Groundwater flow in an arsenic-contaminated aquifer, Mekong Delta, Cambodia

Applied Geochemistry 23 (2008) 3072–3087

Contents lists available at ScienceDirect

Applied Geochemistry

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

Groundwater flow in an arsenic-contaminated aquifer, Mekong Delta,

Cambodia

Shawn G. Benner a,*, Matthew L. Polizzotto b, Benjamin D. Kocar b, Somenath Ganguly a, Kongkea Phan c, Kagna Ouch c, Michael Sampson c, Scott Fendorf b

a Depart ment of Geo scienc es, Boise State Uni ver sity, Boise, ID 83705, USAb School of Earth Sci ences, Stan ford Uni ver sity, Stan ford, CA 94305, USAc Resource Devel op ment Inter na tional, Kien Svay, Kan dal, Cam bo dia

a r t i c l e i n f o

Article history:

Available online 4 July 2008

0883-2927/$ - see front matter © 2008 Else vier L

doi:10.1016/j.apgeochem.2008.06.013

* Cor re spond ing author.

E-mail address: sben ner@boise state.edu (S.G.

a b s t r a c t

To advance under stand ing of hydro log i cal influ ences on As con cen tra tions within ground-

wa ters of South east Asia, the flow sys tem of an As-rich aqui fer on the Me kong Delta in

Cam bo dia where flow pat terns have not been dis turbed by irri ga tion well pump ing was

exam ined. Mon i tor ing of water lev els in a net work of installed wells, extend ing over a

50 km2 area, indi cates that ground wa ter flow is dom i nated by sea son ally-var i able gra di-

ents devel oped between the river and the inland wet land basins. While the gra di ent inverts

annu ally, net ground wa ter flow is from the wet lands to the river. Hydrau lic param e ters of

the aqui fer (K t 10¡4 ms¡1) and over ly ing clay aqui tard (K t 10¡8 ms¡1) were deter mined

using grain size, per me am e ter and slug test anal y ses; when cou pled with observed gra di-

ents, they indi cate a net ground wa ter flow veloc ity of 0.04–0.4 ma¡1 down ward through

the clay and 1–13 ma¡1 hor i zon tally within the sand aqui fer, pro duc ing aqui fer res i dence

times on the order 100–1000 a. The results of numer i cal mod el ing sup port this con cep tual

model of the flow sys tem and, when inte grated with observed spa tial trends in dis solved

As con cen tra tions, reveal that the shal low sed i ments (upper 2–10 m of fine-grained mate-

rial) are an impor tant source of As to the under ly ing aqui fer.

© 2008 Else vier Ltd. All rights reserved.

1. Intro duc tion

There has been exten sive inves ti ga tion into the geo-

chem i cal pro cesses lead ing to ele vated As in the del taic

ground wa ters in Asia and there is emerg ing con sen sus

that reduc tive dis so lu tion of Fe oxides is an impor tant

mech a nism lead ing to As release to the pore wa ters of the

aqui fers in the region (Nick son et al., 2000; Har vey et al.,

2002; McAr thur et al., 2004; Islam et al., 2004). In con trast,

there has been less agree ment regard ing the both the role

of hydrol ogy in As release and the impor tance of sur fi cial

pro cesses as a source of As and/or organic C (see Har vey

et al 2002 and asso ci ated dis cus sion). Many stud ies pres-

td. All rights reserved.

Benner).

ent evi dence that As is released from sed i ments at depth

(e.g. McAr thur et al., 2004; Islam et al., 2004), while near-

sur face pro cesses have also been evoked (Po lizz ot to et al.,

2005; Har vey et al., 2006). Based on work at a field site

in Cam bo dia, the authors recently pro posed that sig nifi -

cant As is mobi lized from shal low sed i ments with release

and trans port strongly influ enced by sur face hydro logic

pro cesses and cou pled ground wa ter flow (Po lizz ot to et al.,

2008). A tight link age between As behav ior and hydrol ogy

implies that ongo ing and future anthro po genic activ i ties

(i.e. land use change, ground wa ter pump ing, sed i ment

exca va tion) will likely influ ence As behav ior, poten tially

alle vi at ing or exac er bat ing this mas sive health cri sis.

Elu ci dat ing the role of the hydro logic flow sys tem on

release (and trans port) has been ham pered by the flow

com plex ity of the heav ily pumped Ban glade shi aqui fers

where most research has been focused. Such local flow

http://www.sciencedirect.com/science/journal/08832927

www.elsevier.com/locate/apgeochem

mailto:[email protected]

S.G. Ben ner et al. / Applied Geochemistry 23 (2008) 3072–3087 3073

com plex ity has pre cluded the devel op ment of a site-spe-

cific, spa tially dis trib uted flow model that can enable eval-

u a tion of changes in geo chem i cal param e ters, inclu sive of

As con cen tra tions, along a con strained flow path. Despite

this com plex ity, a num ber of recent obser va tions indi cate

that hydro logic con di tions influ ence the result ing dis tri-

bu tion of dis solved As within the aqui fer (Har vey et al.,

2006; Klump et al., 2006; Post ma et al., 2007; Stute et al.,

2007).

Har vey et al. (2006) over came the chal lenges of a

spa tially dis trib uted mod el ing approach by uti liz ing a

detailed, lumped-param e ter model to dem on strate that

the aqui fer at their Mun sh i ganj site in Ban gla desh was rap-

idly flushed (res i dence times of <100 a), and that pump ing

for irri ga tion had altered the phys i cal and chem i cal nature

of the ground wa ter sys tem. Fur ther more, this eval u a tion

sug gested that ponds, which have been exca vated into the

sur fi cial clay layer at the site, may be a primary source of

recharge to the aqui fer. Klump et al. (2006), also work ing

in the Mun sh i ganj dis trict of Ban gla desh, cou pled 3H/3He

ground wa ter dat ing with con cep tual flow mod el ing to pro-

pose that the high est As con cen tra tions are found within a

zone of con ver gent flow pro duced by irri ga tion pump ing.

