosis were judged to overestimate the general levels ofcontamination. Field-test methods (based on the mercuric

bromide stain method) tend to underestimate arsenicconcentrations, and only reliably indicate exceedance of50 �g/l when arsenic concentrations actually exceed180 mg/l (DPHE 1999). Where statistical calculations areperformed on arsenic concentration data, results belowdetection limits have been processed as half the detectionlimit. The Regional Survey was considered to give themost regionally representative view of arsenic concentra-tions, on account of the unbiased sampling strategy usedand the quality control of the analytical results.

The Meherpur Survey (Burren 1998; Perrin 1998) wasbased on 150 groundwater samples from HTWs, irrigationtubewells and deep public supply tubewells (DTWs) overan area of 15 km2. Arsenic concentration was tested usinga field-test kit at 125 sites, from which 76 were sampledfor full hydrochemical analysis, selected to represent thefull range of arsenic concentration, borehole depth, andpumping regime. Samples were filtered at the wellheadusing 0.45 �m membrane filters (for anion analysis), orfiltered and acidified to pH 2 (for cation analysis). On-sitemeasurements were made of electrical conductivity, pH,alkalinity, dissolved oxygen and temperature. Purgingprior to sampling was limited by time constraints toapproximately 40% of the HTWs volume. Shallow anddeep irrigation tubewells with motorised pumps werepumped for at least 10 minutes to allow purging of at leastthree well volumes. Anion analysis was by IC and cation

Fig. 2 Simplified hydrogeological section through north-centralBangladesh (after Ravenscroft 2003). The lithological section isderived from several hundred individual logs reported by BADC(1992) that have been averaged as the most probable lithology ineach 3-m depth slice within the local administrative unit (union)through which the line of section passes. Lower case ‘g’ (grey) and‘b’ (brown) denote the dominant sand colour. The bold dotted line

shows the inferred position of the land surface during the lastglacial maximum. Large arrows show the general direction ofregional groundwater flow. The depth of the section represents thevarying average depth of wells in each area, which is governed bythe greater thickness of clays and lower permeability of sandsbeneath the Madhupur Tract. Permeability and EC data are alsofrom BADC (1992). Iron data are from Davies and Exley (1992)

733


analysis was by ICP-AES. Arsenic was determined byhydride-generation AAS (detection limit 1 �g/l). Fullanalytical details are given in Burren (1998) and sum-marised in Burgess et al. (2002).

Sediment samples were preserved immediately fol-lowing recovery on site by waxing the ends of the PVCcore sleeves. On extraction in the laboratory, approxi-mately 100 g of sediment were mixed with distilled water,stirred for five minutes and allowed to settle. Theporewater/distilled water mixture was then decanted,and filtered or centrifuged prior to analysis. The resultsmust be interpreted in the light of the inevitable possi-bility of oxidation occurring during porewater extractionprior to analysis. However, dissolved iron is generallyhigh in the porewaters, up to 30 mg/l, and has not ap-parently been oxidised and removed from solution. Also,porewater calcium concentration is similar to that of localgroundwaters, which suggests that calcite precipitationhas not occurred to any great extent. Porewater compo-sition is in general similar to pumped groundwater in theregion. Together these factors suggest that the integrity ofthe porewater has been adequately preserved duringtreatment, and the porewater hydrochemical profiles havenot been obscured.

Sediment samples were extracted from the wax-sealedPVC core-sleeves, oven-dried, disaggregated and mixedprior to analysis. One sub-sample from each core wassubjected to standard fusion with lithium metaborate;another sub-sample was treated with hot 6 M HNO3 toextract the readily soluble minerals (Perrin 1998). Anal-ysis was by ICP-AES for cations, by IC for anions, and byhydride-generation AAS for arsenic. Full analytical de-tails are given in Perrin (1998) and summarised inBurgess et al. (2002). It is emphasised that despite beingwax-sealed immediately on recovery at the drilling site,and oven dried over a 24-hour period immediately onextraction in the laboratory, some oxidation of the coredsediments may have occurred. Breit (2001) demonstratedthe sensitivity of iron and arsenic to the redox environ-ment by measuring oxidation of 50% of the extractableiron and arsenic in grey Holocene sediment on exposureto humid air for one week. Oxidation might therefore leadto an increase in the reported iron oxide concentrations,

and could potentially result in oxidation of As(III) on thesediments, but would not affect the total arsenic analysis.

Distribution of Arsenic in Groundwater

Regional DistributionDPHE (1999) compiled the results of more than 30,000arsenic tests from the Pre-existing Surveys, with results of3,534 analyses for arsenic from the Regional Survey, andthe data were compiled into a GIS-database. The Pre-existing Surveys, of which 72% of analyses were by field-test kit, covered all parts of the country except theChittagong Hill Tracts and some offshore islands. TheRegional Survey, conducted in two phases, covered thewhole country except for the Chittagong Hill Tracts. Asummary of the results, by region (administrative divi-sion), as the percentage of wells exceeding 50 �g/larsenic, is shown in Table 2. The 25% exceedance asindicated by the Regional Survey is the same as that by alldata sets combined. The Pre-existing Survey field testingindicated a slightly lower percentage exceedance of the50 �g/l arsenic limit, at 21%. The Pre-existing Surveylaboratory test data indicated a higher percentage ex-ceedance, at 34%. Inconsistencies between the data setsmight be expected due to differences in sampling strategyand analytical methods. The field-testing had a relativelyunbiased geographical sampling frame but suffered fromunreliable detection of arsenic in the range of 50–180 �g/l. The laboratory results are analytically superior but inmany cases the Pre-existing Survey samples were selectedfrom known contaminated areas. The Regional Surveygives the most realistic estimate of the frequency distri-bution of arsenic concentrations on account of the un-biased, consistent sampling strategy employed. The fre-quency distributions over seven concentration ranges aregiven in Table 3.

Variations of arsenic concentrations with depth andtime (see below) must be considered when mapping itsdistribution. Changes over time occur sufficiently slowlyto allow mapping of arsenic data collected within a fewyears, even though the tubewell age is a significant factor(DPHE 1999; Cuthbert et al. 2002). Variations of arsenicwith depth may be accounted for by mapping data from

Table 2 Summary of arsenic testing of water wells in Bangladesh

Number of tests Percent of wells with more than 50 �g/l of arsenic (%)

Division Field kittests

Pre-existingsurvey labo-ratory tests

DPHERegionalsurvey

All arsenictests

Field kittests

Pre-existingsurvey labo-ratory tests

DPHERegionalsurvey

All arsenictests

Barisal 1,396 403 295 2,094 9 31 14 14Chittagong 3,232 1,094 445 4,771 51 74 50 56Dhaka 6,175 2,189 988 9,353 17 34 31 22Khulna 4,819 2,036 474 7,329 30 30 41 31Rajshahi 5,891 1,712 1072 8,675 8 17 6 10Sylhet 1,264 1,440 260 2,964 11 35 21 24Bangladesh 22,777 8,874 3,534 35,185 21 34 25 25

Source: DPHE (1999) and <exref type=”URL”>http://.www.bgs.ac.uk/arsenic/bangladesh</exref>

734


wells with limited depth ranges, but the effect of verticalflow within individual tubewell catchment areas must beacknowledged (Burgess et al. 2002). Within these con-straints, the distribution of arsenic can be mapped eitheras concentration or as the frequency of exceedance of athreshold concentration. Figure 3 shows the percentage ofwells exceeding the Bangladesh drinking water standard(50 �g/l) in each upazila with 10 or more tests, using data

from the Pre-existing Survey and the Regional Survey(DPHE 1999). Despite the coarse resolution and thegeologically arbitrary boundaries, chloropleth mappingbased on administrative unit allows incorporation of themuch larger number of non-georeferenced data andqualitative field-kit results available from the Pre-existingSurveys, and hence gives a fuller perspective on humanexposure. A geostatistical interpolation of the ‘mostprobable’ arsenic concentration, interpolated from log-transformed arsenic concentrations at wells less than150 m deep in Fig. 4, using data from the RegionalSurvey, gives more insight into the geological controls.The log-transformation is justified by the demonstrationof Johnston (1998) that arsenic concentrations approxi-mate a lognormal distribution within discrete geologicalunits.

In whatever way the arsenic data are represented, thebroad regional patterns are the same. Arsenic concentra-tions exceeding 50 �g/l are encountered in most parts of thecountry, but most commonly in the southeast, southwestand the Sylhet Basin (northeast) (Fig. 3). In many areasadjacent to the Lower Meghna Estuary more than 80% ofwells exceed the 50 �g/l concentration limit. For example,Jakarya et al. (1998) have reported that 93% of 12,000wells in Hajiganj upazila in southeast Bangladesh exceedthe limit. The probability of encountering extreme arsenicconcentrations, above 250 �g/l, is also highest in the southand southeast (Fig. 4). High concentrations of arsenic arefound in the lower catchments of all three major rivers ofthe GBM system, indicating the existence of multiplesource areas and the likelihood of related mechanisms ofmobilisation across the entire Bengal Basin. Tongues ofhigh concentration of arsenic in groundwater extend up-stream along the major river courses, expanding along thecourse of the Meghna River into the subsiding Great HaorBasin of Sylhet. The band of lower arsenic concentrationextending NNW-SSE along the Khulna-Jessore ridge isalso of note as it may indicate an accumulation of coarsersediment along a Holocene course of the River Ganges.

Sub-regional DistributionArsenic concentrations vary systematically at a numberof smaller scales. Figure 5 shows the distribution ofarsenic in groundwater at Nawabganj upazila in northwest

Table 3 Frequency distribution of arsenic concentrations

Concentration range (�g/l) Classification No. in class Percent in class (%) Percent exceeding lowerthreshold (%)

<10 Below WHO guideline 2,041 58 -10–50 Below Bangladesh standard 606 17 4250–100 Above Bangladesh standard 315 8.9 25

100–250 324 9.2 16250–500 178 5.0 6.9500–1,000 61 1.7 1.8

>1,000 3 0.09 0.09Total 3,534

Source: DPHE (1999) and <exref type=”URL”>http://.www.bgs.ac.uk/arsenic/bangladesh</exref>

Fig. 3 Percentage of wells with > 50mg/l arsenic. The map sum-marises >33,000 field and laboratory test data from drinking waterwells as compiled by DPHE (1999). The unit of aggregation isthe upazila, of which there are 490 in Bangladesh. A minimumcriterion of 10 results per upazila was applied. No depth classifi-cation was applied in selecting wells

735


Bangladesh, and a summary of arsenic concentrations isgiven in Table 4. The survey of Nawabganj (DPHE 1999)incorporated 58 groundwater samples over an area of400 km2, approximately one sample per 7 km2. Nawab-ganj lies beyond the worst affected areas of south andsoutheast Bangladesh (Fig. 4), but is an example of aspatially restricted instance of excessive arsenic concen-tration, a “hot spot”, which was identified following in-vestigations of high rates of arsenicosis in 1993. Extremearsenic concentrations, >250 �g/l and even exceeding1,000 �g/l, are restricted to an area close to the urbancentre. A strong geological control is evident: the DupiTila sands that underlie the Barind Clay on the BarindTract and beneath the River Mahananda floodplain areunaffected by arsenic. Arsenic concentrations in the Holo-cene strata are spatially variable.

Although Table 4 lists a similar number of contami-nated wells on the Alluvial Sand and the Alluvial Silt, themap shows that most of the contaminated wells on theAlluvial Sand are located near where it pinches out. These

wells are probably screened in the underlying Alluvial Siltunit (DPHE 1999). The Alluvial Sand corresponds to theactive floodplain of the River Ganges, while the older andthinly bedded Alluvial Silt unit is believed to have beendeposited in the early Holocene when an estuary extendeddeep along the Ganges valley (DPHE 1999). The contrastbetween the Holocene units demonstrates how bothdepositional environment and geological age are impor-tant factors in controlling arsenic mobilisation.

Arsenic concentrations have been mapped at a yet moredetailed scale at Faridpur by Safiullah (1998), Jessore byAAN (1999), and Meherpur (Burren 1998), Chaumohanipart of the Noakhali urban area (Mather 1999), Magura(Cobbing 2000) and Manikganj (McCarthy 2001), underthe London-Dhaka Arsenic in Groundwater Programme.Arsenic concentration in groundwater from HTWs lessthan 60 m deep at Meherpur, on the Ganges floodplain inwestern Bangladesh, is shown in Fig. 6. The MeherpurSurvey, incorporating five samples per km2 over 15 km2,illustrates arsenic variability at a scale nearly two orders ofmagnitude more detailed than the survey at Nawabganj.Arsenic concentration in groundwater around Meherpurranges from less than 1 �g/l to nearly 900 �g/l. Of theboreholes sampled, 55% have arsenic concentrationsgreater than 50 �g/l and only 18% have arsenic concen-trations of less than 10 �g/l. The spatial distributionindicates a belt of low arsenic concentration (<50 �g/l),500 m wide, with a north-south trend between Ujjalpur andMeherpur. Within this belt the average arsenic concentra-tion is 15 �g/l; beyond it the average is 200 �g/l (maximum890 �g/l). The pattern of this low-arsenic belt is suggestiveof a palaeochannel cutting across the present-day flood-plain of the Bhairab River. The apparent “channel” isdefined by the extreme spatial variability along its margins.For example to the northeast of Ujjalpur one boreholebeyond the “channel” pumps water with 250 �g/l arsenic,and lies just 105 m from a borehole within the “channel”with arsenic at less than 1 �g/l. Within the “high-arsenic”regions on either side of the “channel”, occasional indi-vidual boreholes or small groups of boreholes have arsenicconcentrations of less than 50 �g/l.

Safiullah (1998), AAN (1999), Mather (1999) andMcCarthy (2001) have also mapped lenticular bodies ofhigh and low arsenic concentration with widths measuredin hundreds of metres, that may correspond to palaeo-channel and oxbow-lake deposits. In these cases, a

Fig. 4 Arsenic concentration in groundwater in drinking waterwells less than 150 m deep. The input data are laboratory analysesfrom the Regional Survey (DPHE 1999 and http://bgs.ac.uk/arsenic/Bangladesh). The areal distribution of arsenic concentra-tions was interpolated from logarithms of arsenic concentration at3,500 evenly spaced points using the ArcView Spatial Analyst�software with the Inverse Distance Weighted method and includingthe eight nearest neighbours. Areas with no shading have no orinsufficient data

Table 4 Summary of arsenic determinations in Nawabganj Upazila

Surfacegeological unit

No. ofwells

Arsenic concentration (�g/l)

Mean1 Max. >50 >10

Alluvial Sand 21 7.0 742 24% 43%Alluvial Silt 22 4.9 1,524 14% 45%Barind Clay /Dupi Tila

15 <0.6 <0.6 0% 0%

1 Calculated as the mean of the logarithms of arsenic concentration.All analyses by ICP method on acidified, filtered samples. Datasource DPHE (1999)

736


meander-belt sedimentary model could explain the rapidlateral and vertical variations of arsenic occurrence ingroundwater, reflecting the contrast between relativelyoxic channel sands compared to more reducing overbankmuds. In general, these observations demonstrate thepossibility of linking spatial patterns of arsenic concen-tration in groundwater, and their variability, to sedimen-tological features that may have geomorphological man-ifestation, and hence the importance of such detailedspatial surveys.

Depth DistributionThe majority of wells sampled in the Regional Survey andthe Meherpur Survey were hand-pumped wells with atypical screen length of 3 m, and low discharge. Inaquifers hundreds of metres thick these wells approximatepoint data, and hence it is reasonable to interpret theconcentration data in terms of the depth distribution ofarsenic in groundwater. The occurrence of arsenic con-centrations exceeding 50 �g/l is shown in relation to well

depth in Fig. 7 and is summarised in Table 5. There is astrong correlation between the occurrence of arsenic ingroundwater and the depth of wells, although the precisepattern varies among regions. In general, the highestconcentrations, and also the greatest spatial variability,occur a few tens of metres below the ground surface, anddecrease rapidly below about 100 m. In wells deeper than200 m, arsenic concentration is generally negligible.

Fig. 5 Distribution of arsenicconcentrations in wells atNawabganj Upazila in relationto surficial geology. Surveydata, all analysed at the BGSlaboratories, are taken fromDPHE (1999). The main chan-nel of the River Ganges and theBarind Tract (underlain byBarind Clay over Dupi Tilasands) are indicated by shading.The Holocene floodplain unitsare labelled as follows: asd,Alluvial Sand, asl, Alluvial Silt,asc, Alluvial Silt and Clay, ppc,Marsh Clay and Peat. The geo-logical boundaries were digi-tised from Alam et al. (1990). Inthis area, the Alluvial Sand isequivalent to the active Gangesfloodplain, and the Alluvial Siltand Alluvial Silt and Clay areequivalent to the Mahanandafloodplain. The map grid is theBangladesh Transverse Merca-tor projection

Table 5 Arsenic distributions in groundwater by well depth

Well depth No. ofwells


Mean1 Max. >50 >10

<10 m 36 20 260 33% 69%10–30 m 576 34 1,090 52% 78%30–100 m 1,033 4.9 1,670 32% 49%

100–200 m 92 5.6 250 20% 54%>200 m 283 0.7 110 0.7% 3%All 2,023 6.7 1,670 35% 51%

Source: GSACB Regional Survey, Phase I (DPHE 1999). 1Calcu-lated from the mean of the logarithms of arsenic concentration

737


Wells shallower than 5 m, and especially dug wells, arecommonly uncontaminated. The absolute range of con-centrations in most depth intervals spans several orders ofmagnitude. Thus in Fig. 7 the data have been representedas the proportion of wells in each 10-m interval (withmore than ten data points) that exceed 50 �g/l. Figure 7also shows profiles within three geomorphic sub-unitssampled in the Regional Survey. The profile from the OldMeghna Estuarine Floodplain shows the simplest patternand the sharpest reduction with depth. Samples from theGanges floodplains suggest a second peak below 50 mdepth. The Sylhet Basin, where there is no overalldecrease of arsenic concentration to a depth of 150 m,shows the greatest variation to this trend. It has beenproposed (Ravenscroft et al. 2001) that the depth peaks of

arsenic concentration can be related to the diachronousformation of paludal basins during the Holocene trans-gression, where degradation of peat provides the strongredox driver required to account for the extreme arsenicconcentrations observed (McArthur et al. 2001).

The local arsenic depth profile at Meherpur is consis-tent with the regional picture (Burgess et al. 2002).Groundwater with the highest arsenic concentrations (200to 890 �g/l) is pumped from depths shallower than 45 m.Variability is highest in the shallow groundwater, wherethe maximum range of arsenic concentrations, frombelow the analytical method detection limit to 890 �g/l,is observed. Arsenic appears to occur at a lower and lessvariable concentration, commonly around 100 �g/l, ingroundwater pumped from depths greater than 45 m.