Sim i larly, Stute et al. (2007) sug gest, again based on 3H/3He

ground wa ter dat ing in Ban gla desh ground wa ter, that

ground wa ter flush ing (driven by pump ing) may explain

the spa tial dis tri bu tion of As in the aqui fer. At a sep a rate

site in Ban gla desh, van Geen et al. (2006) dem on strated a

cor re la tion between hydrau lic soil prop er ties and the con-

cen tra tion of As in the under ly ing aqui fer; areas with more

sand and less clay were cor re lated to lower As lev els. At

an inten sively instru mented field site on the Red River in

Viet nam, Post ma et al. (2007) doc u mented steep gra di ents

in As con cen tra tions along flow paths in the upper 10 m

of the aqui fer and detailed bio geo chem i cal changes along

inferred flow paths. In Po lizz ot to et al. (2008), the authors

pro posed a tightly cou pled hydro logic and geo chem i cal

model indi cat ing that ele vated As con cen tra tions within

the aqui fer are, at least in part, released from shal low sed-

i ments and trans ported to depth by infil trat ing sur face

water. While not exclu sion ary of other release pro cesses

and loca tions, the pro posed model pro vides a mech a nism

to explain As pro files through out the aqui fer. Here, a more

detailed descrip tion of the phys i cally-based hydro log i cal

frame work of ground wa ter flow that enabled this inter pre-

ta tion of observed spa tio tem po ral trends in ground wa ter

geo chem is try is pre sented. In a com pan ion sub mis sion,

the bio geo chem i cal pro files and mech a nisms lead ing to

As release from shal low sed i ments are described in detail

(Ko car et al., 2008).

The hydro logic com plex ity induced by ground wa ter

pump ing for irri ga tion has been over come by devel op ing

a model for a regional scale, As-rich, del taic aqui fer in

Cam bo dia where ground wa ter flow is min i mally dis turbed

by ground wa ter pump ing (Polya et al., 2005). Spe cifi cally,

an exten sive hydro logic data set, of regional extent, has

been cre ated and inter preted using sim ple mass bal ance/

flux cal cu la tions and numer i cal flow mod el ing, to develop

a site-spe cific, spa tially dis trib uted, model of the ground-

wa ter flow sys tem. The hy drog e ol o gy at the site is driven

by dra matic, sea son ally-var i able fluc tu a tions in sur face

water lev els and is strongly influ enced by a sur fi cial clay

depo si tional unit. It is dem on strated that, despite a sea-

sonal rever sal of flow direc tion, the net annual ground wa-

ter flow is down wards from wet land basins to the under-

ly ing aqui fer and hor i zon tal from the aqui fer to adja cent

riv ers.

1.1. Field site descrip tion

The 50 km2 field area is located on the Me kong Delta in

the Kan dal prov ince of Cam bo dia, SE of Phnom Penh (Lat.

11 31 3.90N, Long. 105 0 41.77E, Fig. 1). The Me kong River

is a broad river that car ries the sev enth larg est sed i ment

load in the world, the major ity of which is dis charged dur-

ing the wet sea son. The site is located at the upper end

of the delta and is bound by two del taic dis trib u tar ies of

the Me kong River (Me kong and Bas sac branches). As on

the other major del tas of Asia, Me kong topog ra phy is very

sub dued (max i mum ele va tion typ i cally <10 m) and is char-

ac ter ized by ele vated nat u ral levees along the riv er banks

reced ing to a native, sea son ally-flooded, wet land basin

between the two riv ers (Bus ch mann et al., 2007). The sed i-

ments of the Me kong Delta were depos ited over the pre vi-

ous 7.5 ka fol low ing the last sea level high-stand (Ngu yen

et al., 2000; Ta et al., 2002; Tam ura et al., 2007). At the site,

the sed i ment sequence is com posed of >30 m of aqui fer-

form ing sands over lain by approx i mately 15 m of silts and

clays (JICA, 2002; Tam ura et al., 2007). The aqui fer of the

low-lying Me kong flood plain in Cam bo dia is the primary

drink ing water source for much of the rural pop u la tion,

yet ele vated dis solved As con cen tra tions are com mon

(Polya et al., 2005; Berg et al., 2007). Within the aqui fer

at the field site, dis solved As con cen tra tions are nearly

uni ver sally ele vated with almost all wells within the aqui-

fer exceed ing 100 lg L¡1 (Fig. 2). Dur ing the study period,

ground wa ter pump ing through out the field site was lim-

ited to domes tic extrac tion.

2. Meth ods

2.1. Ground wa ter and sur face water mon i tor ing net work

A net work of 80 installed wells and 10 sur face water

mon i tor ing sites are dis trib uted through out the field area

(Fig. 1). At two loca tions addi tional near-sur face lysi me ters

and a sub set of closely spaced wells pro vide more detailed

spa tial anal y ses (see inten sive sites, Fig. 1). At most loca tions

outside the detailed mon i tor ing sites, wells are installed as

nested wells with one well screened across the water table

within the clay unit (4–12 m), and two wells screened in

the upper (20–30 m) and lower (36–60 m) por tions of the

aqui fer sands. Wells were installed using a locally devel-

oped man u ally-driven, direct rotary method (1.50, 3.8 cm,

diam e ter pipe with a 40, 10.2 cm, diam e ter open cut ting

tip) where a water-based slurry is pumped down the cen ter

pipe of the drill, lift ing drill cut tings to the sur face along

the annu lus between the drill pipe and drill hole. Wells are

com posed of 3.2 cm diam e ter PVC pip ing, typ i cally screened

over the lower 4 m. After instal la tion, wells were back filled

with sand over the screened inter val and capped with clay

or cement to the sur face.