Fig. 6 Small-scale variation ofarsenic concentration near Me-herpur Town (after Burren1998). The location of sampledwells less than 60 m deep areshown with their arsenic con-centrations divided into fourclasses. Isopleths of arsenicconcentration show the align-ment of a low-arsenic (<50 �g/l)band roughly parallel to thepresent river channel. The mapgrid is the Bangladesh Trans-verse Mercator projection

738


However, the data at these greater depths are too sparse todraw firm conclusions on a local level.

At Meherpur, a more precise, site-specific view of thedepth profile of arsenic in the aquifer is illustrated by thearsenic concentration of porewater eluates from a coredborehole at Ujjalpur Village located away from HTWpumping influences (Figs. 6 and 8). There is a single,distinct peak concentration of arsenic in porewater be-tween 18 and 21 m depth, where arsenic concentrationexceeds 300 �g/l (range 50 to 500 �g/l). In contrast, atdepths of less than 10 m there are elevated chlorideconcentrations of between 80 and 90 mg/l. The chlorideprofile probably reflects the limiting depth of activegroundwater circulation with anthropogenic influence onchloride concentration, controlled partly by the subduedtopography and partly by the occurrence of a silty claylayer just below 10 m depth.

Characteristic depth profiles of arsenic in groundwaterhave been described at Lakshmipur and Chaumohani (partof the Noakhali urban area), in the coastal region ofsoutheast Bangladesh (DPHE 1999; Mather 1999). Herethe shallow Holocene aquifer, <30 m thick, containsgroundwater with a high arsenic concentration, and adeeper Pleistocene aquifer, >150 m deep, is almost free ofarsenic. In the Holocene aquifer at Lakshmipur, thearsenic concentration in 87% of wells exceeds 10 �g/l andin 73% it exceeds 50 �g/l. In 84% of wells in theHolocene aquifer at Chaumohani it exceeds 50 �g/l. In the

Pleistocene aquifer, the arsenic concentration exceeds50 �g/l in only one out of 17 samples at Lakshmipur, andnone of 10 exceed it at Chaumohani. In all other samplesfrom the Pleistocene aquifer in these areas the arsenicconcentration is below 10 �g/l. Between 30 and 150 mdepth, there are few wells because the groundwater isgenerally brackish.

Temporal TrendsThe Eighteen District Towns project (R. Dierx, pers.comm. 1999), which covers towns in most of Bangladesh,has monitored public water supply production wells forarsenic since 1996. While some wells show no clear trend,some wells do show an increase of arsenic over this shortperiod (DPHE 1999; Burgess et al. 2002). However, thereare no monitoring data of longer-duration on arsenic ingroundwater in Bangladesh. It is appropriate therefore touse tubewell age as a surrogate time parameter. TheRegional Survey data are suitable for analysis because theage of the sampled wells is reliably known and had notbeen a factor in their selection. In addition, the wells weresampled and analysed in a consistent fashion. The datawere transformed into the proportion of wells exceeding50 �g/l in each age group with 10 or more data points.The results of this analysis (Fig. 9) suggest that in manywells arsenic concentration increases to >50 �g/l duringthe first 5–10 years after installation, after which con-

Fig. 7 Depth distribution of arsenic in groundwater. The tracesrepresent the average percentage of wells with total depths fallingin 10 m intervals and the arsenic concentrations exceeding 50 �g/l.The solid trace represents 1,786 samples from the Regional Survey

of DPHE (1999), while the ornamented traces are geologicallyclassified subsets from the survey, where triangles indicate wells onthe Ganges River Floodplain (n=772), squares the Old MeghnaEstuarine Floodplain (n=266), and circles the Sylhet Basin (n=97)

739


centrations level off. An independent analysis of the samedata using a semi-variogram approach (Richard Howarth,Visiting Professor of Mathematical Geology, UniversityCollege London, pers. comm. 1998) reached the sameconclusion and identified a sill after about eight years andsignificant increases (at the 99% level) in the scale ofexceedance of the 50, 100, 150 and 200 �g/l thresholds.At a regional scale the data are vulnerable to bias ifyounger tubewells have been installed in areas of lowerarsenic concentration. The Meherpur Survey data, fromthe same aquifer as investigated under the RegionalSurvey, are illustrated in Fig. 8. Although the inter-annualvariations are unsurprisingly greater with a smaller dataset such as obtained in the Meherpur Survey, the sameoverall trend of arsenic concentration increasing withtubewell age is apparent at this local scale, which isunaffected by spatial bias of the timing of tubewellinstallation.

Increasing arsenic concentrations with time could beattributed to lateral migration of arsenic in the aquifer,leakage from adjacent or overlying aquitards, or a change

in redox conditions. There are no data to suggest redoxchanges during pumping, though they may occur. Mod-elling of generalised groundwater flow scenarios by DPHE(1999) suggests that arsenic is unlikely to move laterallyby more than a few metres to a few tens of metres a yearunder the very low prevailing horizontal hydraulic gradi-ents. Modelling of the specific conditions at Meherpur(Cuthbert 1999; Cuthbert et al. 2002) suggests that fortubewells screened below an arsenic-rich aquitard, arsenicbreakthrough would occur within 2–20 years of the onsetof pumping, depending on the sorption parameters spec-ified. This timing of arsenic breakthrough is consistentwith the field data that indicate arsenic concentrationsincrease with tubewell age, and supports the hypothesisthat vertical leakage is the principal cause of changingarsenic concentration in tubewell discharge with time.

Fig. 8 Arsenic concentration profiles in sediment and water atUjjalpur village in Meherpur (after Perrin 1998) are compared withthe lithological section at the site. Squares are porewater arsenicconcentrations (as ppb, equivalent to �g/l); open diamonds repres-ent the arsenic content of HNO3 extracts of the sediment; and

triangles represent the total arsenic content of the sediments (byfusion, as ppb). The discrete bars represent the arsenic concentra-tion in the vicinity of the cored borehole, and screened intervals ofhand tubewells

740


Relation of Arsenic in Groundwater to the Geology

Stratigraphic OccurrenceTo identify general relationships between surficial geol-ogy and arsenic in groundwater, all georeferenced welllocations were superimposed on a digitised version of theGeological Map of Bangladesh (Alam et al. 1990) usingthe ArcView GIS software. Detailed analyses are given inDPHE (1999) and Ravenscroft (2001), and a summary isgiven in Table 6. In Bangladesh, the surface geologicalunit generally has a depositional association with theaquifer that underlies it, to depths of about 100 m. TheMadhupur and Barind clays and Dupi Tila units have beencombined because they refer to underlying aquifers of asimilar age, depositional environment and diagenetichistory. Some errors may result from the coarse resolutionof the geological map, and also from situations wherewells are sunk in valleys filled by younger sedimentwithin the Tertiary hills and Pleistocene terraces, or

penetrate oxidised aquifers beneath Holocene sedimentson the GBM floodplain.

Groundwater associated with the Holocene deposits ismost affected by arsenic. However, it is clear thatprovenance and depositional environment are additionalcontrols on arsenic distribution. The map of Alam et al.(1990) is based on lithology, and its units are not uniqueto individual river systems. The same GIS overlaytechnique was applied to a map of physiographic units(FAO/UNDP 1988) and showed significant differencesbetween the various floodplains. Beneath the Gangesfloodplains, water in 35% of 1,747 wells had arsenicconcentrations exceeding 50 �g/l, in contrast to 25% of524 wells beneath the Brahmaputra (and Tista) flood-plains, and 53% of 810 wells on the Meghna floodplains(including the Sylhet Basin). It should, however, be notedthat provenance and grain size are related. In particular,large parts of the Brahmaputra and Tista floodplains are

Fig. 9 Temporal trends of ar-senic in groundwater. The per-centage of wells with arsenicconcentrations exceeding 50 �g/l in each year of construction isshown by squares for wells inthe Regional Survey (afterDPHE 1999) and by crosses forwells in the Meherpur survey(Burren 1998). The figure alsoshows the results of polynomialregressions fitted to each dataset

Table 6 Occurrence of arsenic in groundwater related to surficial geology

Geological unit(s)1 Age Geomorphic equivalent orlocation


No. ofwells2

Mean3 Max. >50

Alluvial Sand U. Holocene Active floodplains 144 7 890 27%Alluvial Silt and Clay Holocene Brahmaputra and Meghna River

Floodplains and Sylhet Basin668 8 700 32%

Deltaic Silt Holocene Ganges Floodplain 544 15 1,670 45%Alluvial Silt Holocene Ganges and Brahmaputra River

Floodplains428 5 1,450 26%

Chandina Formation L.-M. Holocene Old Meghna Estuarine Floodplain 152 88 1,090 77%Old and Young Gravellysand

U. Pleistocene–Holocene Tista Fan and floodplain 45 19 70 16%

Dupi Tila and DihingFormations

L. Pleistocene–Pliocene Madhupur and Barind Tracts 151 1 140 13%

Surma Series Tertiary Sylhet and Chittagong Hill Tracts 26 2 130 4%

Data source: DPHE (1999). 1Geological units as per the map of Alam et al. (1990). The Dhamrai Fm underlies parts of the ‘Alluvial Silt’and ‘Alluvial Silt and Clay’ units. Other Holocene units not referred to in Table 1 are stratigraphically ‘unclassified’. 2All wells less than100 m deep. 3Calculated from the mean of the logarithms of arsenic concentration

741


underlain by medium and coarse sand, while the sedi-ments beneath the Meghna floodplains are finer grained.