3074 S.G. Ben ner et al. / Applied Geochemistry 23 (2008) 3072–3087

Mekong River

Wetland Basin

Bassac River

AbandonedRiver Channel

Ponds

0 1 2

kilometers

N

Well nest

Surface water

Core locations

Intensive sitesCross section

TC

TC51

TE51

TE11

MI11

BD11

PE51

PW31

BD51

PE31

PE21

PW11

TH11

TH71TC11

TC31

Mekong River

Wetland Basin

Bassac River

AbandonedRiver Channel

Ponds

0 1 2

kilometers

N

Well nest

Surface water

Core locations

Intensive sitesCross section

TC45

TC51

TE51

TE11

MI11

BD11

PE51

PW31

BD51

PE31

PE21

PW11

TH11

TH71TC11

TC31

Fig. 1. Field area map. Phnom Penh, the cap i tal of Cam bo dia, is located approx i mately 15 km to the west of the field area.

2.2. Water lev els

All ref er ence points, wells, and sur face water mon i tor-

ing sites used for water level mea sure ments were spa tially

linked by ver ti cal sur vey ing with an auto level (2.5 mm

accu racy per dou ble km run); field error was <5 mm on

closed loops.

Me kong River stage lev els were pro vided by the Me kong

River Com mis sion (MRC) and are rep re sen ta tive of the

recent his tor i cal aver age. Dis tance-weighted aver ages

from the Phnom Penh Port (upstream of the site) and the

Neak Lu ong (down stream) sta tions were used to cal i brate

the abso lute ele va tion of the water level mon i tor ing site in

the Me kong River, and ele va tions through out the field area

were ref er enced with these MRC data. Hydrau lic heads in

each well were mea sured weekly with an elec tronic water

level tape, and sur face water lev els were mea sured weekly

with a weighted mea sur ing tape from points of fixed ele-

va tion. Water lev els are reported as meters above mean

sea level (mASL).

m 0m 0001m 0002m 0003

0 m

-50 m

10 m

-20 m

-30 m

-40 m

0 m

-50 m

10 m

-20 m

-30 m

-40 m

-50 m

10 m

-20 m

0 m

-30 m

-40 m

-50 m

10 m

-20 m

0 m

-30 m

-40 m

-50 m

10 m

-20 m

0 m

-30 m

-40 m

wetlands ponds Mekong

240

10

100

400

650

0-7

80-120

390

540

600-750

900-1100

130

100

170-220

320-1200

0-30

300-550

920

1200

480

960

12-75

460670-850

450-930

950

920-1200

40-50

510-1030

770-1300

(0-100)

(100-400)

(400-850)(>850)

Fig. 2. Cross sec tional pro file along tran sect shown in Fig. 1, with As con cen tra tions, all val ues in lg L¡1. The num bers in paren the ses indi cate the range for

shaded con tours. The ver ti cal axis is expressed as meters above mean sea level (mASL).


2.3. Tidal mon i tor ing

Daily, tid ally-driven changes in water lev els in the

Me kong and within ground wa ter wells were mea sured

using In-Situ Level Troll 500 and 700 Series pressure

trans duc ers. A trans ducer, set to record at a rate of once

per min ute, was placed in the Me kong approx i mately 2 m

below the river sur face. A sec ond pressure trans ducer was

then sequen tially placed in wells TC11, TH11, and TC31 for

approx i mately 48 h time peri ods, pro duc ing a series of

river-aqui fer paired data sets.

2.4. Stra tig ra phy and sed i ment hydrau lic prop er ties

A cross-sec tional pro file of the site stra tig ra phy (Fig. 3)

was con structed by inter pre ta tion of well log data (visual

and tex tur ally-based doc u men ta tion of well cut tings

dur ing well instal la tion) and lim ited sed i ment core data

(Fig. 1) and was inter preted within the con straints of the

detailed sed i men tary his tory pro posed by Tam ura et al.

(2007). An intact cor ing pro ce dure was used at selected

loca tions for sed i ment retrieval. A 0.750 (1.9 cm) open core

device fit ted with a poly car bon ate sleeve and core-catcher

was deployed through the dril ling pipe at 3 m inter vals and

driven into the undis turbed sed i ment below the dril ling

tip. Once retrieved, sam ples were imme di ately capped and

sealed in O2-imper me able anaer o bic pouches to pre vent

oxi da tion of the sed i ments. Sed i ment sam ples were stored

and trans ported at 4 °C.

Par ti cle size sep a ra tion was per formed on 14 aqui fer

sand sam ples and the result ing grain size dis tri bu tions

were used to cal cu late hydrau lic con duc tiv ity val ues. The

Bre yer method was used to cal cu late hydrau lic con duc tiv i-

ties of the fine sand aqui fer sed i ments; the Ha zen method

was used to obtain K val ues for medium and coarse aqui fer

sand sam ples, (Kre sic, 1997).

Con stant head per me am e ter tests (Dome nic o and

Sch wartz, 1998) were per formed on 10 sam ples from the

upper clay layer (com posed of silts and clays), with depths

rang ing from 6 m to 18 m, in order to deter mine hydrau-

lic con duc tiv ity val ues. Sam ples were dried and repacked

into sealed flow cells, with lengths of 3.3 cm and cross-sec-

tional areas of 24 cm2. Con stant flow was main tained and

mon i tored. For each sam ple, the test was repeated at 3–6

dif fer ent pressure heads; for each trial at all depths at each

site, the range of cal cu lated K val ues was within a fac tor

of 2.

Slug tests were per formed on 15 shal low (8–12 m)

wells through out the field area and the result ing head

changes were ana lyzed accord ing to the Hvors lev method

(Hvors lev, 1951). The result ing K val ues for the fall ing head

and ris ing head tests were within a fac tor of 2 from each

other in all but one case. The results of hydrau lic con duc tiv-

ity deter mi na tion are sum ma rized in Table 1.

2.5. Water chem is try data col lec tion and plot ting

All mon i tor ing wells were sam pled once over the time

period Jan u ary 2005 to March 2006. Wells were sam-

pled and ana lyzed for As by induc tively cou pled plasma

opti cal emis sion spec trom e try using hydride gen er a tion.

Select wells were sam pled four times over the sea son,

TE 11TE 51TC 45TC 51

fine, coarse sands

with clay

clay/siltwith sand

-50 m

10 m

-20 m

0 m

-30 m

-40 m

-50 m

10 m

-20 m

0 m

-30 m

-40 m

wetlands ponds MekongMI 11

m 0m 0001m 0002m 0003

TransientRiver Boundary

Transient Wetland BoundaryNo flow

boundary

No flow boundary

No

flow

bou

ndar

y

Clay

Sand

Idealized Stratigraphic Cross Section

Idealized Model Domain

Fig. 3. Sed i men tary cross-sec tion of field area and ide al ized model domain (see Fig. 1 for loca tion). The ver ti cal axis is expressed as meters above mean sea

level (mASL). The model domain is 4000 m (400 cells) £ 60 m (20 cells).


DateNov-04 Mar-05 Jul-05 Nov-05 Mar-06

Ele

vatio

n A

bove

Sea

Lev

el (

m)

0

2

4

6

8Surface Water PondsMekong RiverWell TC31 Well TE11Well TC45 Well TC51

River > PondsPonds > River Ponds > River

Fig. 4. Sea sonal Me kong River, wet land and well hy dro graphs. Arrow indi cates dis tinct river hyd ro graph peak that is observed in ground wa ter wells and

cap tured in numer i cal mod el ing results.

estab lish ing that there was little var i a tion in con cen tra tion

within the aqui fer. The excep tion was observed tem po ral

vari a tions in the recharge and dis charge zones. Col lected

data from multiple tran sects was pro jected onto a sin gle

plane based on dis tance-from-the-river. The result ing two

dimen sional, cross-sec tional, plot of the data was then

con toured using visual inter po la tion.

2.6. Numer i cal mod el ing

Numer i cal mod el ing was under taken to fur ther eval u-

ate and refine (or revise) the emerg ing ground wa ter flow

con cep tu al i za tion. Model cal i bra tion was con ducted by

seek ing to repro duce the sea sonal trends in hydrau lic

heads observed in the mon i tor ing wells within the aqui-

fer. Aqui fer prop er ties were con strained by field-mea sured

or doc u mented lit er a ture val ues. It was also sought to

pro duce a net aver age flow field con sis tent with that dic-

tated by the mea sured annual gra di ents. In practice, this

entailed adjust ing the hydrau lic con duc tiv ity val ues of the

clay and the sand (within the con straints of the observed

range of val ues) to opti mize the rel a tive influ ence of the

sea sonal wet land and river hydrau lic head changes on the

aqui fer wells.

Numer i cal mod el ing was con ducted using MOD FLOW

and MOD PATH (McDon ald and Harb augh, 1988) under

both tran sient and steady-state (net annual) con di tions.

Sim u la tions were con ducted in a two-dimen sional cross-

sec tional grid (60 m, 20 cell depth; 4000 m, 400 cell width,

Fig. 3). An assump tion of sym me try between the Bas sac

and Me kong Riv ers enabled the estab lish ment of a no flow

bound ary at the cen ter of the wet lands and the mid dle of

the Me kong River. The base of the aqui fer (assumed to be

at 60 m depth) was also des ig nated a no flow bound ary

con di tion. Tran sient bound ary con di tions were defined by

a fit ted func tion of the 5 day run ning aver age of mea sured

water wet land/pond and river lev els over the time period

from Jan u ary 2005 to March 2006 (Fig. 4). These bound ary

con di tions were rep re sented by cells 12 m thick to accom-

mo date sea sonal head fluc tu a tions. The hydrau lic con duc-

tiv ity field was expressed as a sim pli fied 2-layer sys tem

with 13 m of clay over ly ing 36 m of sand.

Mate rial prop er ties were fixed based on field and lit-

er a ture val ues (Table 2) with the excep tion of hydrau lic

con duc tiv ity val ues, which were treated as the primary

vari ables dur ing tran sient model cal i bra tion. The range

of hydrau lic con duc tiv ity val ues was con strained by the

range of field-mea sured val ues and by the observed gra-

di ents across the clay (between the wet lands and the aqui-

fer) and across the sand (aqui fer to the river).

Model cal i bra tion was con ducted by adjust ing hydrau-

lic con duc tiv ity val ues for the clay and sand to repro duce

field-observed tem po ral trends (both sea sonal and daily-

tidal) in well water lev els at vary ing dis tances from the

Me kong River. The assump tion of two-dimen sion al ity (i.e.

all flow per pen dic u lar to the river) was eval u ated by com-

par ing the model results with observed trends along par al-

lel well tran sects.

The result ing hydrau lic con duc tiv ity val ues were used

in steady-state (annual aver age) flow sim u la tions in which

Table 1

Hydrau lic con duc tiv i ties for aqui fer sand and sur face clay sed i ments,

based on grain size anal y sis, con stant head per me am e ter test ing and slug

test anal y sis

Method Aqui fer sands Sur face clays

Grain size Per me am e ters Slug tests

Aver age K (ms¡1) 3.15E¡04 4.08E¡07 1.22E¡06

Stan dard

devi a tion (ms¡1)

3.02E¡04 5.94E¡07 1.89E¡06

Median K (ms¡1) 2.70E¡04 5.02E¡08 4.76E¡07

Num ber of sam ples 14 10 15


the wet land and river heads were fixed to pro duce the

observed aver age annual head dif fer ences.