Groundwaters in the Pleistocene and older aquifers arelargely free of arsenic. No evidence has been found of anyextensive or severe contamination in aquifers pre-datingthe LGM. It is yet to be established whether arsenic thatmay originally have been present in groundwater has beeneither immobilised in the solid phase or removed byflushing.

Arsenic in SedimentsThe occurrence of arsenic in alluvial sediments is notunusual (Welch et al. 1988), and the arsenic content of theGBM alluvial sediments is not particularly high, but it isunusual for arsenic to be mobilised into groundwater soextensively and at such high concentrations. The averagearsenic content of the Earth’s crust is 1.8 ppm and arsenicis most abundant in shales (Mason 1966). Based on a datacompilation from West Bengal and Bangladesh, DPHE(1999) report average total arsenic contents of fluvio-deltaic sediments of 15.9 ppm for 134 onshore samplesand 10.3 ppm for 96 offshore samples. Although most ofthe onshore samples were collected in areas of higharsenic concentration in groundwater, this alone does notaccount for the extensive and extreme contaminationencountered in the Bengal Basin. Based on a limited dataset, Datta and Subramanian (1994) report average arseniccontents of riverbed samples to be 2.03 ppm in theGanges, 2.79 ppm in the Brahmaputra and 3.49 ppm inthe Meghna.

PHED (1991), Nickson et al. (1998) and Perrin (1998)all note discrepancies between the arsenic content mea-sured by selective extraction and by microprobe analysisof individual mineral grains. A sedimentological study byImam et al. (1997) noted the ubiquitous presence of iron-rich coatings on the sands, while analysis of grain coatingsfrom West Bengal found more than 2,000 ppm of arsenic(PHED 1991). Nickson (1997), Perrin (1998) and AAN(1999) all identified pyrite in the framboidal crystalform indicating that it is of diagenetic origin and henceprimarily a sink rather than a source for arsenic underpresent conditions. Furthermore, total arsenic in sedimentcorrelates strongly with iron but not with sulphur, sup-

porting the view that arsenic is primarily associated withoxyhydroxides and not sulphides (Nickson et al. 1998 and2000).

Nickson (1997), Perrin (1998), AAN (1999) and DPHE(1999) all analysed sediments from cored boreholes on theGangetic Plains of Bangladesh, demonstrating the litho-logical and stratigraphical controls over the depth profilesof arsenic in groundwater. The core analyses show thatarsenic content in sediments is greatest in fine-grainedstrata, and usually also within the first few tens of metresdepth. At Meherpur, Perrin (1998) measured arseniccontent of the bulk sediment and of selective extractions,in addition to porewater. Total arsenic content of sedi-ments at Meherpur ranges from 1.4 to 35 ppm, spanningthe range of sediment analyses from Faridpur, 11 to28 ppm (Nickson 1997), and Nawabganj, 2 to 11 ppm(DPHE 1999). The depth profiles of arsenic in the aquiferat Meherpur (Fig. 8) demonstrate a close agreementbetween arsenic concentrations in porewater, the arseniccontent of selective extractions, and the arsenic concen-tration measured in nearby wells. The bulk sedimentarsenic content demonstrates the same trend but is muchhigher in all cases.

Arsenic concentrations in the selective extractionswere positively correlated with iron, manganese and to alesser extent aluminium, and the bulk sediment analysesindicated strong correlations of arsenic with both iron andaluminium. This is possibly due to an association of bothiron oxyhydroxides and clay minerals within the finersediment fraction, but raises the possibility that the Fe-Asassociation could be due to dissolution of an alumino-silicate phase, as observed by Breit (2001).

Relation to Groundwater Chemistry

Groundwater in the Holocene aquifers beneath the GBMfloodplains characteristically contains negligible dis-solved oxygen and registers low redox potentials on aPt-electrode under field conditions (NRECA 1997; alsosee Table 7), with iron being extensively mobilised intosolution. High arsenic concentrations are restricted tothese strongly reducing conditions, though not all reduc-ing waters contain arsenic (DPHE 1999). Figure 10 shows

Table 7 Groundwater chemistry in an arsenic-affected area: Meherpur, Western Bangladesh

Site Depth(m)

T(C)

O2(%)

EC(�S/cm)

pH As(�g/l)

Ca Na K Mg Fe Mn Cl NO3 SO4 HCO3

S4 15 27.2 0 1,055 6.86 3 170 31.7 6.9 56.7 0.0 0.85 94.6 48.2 48.2 603S3 29 27.1 0 750 6.82 11 141 14.2 4.0 35.0 6.2 0.63 1.4 0.0 1.5 693S104 33 27.2 18 1,000 7.02 14 161 44.5 13.0 52.5 0.0 1.04 67.3 7.1 55.2 788S107 17 26.8 0 710 6.99 47 131 14.3 4.4 31.2 0.8 0.45 8.0 0.0 3.1 639S6 24 27.1 0 720 6.99 76 128 12.3 5.2 39.5 1.3 0.71 27.2 0.0 2.7 581S1 30 26.8 3 660 6.97 110 110 13.7 4.1 26.5 8.5 0.35 1.0 0.0 2.1 507S101 18 26.8 12 630 7.12 135 99 24.6 3.8 25.4 1.7 0.34 3.0 0.0 0.9 503S205 45 26.9 0 550 7.04 243 91 11.9 4.6 24.6 5.8 0.48 2.0 0.0 1.0 481S210 23 26.6 0 730 6.90 775 122 24.3 5.3 30.3 10.0 0.48 14.1 0.0 0.1 578

All units are mg/l except where stated otherwise. Source: Burren (1998)

742


the Pt electrode measurements, pH and arsenic in ground-water across Bangladesh.

The hydrochemical associations of arsenic are illus-trated by reference to the Meherpur study of Burren (1998)in Table 7 and Fig. 11. The tabulated data relate to wellswithin an area of about 5 km2, centred on the location ofthe cored borehole at Ujjalpur. Groundwater at Meherpuris predominantly anoxic, and of Ca-(Mg)-HCO3 type. Thehigh background values of total dissolved solids (140–1290 mg/l) are typical of groundwater in young, reactive,alluvial sedimentary sequences (Hem 1985). Dissolvedoxygen is generally less than 6% saturation (Burren 1998),suggesting that anoxic conditions are common, if notpervasive, in the aquifer. Dissolved iron ranges up to10 mg/l, reflecting the reducing conditions and the avail-ability of iron in the sediments. Iron concentrations showan approximately inverse relationship with nitrate andchloride (Fig. 11). Nitrate is generally absent, except atMeherpur town (NO3

- 20 to 88 mg/l) and Ujjalpur village(NO3

- 20 mg/l), where it may result from on-site sanitationin areas of dense human settlement (Burgess et al. 2002).Chloride has a distribution similar to nitrate in the shallowaquifer. Chloride reaches 150 mg/l at Meherpur and55 mg/l at Ujjalpur, but is less than 50 mg/l outside themain areas of settlement. Median bicarbonate values arearound 500 mg/l, with values greater than 700 mg/lrecorded beneath Meherpur and Ujjalpur.

At Meherpur, high arsenic concentrations in ground-water are associated with reducing conditions underwhich oxygen is limited, nitrate is absent, and iron andbicarbonate are at high concentrations (Fig. 11). Howev-er, in many cases groundwater with high iron contentcontains negligible arsenic. Nitrate ranges from zero to88 mg/l and has a strongly inverse relationship witharsenic. The positive correlation between iron and arsenicand the pervasively elevated bicarbonate concentrationsare similar to those recorded over broader geographicalareas (DPHE 1999; McArthur et al. 2001). The resultsfrom Meherpur support the hypothesis (Nickson et al.

2000) that desorption of arsenic has accompanied reduc-tive dissolution of iron oxyhydroxides in the aquifersediments, and suggest this is the principal mechanism bywhich arsenic is released to groundwater. Instances wherearsenic is present in groundwater despite the iron con-centration being low (Nickson et al. 2000) may be relatedto the precipitation of Fe carbonates (Mather 1999;Nickson et al. 2000).

Relation to Irrigation Pumping

Das et al. (1994) and Mallick and Rajagopal (1996)suggested that elevated arsenic concentrations in WestBengal and Bangladesh are caused by extensive pumpingof groundwater for irrigation. In this scenario, pumpinglowers the water table, and arsenic-rich pyrite in shallowsediments is oxidised, releasing iron, arsenic and sulphateinto solution. Certainly there is a temporal associationbetween the reports of arsenicosis and increases ingroundwater pumping for irrigation. However, there areno arsenic analyses dating from before about 1983 inIndia, or 1990 in Bangladesh, so such a hypothesis cannotbe directly tested. Statistical tests have been carried outon an upazila-based compilation of data to identify anyspatial association between elevated arsenic concentrationin groundwater and the extent of groundwater pumpingfor irrigation. The results are summarised in Fig. 12. Thepercentage of arsenic concentrations exceeding 50 �g/l ineach upazila was used as the measure of elevated arsenicconcentrations. The intensity, or impact, of groundwaterpumping was represented by two measures. First, sinceirrigation accounts for more than 90% of groundwaterpumping in Bangladesh (UNICEF 1994), the spatialimpact of groundwater pumping was first described byreference to the maximum recorded depth to the watertable over the period 1961–93. The second measure ofintensity was the percentage of the area of each upazilairrigated by groundwater in 1996. The latter measure

Fig. 10 Redox conditions andarsenic concentration showingsurvey data from NRECA(1997) compared to stabilitylines from Welch et al. (1988)where the shadowed boxes in-dicate the thermodynamicallyfavoured form of arsenic inwater

743


reflects the gross abstraction of groundwater per unit area.Both measures are negatively correlated with arseniccontamination. Although the proportion of variation ex-plained by the regression equations is small, both rela-tionships are significant at the 99% level and arguestrongly against irrigation pumping being a primary causeof the elevated arsenic concentrations in groundwater.