3. Results and dis cus sion

3.1. Stra tig ra phy

Through out the field area, a silt-clay aqui tard over ly ing

the sand aqui fer was observed at all drill loca tions (Fig. 3).

While dril ling and cor ing tech niques did not per mit col lec-

tion of con tin u ous, undis turbed cores, the observed stra-

tig ra phy is con sis tent with the sed i men tary pro file and

his tory of the upper Me kong Delta pro posed by Tam ura et

al. (2007). The upper silt/clay unit ranges in thick ness from

5 to 22 m, is typ i cally orange col ored at the sur face, and is

green to grey at its base. Car bon dat ing of this fine-grained

sequence indi cates that the entire unit was depos ited over

the pre vi ous 7.5 ka; the upper, orange col ored sed i ments

were likely depos ited over the pre vi ous 0.7 ka (Tam ura,

2007). Below the upper clay/silt unit, a dis tinct tran si tion

to sand-rich sed i ments is observed. The aqui fer sand unit

extends from an aver age depth of 15 m to the max i mum

depth of dril ling (60 m), and this unit is char ac ter is ti cally

fine to course grained, is grey in color, and con tains clay

lenses. Near the base of the sand unit at some loca tions

(par tic u larly prox i mal to the Me kong and Bas sac Riv ers),

coarse grained sands and grav els are observed. Depth to

bed rock is, how ever, poorly con strained. Based on ground-

penetrating radar anal y ses, JICA (2002) esti mated that

con sol i dated bed rock ranges from 120 to 160 m below the

sur face through out the upper delta region (JICA, 2002),

but this has not been con firmed by geo phys ics nor drilled

wells in the field area. In fact, one JICA bore hole near the

Bas sac River and south of the field area reached bed rock

at 70 m (JICA, 2002). Addi tion ally, exca va tion pits near

Phnom Penh, »20 km NW of the field area, encoun ter bed-

rock at 40 m. Local well dril ling, which was used to install

the wells in this study, does not extend beyond 60 m depth;

through out the field area bed rock was reached only at

Well PE51 in the cen ter of the wet lands – at 40 m depth.

3.2. Sur face water trends and gra di ents

The delta-scale gra di ent, as expressed by the slope of

the Me kong River from Phnom Penh to the ocean, is very

low, approx i mately 1 £ 10¡5 mm¡1. In con trast, local, sea-

son ally-var i able dif fer ences in water lev els between the

river and adja cent ponds and wet lands are often 2–6 m, pro-

duc ing gra di ents on the order of 1 £ 10¡3 mm¡1. As is evi-

dent in Fig. 4, the sur face water lev els in the river pro duce

a sea sonal sig na ture that is sim i lar to, but dis tinct from,

that observed in the adja cent wet lands. The river begins

ris ing in May or June and peaks in Octo ber at an ele va tion

of approx i mately 8 m above sea level (mASL). Fol low ing a

steep decline in Decem ber and Jan u ary, the river con tin-

ues to decline to a low of approx i mately 1 mASL, until the

rainy sea son com mences in June and the river again begins

to rise. In con trast, wet land water lev els decline through

August, reach ing a low at approx i mately 4 mASL, but then

sharply rise when the river reaches flood stage in Sep tem-

ber and inun dates the wet lands. Fol low ing inun da tion, the

wet lands exhibit a rapid increase in water lev els, peak ing

in Octo ber to approx i mately 7 mASL (Fig. 4).

The dif fer ences in water lev els between the wet lands

and the river pro duce sig nifi cant, and sea son ally-var i able,

gra di ents. The river begins ris ing before the wet lands and,

from July through Novem ber, the river is 2–4 m higher

than the wet lands. In con trast, the river falls much more

rap idly than the wet lands and there is a period of approx-

i mately 8 months dur ing which the wet lands are 2–6 m

higher than the river.

3.3. Ground wa ter trends and gra di ents

Ground wa ter heads also exhibit dra matic sea sonal var-

i a tion; lev els in wells fluc tu ate up to 8 m annu ally (Fig.

4). In gen eral, ground wa ter lev els exhibit sea sonal trends

that are sim i lar, but of decreased ampli tude and delayed

trend arrival, to those observed in the river. In fact, sea-

sonal changes in ground wa ter lev els fall along a trend

that lies between those observed in the wet lands and the

river. Wells fur ther from the river exhibit less sea sonal var-

i a tion. For exam ple, Well TE11, which is 180 m from the

river, is almost iden ti cal to river trends, expe ri enc ing »8 m

of sea sonal change while Well TC31, which is 920 m from

the river, var ies by 7 m annu ally, and Well TC45, 1800 m

from the river, exhib its approx i mately 5 m of annual head

change. Com par ing wet land and aqui fer water lev els

indi cates there is a strong down ward gra di ent (2–4 m of

hydrau lic head dif fer ence) through out the fall ing limb of

the hyd ro graph. How ever, when the river is ris ing and

higher than the wet lands, that gra di ent is reversed and

when the river peaks, many wells across the site, par tic u-

larly those in lower-ele va tion basins, exhibit arte sian con-

di tions, indic a tive of strong upward gra di ents.

The direc tion of ground wa ter flow within the aqui fer,

as expressed by observed gra di ents, is per pen dic u lar to

the river (Fig. 5). How ever, ground wa ter gra di ents between

the aqui fer and the river invert semi-annu ally; dur ing the

ris ing limb of the hyd ro graph there is gra di ent of approx-

i mately 10¡3 mm¡1 from the river into the aqui fer, and

when the river is fall ing there is a gra di ent of sim i lar mag-

ni tude from the aqui fer towards the river (Figs. 5 and 6).

While the gra di ent var ies sea son ally, the wet lands are

higher than the river for 8 out of the 12 months of the year.

Because the water lev els in the river, wet lands and aqui fer

were mea sured over an entire sea sonal water cycle, it is

Table 2

Mate rial prop er ties used in model sim u la tions

River Wet land Sand Clay

Spe cific stor age (m¡1)a 0 5 £ 10¡4 1 £ 10¡4 5 £ 10¡4

Poros ityb 1 0.5 0.2 0.5

Model cal i brated

param e ters

Hor i zon tal K (ms¡1)c 1 1 £ 10¡8 3.5 £ 10¡4 1 £ 10¡8

Ver ti cal K (ms¡1)c 1 1 £ 10¡8 3.5 £ 10¡4 1 £ 10¡8

a Based on the lit er a ture val ues (Thang ara jan et al., 1999; BGS and

DPHE, 2001; Flec ken stein et al., 2006; Koni kow and Neu zil, 2007; Mast er-

son and Ga rab e dian, 2007). b Based on the lit er a ture val ues (Dome nic o and Sch wartz, 1998). c Derived from tran sient model cal i bra tion.


pos si ble to assess the net annual gra di ents between these

3 water bodies. By cal cu lat ing the time-weighted aver age

of the dif fer ence in water level between the wet lands and

under ly ing aqui fer, a net annual ver ti cal hydrau lic head

dif fer ence of approx i mately 1.2 m is obtained from the

wet lands down wards into the aqui fer. Sim i larly, com par-

ing sea sonal val ues for water lev els in the aqui fer with the

adja cent river indi cates a net annual hydrau lic head dif fer-

ence of approx i mately 0.2 m hor i zon tally from the aqui-

fer towards the river. A direct com par i son of the wet land

lev els to those observed in the river indi cates an aver age

annual head dif fer ence of 1.2–1.4 m. Impor tantly, these

dif fer ences in hydrau lic head are on the scale of deci me-

ters to meters, while the mea sure ment error asso ci ated

with the data col lec tion and ele va tion sur vey ing is <3 cm,

indi cat ing that there is a high degree of cer tainty asso ci-

ated with these net gra di ent cal cu la tions; the sur vey ing

error is less than 1% of the aver age annual head dif fer ence

between the wet lands and the river.

3.4. Aqui fer hydrau lic con duc tiv ity val ues

The mea sured hydrau lic con duc tiv ity val ues for the

clay range from 10¡8 to 10¡6 ms¡1 and median K val ues

range from 10¡8 and 10¡7 ms¡1, for per me am e ter and

slug tests, respec tively (Table 1). The mea sured hydrau lic

con duc tiv ity val ues for the sand unit range from 10¡4 to

10¡3 ms¡1 with a median value of 3 £ 10¡4 ms¡1 (Table 1).

These val ues are con sis tent with those reported in the lit-

er a ture for sim i lar sed i men tary mate ri als (Dome nic o and

Sch wartz, 1998) and are remark ably sim i lar to the val ues

reported for the clay aqui tard and sand aqui fers in Ban gla-

desh (3 £ 10¡8–9 £ 10¡8 ms¡1, and 4 £ 10¡4–9 £ 10¡4 ms¡1,

respec tively) (Rav ens cort et al., 2005).

3.5. Ground wa ter flux esti mates

Net annual fluxes can be esti mated by cou pling gra di-

ents with cal cu lated hydrau lic con duc tiv ity val ues using

Darcy9s Law. Assum ing an area of recharge of 25 km2

(i.e. roughly half the field site dis charges to each river)

and an aqui fer thick ness of 45 m, a net annual gra di ent

of »0.065 mm¡1, a clay hydrau lic con duc tiv ity value of

10¡8–10¡7 m s¡1, and a poros ity of 0.5, the net annual ver-

ti cal flow veloc ity is between 0.04 and 0.4 m per year and

the net annual flux down ward from the wet lands to the

aqui fer ranges from 106–107 m3a¡1. Inde pen dently, the flux

from the aqui fer to the river can also be cal cu lated. The net

annual gra di ent is approx i mately 8 £ 10¡5 mm¡1. Given

the aqui fer hydrau lic con duc tiv ity of 10¡4–10¡3 ms¡1, and

a poros ity of 0.2, net annual hor i zon tal ground wa ter flow

to the river ranges from 1.3 to 13 m and the net annual flux

from the aqui fer to the river is 105–106 m3a¡1. The con cep-

tual model dic tates that the water that recharges the aqui-

fer through the clay must, on an annual basis, equal the

amount of water exit ing the aqui fer to the river. Accord-

ingly, the actual flux through the clay is likely to be on the

lower end of the cal cu lated range while the aqui fer flux to

the river is on the upper end of the reported range to sat-

isfy con ser va tion of mass.

3.6. Net annual flow direc tion

Observed hydrau lic gra di ents cou pled with phys i cally

based, inter nally-con sis tent, flux cal cu la tions indi cate that

on an annual basis, water flows down wards from the wet-

lands into the under ly ing aqui fer and hor i zon tally from

the aqui fer to the river. While this would be con sid ered a

basic obser va tion in most hydro logic set tings, it is believed

that this rep re sents the first dem on stra tion of field mea-

sure ment-derived net flow direc tion for a regional aqui fer

not influ enced by ground wa ter pump ing on the del tas of

Asia. This observed flow direc tion is con sis tent with the

implied pre-pump ing base case used in mod el ing of the

Ban gla desh sys tem by BGS and DPHE (2001) and Klump

et al. (2006). In addi tion, hydrau lic data from a site in the

Mun sh i ganj dis trict of Ban gla desh illus trate sim i lar pat-

terns to those observed here except that the ini ti a tion

of dry sea son ground wa ter pump ing pro duces declines

in aqui fer water lev els below those of the river, alter ing

ground wa ter flow (Har vey et al., 2006).