Mobilisation of Arsenic into Groundwater

Mechanisms of Release of Arsenic to GroundwaterBoth anthropogenic and geological sources have beenproposed to explain the elevated arsenic concentrations ingroundwater in the Bengal Basin. Suggestions for anthro-pogenic sources of arsenic have included mining wastes,industrial pollution, burning of fossil fuels, agrochemicals,

and wood preservatives in electric transmission pylons.However, while some of the hypotheses may account forisolated cases of pollution (e.g. Mazumder et al. 1992) nonecan provide a general explanation (DPHE 1999). Only ageological source can explain the extent and magnitude ofthe observed arsenic occurrence, and the lithological andsedimentological associations described above.

Two main explanations for the mobilisation of geo-logical arsenic have been proposed. The first, ‘pyriteoxidation – overabstraction’, considers that arsenic-richpyrite and arsenopyrite in the floodplain sediments areoxidised due to water-table lowering caused by intensivegroundwater pumping (Das et al. 1996; Mallick andRajagopal 1996). The alternative ‘oxyhydroxide reduc-tion’ hypothesis put forward by Bhattacharya et al. (1997,2001) in India, and Nickson (1997) and Nickson etal. (1998, 2000) in Bangladesh, proposes that adsorbed

Fig. 11 Hydrochemical associ-ations of arsenic in groundwaterat Meherpur (after Burren 1998)

744


arsenic is released by reductive dissolution of ironoxyhydroxides as the floodplain sediments become buriedand reducing conditions develop. This latter explanationemphasises the role of organic matter in generatingstrongly reducing porewaters. The ‘oxyhydroxide reduc-tion’ hypothesis is supported by the field evidence that:

– Arsenic-rich groundwaters are all strongly reducing;– Arsenic-rich groundwaters generally have high iron

and bicarbonate concentrations but very little sulphateor nitrate;

– The spatial distribution of arsenic does not correlatewith either water-table depth or the intensity ofgroundwater irrigation, but is associated with Holo-cene floodplains, and particularly with finer-grainedsediments;

– Maximum arsenic concentrations in groundwater arefound tens of metres below the depth of the deepestwater-table fluctuation, even in areas of little pumping;

– Pyrite is rather rare and where present occurs as anauthigenic rather than detrital mineral, more likelyacting as a sink for, rather than a source of, arsenic;

– There is a strong correlation between the iron andarsenic content of the Holocene sediments, but nocorrelation between iron and sulphur; and

– Sand grains in the Holocene sediments have pervasiveferruginous coatings with appreciable arsenic content.

A third possible mineralogical source for arsenic,which is not exclusive of an iron oxyhydroxide source andmay be consistent with the observed geochemical asso-ciations, is detrital biotite. Biotite is a common con-stituent of the Holocene sediments of the GBM flood-plains, and is known to contain arsenic at sites in easternBangladesh (Breit 2000). While evidence in support ofthe ‘oxyhydroxide reduction’ hypothesis is strong, theremay also be a contribution from the weathering of biotite,and the relative significance of the two processes mayvary with depth. Much remains to be done to identify themineralogical mechanisms of arsenic release in detail.More detailed discussions of the alternative mobilisationhypotheses and redox drivers are given by McArthur et al.(2001) and Nordstrom (2000).

Natural Processes Controlling Arsenic OccurrenceThe distribution of groundwater arsenic in Bangladeshmay be explained by a two-stage model that superimposesthe effects of Quaternary sea-level fluctuations upon acontinuum of fluvial-sedimentary processes, as summa-rised in Figs. 13 and 14. Arsenic enters the fluvial systemsin upland areas of India and Nepal by weathering ofsulphide and/or oxide bearing rocks. Arsenic releasedduring weathering is adsorbed by the iron, and possiblyalso manganese and aluminium, oxyhydroxides. Break-down of sulphides releases sulphate into solution. Thesediment load of the GBM system may be deposited andresuspended many times before reaching the currentsite of deposition. The upper reaches of the riversare characteristically braided; coarse-grained abandonedchannels may be preserved but little fine sedimentaccumulates. In the lower reaches, the mud and organicmatter content of sediments increases, especially inoverbank deposits, allowing accumulation of colloidaloxyhydroxides with their load of adsorbed arsenic.Locally, and sometimes extensively, these are interbeddedwith peat horizons (Brammer 1996). Each sedimentationevent provides an opportunity for fractionation of sul-phate from iron and arsenic (DPHE 1999).

Fig. 12a,b Relationship between elevated arsenic concentrationsand groundwater abstraction explored using two surrogate param-eters to represent the intensity of abstraction. The first surrogate (a)is the maximum depth to the water table, which under the extremelyflat conditions of Bangladesh is largely determined by irrigationpumping. The second parameter (b) is the proportion of the totalarea of each upazila that was irrigated by groundwater during 1996.Graph (a) is based on 27,797 analyses in 309 upazilas with aminimum of 25 tests per upazila. Graph (b) is based on 31,376analyses in 340 upazilas with a minimum of 25 analyses in eachupazila. The data are from DPHE (1999), NMIDP (1997) andUNICEF (1994)

745


Channel incision during the sea-level low-stands of thelate Quaternary divided the Bengal Basin into elongatehills parallel to the main rivers, accounting for the sharpsubsurface discontinuities in age (though not necessarilyfacies), aquifer hydraulic properties, and groundwaterquality in the transverse hydrogeological section por-trayed in Fig. 2. During sea-level low-stands, transversegroundwater flow was more important than at present,driven by the large lateral hydraulic gradients. Suppressedmonsoonal circulation reduced rainfall and the regionalwater table would have stood many tens of metres belowthe surface of the Madhupur and Barind tracts. Rechargewould have percolated rapidly, promoting oxidativeweathering and flushing, leading to the removal oforganic matter, the development of circum-neutral, oxic

conditions, and the recrystallization of amorphous oxy-hydroxides as hematite or goethite. The combined effectwas to immobilise any arsenic that had not previouslybeen flushed from the aquifer system.

The Dupi Tila sands have experienced such conditionsover hundreds of thousands of years, allowing the almostcomplete removal or immobilisation of arsenic. Buried,Late Quaternary terraces in the Jamuna Valley and thesoutheast coastal plain (with ages of >25 to �50 Ka BP)experienced a briefer period of oxidative weatheringduring the 18 Ka BP low-stand. It is anticipated that thiswould have removed or immobilised much arsenic, byadsorption onto residual Fe-oxide, but it is not certain thatthese deposits are completely free of arsenic. Thesesediments may account for the lower, but still significant,

Fig. 13 Present day processes affecting the mobilisation, fate and transport of arsenic in the Bengal Basin. The figure shows an idealisedsequence of events that may occur between the upper catchment and the Bay of Bengal

746


arsenic concentrations in groundwater at depths of be-tween 50 and 120 m beneath the present-day floodplains(Fig. 7).

Sea level rose rapidly from 18 Ka to 7 Ka BP, but amajor change in sedimentation occurred when sea levelintercepted the shallow coastal platform at about 11 KaBP (Goodbred and Kuehl 2000). The combination of abroad, shallow shelf with higher rainfall, greater riverdischarge and higher temperature provided ideal condi-tions for the formation of mangrove swamps and fresh-water peat basins. Such waterlogged conditions providedlittle possibility for the flushing of sediments by meteoricwaters. Fine sands deposited at the delta front and lowerfluvial regime are interbedded with organic-rich mud andpeat, the former providing the source of arsenic and thelatter driving the mechanism to generate strongly reduc-ing groundwater conditions.

After deposition of the Holocene sediments, a se-quence of chemical processes commences that may leadto the mobilisation of arsenic in groundwater. Decompo-sition of organic matter progresses with the microbialconsumption of dissolved oxygen, followed by the re-

duction of any nitrate present, and eventually by thereductive dissolution of solid phase ferric oxyhydroxides,releasing adsorbed arsenic into solution in groundwater. Ifreduction proceeds further and if sufficient sulphate isavailable, iron and arsenic may ultimately be sequesteredin diagenetic pyrite, but this does not appear to haveoccurred extensively. The key factors accounting forwidespread mobilisation of arsenic into groundwater inthe Bengal Basin appear to be (1) the efficiency ofseparating sulphur (as sulphate) in river water fromarsenic and iron in sediment that eventually forms theGBM floodplain deposits; (2) the abundance of organicmatter; and (3) the restricted groundwater flow due to thelow hydraulic gradients that have prevailed since depo-sition. Alternative geochemical pathways, due to varia-tions in sediment composition, can lead to methanogen-esis (Ahmed et al. 1998) or siderite formation, the latteraccounting for iron-deficient arsenic–rich groundwaterunder conditions of siderite saturation (Mather 1999).