3.7. Numer i cal model cal i bra tion

Within the tran sient, two-dimen sional ide al ized

flow model, a fit was obtained to the observed sea sonal

and daily-tidal trends in well lev els by assign ing hydrau-

lic con duc tiv ity val ues for the sand and clay lay ers of

3.5 £ 10¡4 ms¡1 and 1 £ 10¡8 ms¡1, respec tively (Figs. 7–9).

This solu tion is able to repro duce the gen eral sea sonal

trends in water lev els with increas ing dis tance from the

river and is in agree ment with the flux cal cu la tions above.

Sim u lated trends for wells imme di ately adja cent to the

river (e.g. TC11) closely match observed trends, with well

6.46

(7.89)

7.51

7.385.97

4.00

3.15

2.68

6.526.14

6.41

2.20

1.51

1.78

1.852.43

5.34

4.38

3.99

2.122.26

2.22

2.44

4.01

3.44

1.80

August 18, 2005 February 16, 20064 m 5 m

6 m

7 m

4 m

3 m

2 m

Mekong Mekong

(4.37)(3.99)

(3.93)(5.59)(5.67)

(5.66)

Fig. 5. Map view of hydrau lic heads in wells (open cir cles) and sur face water lev els (open tri an gles) on August 18, 2005 (wet sea son) and Feb ru ary 16, 2006

(dry sea son). Num bers indi cate heads in medium depth aqui fer wells (20–30 m), while those in paren the ses indi cate sur face water lev els. Units reported

in mASL. Dashed lines indi cate hand-con toured lines of equi po ten tial, with arrows indi cat ing the direc tion of the hor i zon tal ground wa ter gra di ent.


lev els show ing nearly iden ti cal pat terns to those in the

river. At loca tions fur ther from the river (TC31 and TH51),

the sim u lated trends match the approx i mate 1 m decline

observed in high water peak mag ni tude. At about a half of

the more dis tal wells (TC51, TC45, TH71), the model under-

pre dicts the degree of atten u a tion. The model results are

able to cap ture partial atten u a tion of the dis tinct spike in

July on the river hyd ro graph (see arrow in Fig. 4); fur ther

con firm ing the dom i nant role the river plays in dic tat ing

sea sonal trends in ground wa ter lev els.

The daily-scale tidal sim u la tions repro duce the gen eral

observed trends with sim i lar mag ni tudes in peak heights

near to the river and almost com plete atten u a tion at more

dis tal wells (TC31). Like the sea sonal trends, the tidal sim u-

la tions bet ter emu late trends closer to the river.

Impor tantly, the two-dimen sional tran sient model is

equally suc cess ful match ing well trends along multiple

tran sects (Fig. 8), with dis tance-from-the-river being suf -

cient to obtain sim u la tion results sim i lar to those observed

in the field, con firm ing the assump tion that flow is dom i-

nantly per pen dic u lar to the river. Sim i larly, the model is

able to repro duce trends in water lev els observed in wells

adja cent to the Bas sac (BD11), sup port ing the assump tion

of sym me try with respect to the zone between the two

riv ers.

Tran sient model results shown in spa tially dis trib uted,

cross-sec tional for mat, also exhibit gen eral agree ment

with observed trends (com pare Figs. 6 and 10). The model

cap tures the ground wa ter flow direc tion and the net gra-

di ent between the river and wet lands, but it is less suc-

cess ful at cap tur ing the tran sient spa tial dis tri bu tion of

the gra di ent; the model results for August 18th over es ti-

mate head-drop across the clay.

3.8. Net annual flow sim u la tion

Using the net annual head dif fer ences cal cu lated from

the field data and the hydrau lic con duc tiv ity val ues deter-

mined in the tran sient model, a sim u la tion of the net

annual flow field was obtained (Fig. 11). In this sim u la tion,

the observed rel a tive head loss across the clay and sand

is con sis tently repro duced. As illus trated by the sim u lated

flow lines, water flows ver ti cally down wards through the

clay and nearly hor i zon tally through the aqui fer to the

river. Note that the sim u lated results are pre sented with

approx i mate ver ti cal exag ger a tion of 16 times, which visu-

ally empha sizes the lim ited ver ti cal com po nent of flow in

the sand. The flow veloc i ties and travel times along the

flow lines within the sim u la tion are shown in Tables 3 and

4. Sim u lated travel times through the clay are on the order

of 250 a, while travel times across the aqui fer vary from

600 a for the most dis tal flow lines to less than 140 a for

those close to the river. There fore, total travel times from

the wet land to the river range from about 400–900 a, val-

ues con sis tent with those from mass flux cal cu la tions.

These flow veloc i ties and res i dence times are remark ably

-50 m

10 m

-20 m

0 m

-30 m

-40 m

3000 m 2000 m 1000 m 0 m

3000 m 2000 m 1000 m 0 m

TC

21

TC

11

TC

31

TC

51

TC

45

2.69

2.75

4.9

4.95

6.145.89

5.216.43

6.543.15

5 m

4 m3m 6 m

Mekong7.89 mAugust 18th

2006

7 m

Wetland 4.00 m

-50 m

10 m

-20 m

0 m

-30 m

-40 m

TC

21

TC

11

TC

31

TC

51

TC

45

3 m

Mekong`1.50 m

2 m

Wetland 5.60 m

3.99

3.99

5.63

2.20

2.20

2.20

3.02

1.80

2.265.42

2.124.385 m

February 16th

2006

4 m

Fig. 6. Cross-sec tional view (along tran sect depicted in Fig. 1) of hydrau lic heads in wells and sur face waters on August 18, 2005 and Feb ru ary 16, 2006,

illus trat ing sea sonal shifts in gra di ents. Open cir cles indi cate well screen loca tions, and asso ci ated head val ues are in mASL. Dashed lines indi cate hand-

con toured lines of equal potential. Note that despite the ver ti cal exag ger a tion, cross-sec tions are to scale.


sim i lar to those esti mated by ide al ized flow mod el ing

in Ban gla desh of 300 a to travel over 1000 m under pre-

pump ing con di tions (BGS and DPHE 2001).