It is probable that the events described during, andsubsequent to, the terminal Pleistocene transgressionconstitute a cyclical process that occurred many times

Fig. 14 Palaeohydrologicalprocesses controlling the accu-mulation, removal and immo-bilisation of arsenic in theBengal Basin. The ages (BP)are indicative of the events thatmay have occurred during andafter the last Glacial Maximum,but in generalised form arelikely to have occurred manytimes during the Quaternary

747


during the Quaternary, and would have affected manylarge alluvial basins throughout the tropical world.Recently, arsenic occurrence in groundwater at a similarscale to that observed in the Bengal Basin, has beendescribed in the Red River Basin of Vietnam (Berg et al.2001).

Human Influences on Arsenic MobilityHuman activities may modify the natural distribution ofarsenic in groundwater at a local scale, but none appear tobe regionally significant. Human waste (as sewage) mightbe a source of nitrate and sulphate in groundwater beneathareas of dense human settlement (Burren 1998); nitratemay then oxidise ferrous iron to ferric oxyhydroxides(Burren 1998) and partially remove arsenic from solutionby adsorption. Nitrate fertilisers may contribute to theeffect. Phosphatic fertilisers, on the other hand, maycompete with arsenate for adsorption sites and displacearsenic into solution (Acharrya et al. 1999). In northcentral and southeast Bangladesh, Davies and Exley(1992) showed that phosphate in groundwater beneath theJamuna floodplain has concentrations in the range of 3–8 mg/l. At a national scale, the spatial distribution ofphosphate is similar to that of arsenic (Ravenscroft et al.2001), but NRECA (1997) and AAN (1999) show poor(well by well) correlations between arsenic and phos-phate. While high phosphate and high arsenic are bothrestricted to the younger aquifers, it appears that phos-phate and arsenic have a common origin rather thanphosphate playing a role in mobilising arsenic.

Pumping for water supply and irrigation has increasedaeration of the upper aquifer, possibly leading to theprecipitation of iron oxyhydroxides and immobilisation ofarsenic by re-sorption in the very shallow zone of water-level fluctuation. Arsenic in groundwater pumped forirrigation is oxidised in the water distribution channelsand precipitated along with ferric iron in the fields(BADC 1992). Where rice is irrigated from arsenic-bearing aquifers, the transfer of arsenic to the soil zonecould be of the order of 1 Kg/ha/yr (assuming a grossirrigation requirement of 1000 mm/yr for rice and aninput concentration of 100 �g/l). The extent to whicharsenic added to the soil in this way might be leached togroundwater or transferred to the atmosphere bybiomethylation is presently unknown.

Models of arsenic transport in groundwater flowing tohand-pumped tubewells from an arsenic source zone at20 m depth (Cuthbert et al. 2002) have demonstrated thatvertical leakage will tend to increase the arsenic concen-tration in the tubewell discharge with time, where thetubewell screen is below the arsenic source zone. Wherethe tubewell is located within the wider catchment area ofa deeper, more productive, water supply or irrigationtubewell, arsenic will appear at the HTW more quickly,and seasonal discharge from the irrigation tubewell couldlead to seasonal variations in the arsenic content of theshallower HTW. These are, however, only secondaryinfluences on the occurrence of arsenic at HTWs.

Mitigation and Resource Management Issues

Mitigation activities will involve extensive arsenic sur-veys, long-term groundwater quality monitoring, com-munity awareness and mobilisation activities, and a rangeof possible physical interventions including (1) treatingarsenic-rich groundwater at the source, (2) developingalternative groundwater sources and (3) developing sur-face-water sources such as rivers, ponds or rainwater.Existing uncontaminated shallow wells will continue tobe an important source of drinking water for many years.However, with 25% of all wells containing >50 �g/l,critical questions relate to the sustainability of wells thatare presently arsenic-free despite being in affected areas,especially given that arsenic concentrations appear toincrease over time (DPHE 1999; Cuthbert et al. 2002).Detailed investigations of selected sites, and systematicmonitoring will be required to manage the risk to humanhealth.

Groundwater in deep aquifers, below about 200 m,presently contains minimal arsenic, but faces a risk ofleakage from shallow aquifers in the long term. Prelimi-nary modelling (DPHE 1999) based on a variety ofsorption scenarios suggests that cross-contamination wouldtake decades, probably longer. However, better definitionof vertical permeabilities and sorption characteristics underchanging redox conditions is required before predictionscan be made with confidence. Until these parameters arebetter defined, the precautionary principle warrants plan-ning on worst-case (e.g. low sorption) scenarios. Thesustainability of abstraction from deep aquifers may also beconstrained by the possibility of saline intrusion in thecoastal area and offshore islands. These aquifers areconfined downgradient by thick muddy sediments and arenot in continuity with the Bay of Bengal. Any negativeimpacts are likely to take many years to develop. However,the renewable yield of the aquifer is unknown, and aquantitative resource assessment and monitoring networkare high priorities.

Conclusions

Groundwater in Holocene alluvial and deltaic aquiferscontains arsenic of geological origin at elevated concen-trations over extensive areas of Bangladesh, threateningthe lives of more than twenty five million people. Arsenicconcentrations are highest in the upper 50 m of thesedimentary sequence. Below 100 m, arsenic concentra-tion reduces, and below 200 m the chances of drilling an‘arsenic-safe’ well approach 99%. The arsenic is derivedfrom multiple source areas in the upper catchments of theGanges, Brahmaputra and Meghna Rivers and is thoughtto have been transported through the river system ad-sorbed onto colloidal iron oxyhydroxides. Detrital arse-nic-bearing phyllosilicates, such as biotite, may alsocontribute to the arsenic content of the sediments and actas an additional source to groundwater. Following depo-sition, degradation of organic matter has led to reductive

748


dissolution of iron oxyhydroxides, releasing adsorbedarsenic to groundwater. There is, however, no evidence tosupport a causal connection between the pumping ofgroundwater for irrigation and widespread mobilisationof arsenic in the aquifers. The sharp contrast betweenarsenic-bearing groundwater in chemically reducingHolocene aquifers and arsenic-free groundwater in theoxidised Pleistocene aquifers is a result of the effects ofQuaternary sea-level fluctuations on the regional palaeo-hydrogeological evolution. Sediments that were not erod-ed during the last sea level low-stand were oxidised andflushed by meteoric water, removing or immobilisingarsenic. The incision of the main river channels andcoastal plains created the space to accommodate younger,Holocene deposits of organic-rich sediment and fine-to-medium sand that form the arsenic-affected aquifersencountered today.

The occurrence of arsenic in groundwater in theBengal Basin provides a number of general lessons.While the Bengal Basin may be exceptional, it is unlikelyto be unique. Groundwater in other open alluvial basins inhumid, and especially tropical, areas is likely to containarsenic at concentrations harmful to human health. Amore holistic and open-minded approach should beadopted in the assessment of groundwater resourceswhere these conditions apply. Conventional scientificwisdom did not recognise the possibility of widespreadand excessive arsenic occurrence in the alluvial aquifersof Bangladesh. Only a rigorous application of the pre-cautionary principle, whereby representative samplesfrom water supplies across the country were tested forall naturally-occurring health-related constituents andproperties, would have identified the situation that iscausing the suffering seen today in Bangladesh.

Acknowledgements PR and KMA thank Mr Kazi Nasir UddinAhmed, Additional Chief Engineer of the Department of PublicHealth Engineering for his support in conducting the GroundwaterStudies for Arsenic Contamination project. We also wish to thankthe project staff and the staff of DPHE for their co-operation duringthe project. We thank David Kinniburgh of the British GeologicalSurvey for planning and co-ordinating the analytical aspects of theRegional Survey. The Groundwater Studies for Arsenic Contam-ination project was financed by the Department for InternationalDevelopment (UK). The Natural Environment Research Councilprovided an Advanced Course Studentship and fieldwork allowanceto Melanie Burren. Jerome Perrin was supported by a fieldworkgrant from the University College London Graduate School. Wethank Mizanur Rahman of the Bangladesh Water DevelopmentBoard for provision of core-samples from the Ujjalpur borehole.Chemical analyses for the Meherpur study were carried out by theRobens Institute for Public and Environmental Health at SurreyUniversity, the Environmental Mineralogy laboratory of the NaturalHistory Museum in London, and the Natural EnvironmentalResearch Council ICP-AES facility at Royal Holloway College,London. Grateful thanks for help with analyses are due to AndrewTaylor, Chris Stanley, Vic Din, Nikki Paige and Tony Osbornfor assistance. Martin Gillham of Mott MacDonald Ltd, MikeMcCarthy of the Department for International Development and DrBabar Kabir of the World Bank are thanked for their support andencouragement. The script has been much improved due to helpfulreviews by Kirk Nordstrom and Alan Welch of the USGS. Last, butnot least, we wish to extend our sympathies to those people in

Bangladesh and West Bengal whose lives have been so tragicallyaffected by arsenic in groundwater.

References

AAN (1999) Arsenic contamination of groundwater in Bangladesh:interim report of the research at Samta Village. Asian ArsenicNetwork, Research Group for Applied Geology and theNational Institute for Preventive and Social Medicine, Bangla-desh

Acharyya SK et al. (1999) Comment on Nickson et al. 1998. Nature401:545

Aggarwal PK, Basu AR, Poreda RJ (2000) Isotope hydrology ofgroundwater in Bangladesh: implications for characterisationand mitigation of arsenic in groundwater. International AtomicEnergy Agency, Vienna, TC Project BGD/8/016, 23 pp

Ahmed KM (1994) Hydrogeology of the Dupi Tila Sand Aquifer ofthe Barind Tract, NW Bangladesh. PhD Thesis (unpub),University of London

Ahmed KM, Hoque M, Hasan K, Ravenscroft P, Chowdhury LR(1998) Occurrence and origin of water well methane gas inBangladesh. J Geol Soc India 51:697–708

Ahmed KM, Hasan MK, Burgess WG, Dottridge J, Ravenscroft P,van Wonderen J (1999) The Dupi Tila aquifer of Dhaka,Bangladesh: hydraulic and hydrochemical response to intensiveexploitation. In: Chilton PJ (ed) International Contributions toHydrogeology 21: Groundwater in the urban environment:selected city profiles. Balkema, Rotterdam, pp 19–30

Alam MK, Hassan AKMS, Khan MR, Whitney JW (1990)Geological Map of Bangladesh. Geological Survey of Bangla-desh

Asian Arsenic Network (1999) Arsenic contamination in Bangla-desh. Interim Report of research at Samta village. AsianArsenic Network

BADC (1982) ADB Second Tubewell Project Feasibility Study, vol3: Groundwater. Mott MacDonald Ltd and Hunting TechnicalServices. Report produced for Bangladesh Agricultural Devel-opment Corporation and the Asian Development Bann.

BADC (1992) Final Report of the Deep Tubewell II Project, vol2.1: Natural Resources. Mott MacDonald Ltd and HuntingTechnical Services. Report produced for Bangladesh Agricul-tural Development Corporation and the Overseas DevelopmentAdministration

Bakr MA (1977) Quaternary geomorphic evolution of the Brah-manbaria-Noakhali area. Geol Surv Bangladesh Rec, VoI 2

Berg M, Tran HC, Nguyen TC, Pham HV, Schertenleib R, Giger W(2001) Arsenic contamination of groundwater in Vietnam: Ahuman health threat. Env Sci Tech 35:132621–2626

Bhattacharya P, Chatterjee D, Jacks G (1997) Occurrence ofarsenic-contaminated groundwater in alluvial aquifers fromdelta plains, Eastern India: options for safe water supply. WaterResources Development 3(1):79–92

Bhattacharya P, Jacks G, Jana G, Sracek A, Gustaffson JP,Chatterjee D (2001) Geochemistry of the Holocene alluvialsediment of the Bengal Delta Plain: implications on arseniccontamination in groundwater. In: Bhattacharya P, Jacks G,Khan AA (eds) Groundwater arsenic contamination in theBengal Delta Plain of Bangladesh. Proceedings of the KTH-Dhaka University Seminar, KTH Special Publication, TRITA-AMI report 3084, pp 21–40

Brammer H (1996) The geography of the soils of Bangladesh.University Press, Dhaka

Breit GN (2000) Arsenic cycling in eastern Bangladesh: the role ofphyllosilicates. Proc Ann Mtg Geol Soc Am, Reno, Nevada,Nov 9–18

Breit GN (2001) Early diagenetic ferric oxide accumulations alongredox gradients: examples from modern and ancient fluvialsedimentary units. Geological Society of America AnnualMeeting, Boston, Nov 1–10, Abstracts Volume 33, No 6

749


Brunnschweiler RO, Khan MAM (1978) With Sherlock Holmes inthe Bengal Basin. Second Offshore Southeast Asia Conference,Singapore, 5 pp

Burgess WG, Ahmed KA, Burren M, Carruthers A, Cobbing C,Cuthbert MO, Mather SE, McCarthy EM, Perrin J, Chatterjee D(2001) The distribution, hydrochemical context and mobility ofarsenic in alluvial aquifers of the Bengal Basin. GeologicalSociety of America Annual Meeting, Boston, Nov 1–10,Abstracts Volume 33, No 6

Burgess WG, Burren M, Perrin J, Ahmed KM (2002) Constraintson the sustainable development of arsenic-bearing aquifers insouthern Bangladesh. Part 1: A conceptual model of arsenic inthe aquifer. In: Hiscock KM, Rivett MO, Davison RM (eds)Sustainable groundwater development. Geological Society,London, Special Publication 193, pp 145–163

Burgess WG, Ahmed KM, Cobbing J, Cuthbert MO, Mather SE,McCarthy E, Chaterjee D (2002) Anticipating changes inarsenic concentration at tubewells in alluvial aquifers of theBengal Basin. In: Bocanegra E, Martinez D, Massone H (eds)Groundwater and human development. Proceedings of theXXXII IAH and VI ALHSUD Congress, Mar del Plata,Argentina, October 2002, pp 365–371

Burren M (1998) Small-scale variability of arsenic in groundwaterin the District of Meherpur, Western Bangladesh. MSc Thesis(unpub), University College London

Cobbing J (2000) The spatial variation of arsenic in groundwater atMagura, western Bangladesh. MSc Thesis, University CollegeLondon (unpub)

Cuthbert M (1999) Modelling the transport of arsenic to handtubewells in the Holocene alluvial aquifer of Bangladesh. MScThesis (Unpub), University College London

Cuthbert M, Burgess WG, Connell L (2002) Constraints on thesustainable development of arsenic-bearing aquifers in southernBangladesh. Part 2: Preliminary models of arsenic variability ingroundwater. In: Hiscock KM, Rivett MO, Davison RM (eds)Sustainable groundwater development. Geological Society,London, Special Publication 193, pp 165–179

Das D, Chatterjee A, Samanta G, Mandal BK, Chowdhury TR,Chowdhury PP, Chanda CR, Basu G, Lodh D, Nandi S,Chakraborti T, Bhattacharya SM, Chakraborty D (1994) Arse-nic in groundwater in six districts of West Bengal, India: thebiggest arsenic calamity in the world. Analyst 119:168–170

Das D, Chatterjee A, Samanta G, Mandal BK, Chowdhury TR,Chanda CR, Chowdhury PP, Basu GK, Chakraborti D (1996)Arsenic in groundwater in six districts of West Bengal, India.Environ Geochem Health 18(1):5–15

Datta DK, Subramanian V (1994) Texture and mineralogy ofsediments from the Ganges-Brahmaputra-Meghna River systemin the Bengal Basin and their environmental implications.Environ Geol 30(3/4):181–188

Davies J (1989) Pilot study into optimum well design: IDA 4000Deep Tubewell Project. Vol 2: The geology of the alluvialaquifers of Central Bangladesh. British Geological SurveyTechnical Report WD/89/9

Davies J (1995) The hydrochemistry of alluvial aquifers in centralBangladesh. In: Nash H, McCall GJH (eds) GroundwaterQuality. Chapman and Hall, London, pp 9–18

Davies J, Exley C (1992) Hydrochemical character of the mainaquifer units of central and northeastern Bangladesh andpossible toxicity of groundwater to fish and humans. FinalReport. British Geological Survey Technical Report WD/92/43R

Dawson AG (1992) Ice Age Earth: Late Quaternary geology andclimate. Routledge, London

DPHE (1996) Presence of arsenic in groundwater in the 18 districttown projects. Short Mission Report. Department of PublicHealth Engineering, Bangladesh (unpubl)

DPHE (1999) Groundwater studies for arsenic contamination inBangladesh. Rapid Investigation Phase. Final Report. MottMacDonald Ltd and British Geological Survey. Report for theDepartment for Public Health Engineering and the Departementfor International Development

DWASA (1991) Dhaka region groundwater and subsidence study.Final Report. Engineering and Planning Consultants, Dhakaand Mott MacDonald Ltd, UK. Report for Dhaka Water andSeware Authority and the World Bank

FAO/UNDP (1988) Land resources appraisal of Bangladesh. Foodand Agriculture Organisation/United Nations DevelopmentProgramme, Ministry of Agriculture, Bangladesh

Friedman GM, Sanders JE (1978) Principles of sedimentology.Wiley, New York

Goodbred SL Jr, Kuehl SA (1999) Holocene and modern sedimentbudgets for the Ganges-Brahmaputra river system: evidence forhighstand dispersal to floodplain, shelf and deep-sea depocen-ters. Geology 27(6):559–562

Goodbred SL Jr, Kuehl SA (2000) The significance of largesediment supply, active tectonism, and eustasy on marginsequence development: Late Quaternary stratigraphy andevolution of the Ganges-Brahmaputra delta. Sediment Geol133:227–248

Hasan MK, Ahmed KM, Burgess WG, Dottridge J, AsaduzzamanM (1998) Limits on the sustainable development of the DupiTila aquifer, Bangladesh. In: Wheater H, Kirkby C, RushtonKR, Reid I (eds) Hydrology in a changing environment, vol II.Proceedings of the British Hydrological Society InternationalConference, Exeter, July 1998. Wiley, New York, pp 185–194

Hem JD (1985) Study and interpretation of the chemical charac-teristics of natural water. US Geol Surv Water Supply Paper2254, 3rd Ed

Herbert R, Barker JA, Davies J (1989) Pilot study into optimumwell design: IDA 4000 Deep Tubewell II Project. Volume6:Summary of the programme and results. British GeologicalSurvey, BGS Technical Report WD/89/14

Hoque BA (1998) Biological contamination of tubewell water.Environmental Health Programme Report, International Centrefor Diarrhoeal Disease Research, Bangladesh. Report forDepartment for International Development (UK)

Hoque M, Hasan MK, Ravenscroft P (2003) Investigation ofgroundwater salinity and gas problems in Southeast Bangla-desh. In: Rahman AA, Ravenscroft P (eds) Groundwaterresources and development in Bangladesh – background tothe arsenic crisis, agricultural potential and the environment.Bangladesh Centre for Advanced Studies. University Press,Dhaka

Huq SMI, Ara QAJ, Islam K, Zaher A, Naidu R (2001) Thepossible contamination from arsenic through food chain. In:Bhattacharya P, Jacks G, Khan AA (eds) Groundwater arseniccontamination in the Bengal Delta Plain of Bangladesh.Proceedings of the KTH-Dhaka University Seminar, KTHSpecial Publication, TRITA-AMI Report 3084, pp 9–96

Imam B, Alam M, Akhter SH, Choudhury SQ, Hasan MA, AhmedKM (1997) Sedimentological and mineralogical studies onarsenic contaminated aquifers in Bangladesh. Department ofGeology, Dhaka University for Bangladesh Water Develop-ment Board

Jakariya M, Choudhury M, Tareq MAH, Ahmed J (1998) BRAC:Village health workers can test tubewell water for arsenic.Bangladesh Rural Advancement Committee. Available at:http://wso.net/wei/dch/acic/infobank

JICA (1976) Feasibility study report for Jamuna River bridgeconstruction project. Volume VI: Geology and stone material.Japan International Co-operation Agency. Report for theJamuna Multipurpose Bridge Authority

Johnston R (1998) Arsenic contamination of groundwater inBangladesh: a quantitative secondary analysis of the NRECA/REB dataset. UNICEF, Dhaka

Khan FH (1991) Geology of Bangladesh. University Press, DhakaKudrass HR, Spiess V, Michels M, Kottke B, Khan SR (1999)

Transport processes, accumulation rates and a sediment budgetfor the submarine delta of the Ganges – Brahmaputra and theSwatch of No Ground, Bangladesh. International Seminar onthe Quaternary Development and Coastal Hydrodynamics ofthe Ganges Delta in Bangladesh, Dhaka, 20–21 September.Geological Survey of Bangladesh

750


Lovley DR (1987) Organic matter mineralization with the reductionof ferric iron: a review. Geomicrobiol J 5:375–399

Mallick S, Rajagopal NR (1996) Groundwater development in thearsenic-affected alluvial belt of West Bengal – Some Questions.Current Science 70(11):956–958

Mason B (1966) Principles of Geochemistry. Wiley, New YorkMather S (1999) The vertical and spatial variability of arsenic in the

groundwater of Chaumohani, Southeast Bangladesh. MScThesis (unpub), University College London

Mazumder DNG, Chakraborty AK, Ghose A, Gupta JD, Chakra-borty DP, Dey SB, Chattopadhay N (1988) Chronic arsenictoxicity from tubewell water in rural West Bengal. Bulletin ofthe World Health Organisation 66(4):499–506

Mazumder DNG, Das Gupta J, Chakraborty AK, Chatterjee A, DasD, Chakraborty D (1992) Environmental pollution and chronicarsenicosis in South Calcutta. Bulletin of the World HealthOrganisation 70(4):481–485

McArthur JM, Ravenscroft P, Safiullah S, Thirlwall MF (2001)Arsenic in groundwater: testing pollution mechanisms forsedimentary aquifers in Bangladesh. Water Resources Research37(1):109–117

McCarthy EMT (2001) Spatial and depth distribution of arsenic ingroundwater of the Bengal Basin. MSc Thesis (unpubl),University College London

Monsur H (1995) An introduction to the Quaternary geology ofBangladesh. International Geological Correlation ProgrammeIGCP-347, Rehana Akhter, Dhaka

Morgan JP, McIntire WG (1959) Quaternary geology of the BengalBasin, East Pakistan and India. Geol Soc Amer Bull 70:319–342

MPO (1987) Groundwater resources of Bangladesh. TechnicalReport Nr 5, Master Plan Organisation, Dhaka. Harza Engi-neering, USA, Mott MacDonald Ltd, UK, Meta Consultants,USA and EPC Ltd, Bangladesh

National Academy Press (2001) Arsenic in drinking water: 2001update. Sub-Committee to update the 1999 Arsenic in DrinkingWater Report, Goyer R (Chair). National Academy Press,Washington, DC

Nickson R (1997) Origin and distribution of arsenic in groundwa-ter, Central Bangladesh. MSc Thesis (unpub), UniversityCollege London

Nickson R, McArthur JM, Burgess WG, Ahmed KM, RavenscroftP, Rahman M (1998) Arsenic poisoning in Bangladesh ground-water. Nature 395:338

Nickson R, McArthur JM, Ravenscroft P, Burgess WG, AhmedKM (2000) Mechanism of arsenic release to groundwater,Bangladesh and West Bengal. Appl Geochem 15(4):403–413

NMIDP (1997) National Minor Irrigation Census 1995–96. Na-tional Minor Irrigation Development Project, Ministry ofAgriculture, Dhaka

Nordstrom DK (2000) An overview of arsenic mass poisoning inBangladesh and West Bengal, India. In: Courtenay Young (ed)Minor Elements 2000: Processing and Environmental Aspectsof As, Sb, Se, Te and Bi. Society for Mining, Metallurgy andExploration, Littleton, CO, pp 21–30

NRECA (1997) Study of the Impact of the Bangladesh RuralElectrification Program on Groundwater Quality. BangladeshRural Electrification Board. NRECA International with TheJohnson Company (USA) and ICDDR,B (Bangladesh)

Nriagu JO (ed) (1994) Arsenic in the environment. Wiley, NewYork

Perrin J (1998) Arsenic in groundwater at Meherpur, Bangladesh: avertical porewater profile and rock/water interactions. MScThesis (unpub), University College London

PHED (1991) National Drinking Water Mission Submission Projecton Arsenic pollution in groundwater in West Bengal. FinalReport, Compiled by Steering Committee, Arsenic Investiga-tion Project, Public Health Enegineering Department, Govern-ment of West Bengal, India 57 pp

Rashid H (1991) Geography of Bangladesh, 2nd edn. UniversityPress, Dhaka

Ravenscroft P (2001) Distribution of groundwater arsenic in theBangladesh related to geology. In: Bhattacharya P, Jacks G,Khan AA (eds) Groundwater arsenic contamination in theBengal Delta Plain of Bangladesh. Proc KTH-Dhaka UniversitySeminar, KTH Special Publication, TRITA-AMI report 3084,pp 4–56

Ravenscroft P (2003) An overview of the hydrogeology ofBangladesh. In: Rahman AA, Ravenscroft P (eds) Groundwaterresources and development in Bangladesh– background to thearsenic crisis, agricultural potential and the environment.Bangladesh Centre for Advanced Studies. University PressLtd, Dhaka

Ravenscroft P, McArthur JM, Hoque BA (2001) Geochemical andpalaeohydrological controls on pollution of groundwater byarsenic. Chappell WR, Abernathy CO, Calderon R (eds) Proc.4th Int Conf on Arsenic Exposure and Health Effects, SanDiego, June 2000. Elsevier, Oxford

Rus JS (1985) Geohydrological investigations in Khulna. DPHEWater Supply and Sanitation Projects. Department of PublicHealth Enegineering, Netherlands—Bangladesh DevelopmentCo-operation Programme

Safiullah S (1998) Report on monitoring and mitigation of arsenicin the ground water of Faridpur Municipality. CIDA ArsenicProject, Environmental Research Lab, Dept of Chemistry,Jahangirnagar University, Savar, Dhaka, 96 pp

Saha KC (1984) Melanokeratosis from arsenical contamination oftubewell water. Indian J Dermat 29:37–46

Saha KC (1995) Chronic arsenic dermatosis from tubewell water inWest Bengal during 1983–87. Indian J Dermat 40:1–12

Smith AH, Lingas EO, Rahman M (2000) Contamination ofdrinking water by arsenic in Bangladesh: a public healthemergency. Bulletin World Health Org 78(9):1093–1103

Umitsu M (1993) Late Quaternary sedimentary environments andlandforms in the Ganges Delta. Sed Geol 83:177–186

UNICEF (1994) Declining water levels study. Report prepared forthe United Nations Children’s Emergency Fund and Depart-ment of Public Health Engineering, Bangladesh by Engineeringand Planning Consultants, Dhaka, and Mott MacDonald Ltd,UK

UNICEF (1998) Progotir Pathey. UNICEF-Bangladesh, DhakaWadia DN (1975) Geology of India, 4th edn. MacMillan, London,

508 ppWelch AH, Lico MS, Hughes JL (1988) Arsenic in ground water of

the Western United States. Ground Water 26(3):333–347Whitney JW, Pavich MJ, Huq MA, Khorshed Alam AKM (1999)

The age and isolation of the Madhupur and Barind Tracts,Ganges – Brahmaputra Delta, Bangladesh. International Sem-inar on the Quaternary Development and Coastal Hydrody-namics of the Ganges Delta in Bangladesh, 20–21 September.Geological Survey of Bangladesh, Dhaka

WHO (1993) Environmental Health Criteria 18: Arsenic. WorldHealth Organization, Geneva

751


Arsenic in groundwater of the Bengal Basin, Bangladesh: Distribution, field relations, and...

Documents

Transcript of Arsenic in groundwater of the Bengal Basin, Bangladesh: Distribution, field relations, and...