3.9. Model uncer tainty and util ity

Given the ide al ity of the numer i cal model, it pro duces

a remark ably good fit to observed sea sonal and daily-tidal

trends. Impor tantly, how ever, the mod el ing results remain

under-con strained. In par tic u lar, the uncer tainty regard-

ing aqui fer stor ativ i ty val ues, which strongly influ ence

the tran sient sim u la tions, pre clude the use of the flow

mod el ing to mark edly improve esti mates of aqui fer res-

i dence times over those cal cu lated using flux esti mates.

The sim u lated flow lines are, how ever, inde pen dent of

stor ativ i ty val ues and are pri mar ily dic tated by the ratio

of clay to sand hydrau lic con duc tiv i ties (a ratio well-con-

strained within the sim u la tions). There fore, the flow lines

plot ted in Fig. 11 pro vide a sound esti mate of the over all,

generalized net sea sonal flow direc tion.

It is par tic u larly note wor thy that flow lines within the

clay are nearly ver ti cal, while flow lines within the aqui-

fer are nearly hor i zon tal. This flow geom e try sug gests that

track ing changes in chem is try along a ver ti cal pro file will

be most effec tive within the upper 5–20 m of clay at this

site. As those flow lines enter the under ly ing aqui fer and

are refracted to the hor i zon tal, the uncer tainty regard-

ing indi vid ual flow paths increases. There fore, within the

Transient WetlandBoundary

Transient RiverBoundary

No flow Boundary

No

Flo

w B

ound

ary

No F

low B

oundary

60 m

TC 11TC 21TC 21TC 45TC 45TC 51TC 51 TC 31TC 31

ClaySand

4000 m

TC 11

Time (d)50 150 250 350 450

Time (d)50 150 250 350 450

Time (d)50 150 250 350 450

Time (d)50 150 250 350 450

Time (d)50 150 250 350 450

Hyd

raul

ic H

ead

(m)

-2

0

2

4

6

8

10

Hyd

raul

ic H

ead

(m)

-2

0

2

4

6

8

10

Hyd

raul

ic H

ead

(m)

-2

0

2

4

6

8

10

Hyd

raul

ic H

ead

(m)

-2

0

2

4

6

8

10H

ydra

ulic

Hea

d (m

)

-2

0

2

4

6

8

10

ObservedSimulated

TC 21

ObservedSimulated

TC 31

ObservedSimulated

TC 45

ObservedSimulated

TC 51

ObservedSimulated

Fig. 7. Model cal i bra tion results to the TC well tran sect.


deeper aqui fer it is more appro pri ate to exam ine aqui fer-

scale trends in chem i cal param e ters rather than attempt-

ing to track changes along indi vid ual flow lines.

Uncer tainty regard ing aqui fer thick ness also intro duces

uncer tainty to model results. The base of the aqui fer is esti-

mated to be at 60 m and a num ber of wells at the site were

drilled to depths near that range. There fore, there is a high

degree of con fi dence the aqui fer is not sig nifi cantly more

shal low. The aqui fer may, how ever, be thicker than mod-

eled. Depth to bed rock across the region may be as deep as

160 m. If those depths exist at this site, the primary impact

on model results would be an increase in the aqui fer res-

i dence time and/or an increase in the model-esti mated

aqui fer hydrau lic con duc tiv ity.

While the highly ide al ized approach used to develop

this con cep tual model has proven effec tive at the regional

scale, the local influ ences of hydrau lic het er o ge ne ity

should not be dis counted. Aqui fer sed i ments are often

highly het er o ge neous, and that het er o ge ne ity will often

pro duce com plex vari a tions in spe cific flow line paths,

vari a tions that are not cap tured by this con cep tual model.

The potential impact of such het er o ge ne ity (within the

clay) is dis cussed below.

3.10. Potential influ ence of hydrau lic het er o ge ne ity

For wells TC45, TC51 and TH71, the model over-pre dicts

draw down dur ing the fall ing limb of the hyd ro graph, sug-

Transient WetlandBoundary


No flow Boundary

No

Flo

w B

ound

ary

No F

low B

oundary

60 m

TH11BD11TH51 TH71 TH31

Clay

Sand

4000 m

TH 11

Time (d)50 150 250 350 450

Time (d)50 150 250 350 450

Time (d)50 150 250 350 450

Time (d)50 150 250 350 450

Time (d)50 150 250 350 450 550

Hyd

raul

ic H

ead

(m)

-2

0

2

4

6

8

10

Hyd

raul

ic H

ead

(m)

-2

0

2

4

6

8

10H

ydra

ulic

Hea

d (m

)

-2

0

2

4

6

8

10

Hyd

raul

ic H

ead

(m)

-2

0

2

4

6

8

10

Hyd

raul

ic H

ead

(m)

-2

0

2

4

6

8

10

ObservedSimulated

TH 31

ObservedSimulated

TH 51

ObservedSimulated

TH 71

ObservedSimulated

BD 11

ObservedSimulated

Fig. 8. Model cal i bra tion results to the TH and BD well tran sects.


gest ing the lim i ta tions of the ide al ized model for mu la tion.

For TC45, there is also a poor match dur ing and after the

hyd ro graph peak. Due to appre cia ble logis ti cal dif culty in

col lect ing data dur ing flooded con di tions, later time series

data for TC51 and TH71 are quite noisy; how ever at all three

of these sites, which are located prox i mal to one another,

the wells exhibit a sea sonal trend that more closely fol lows

that observed in the over ly ing wet lands (Fig. 3), sug gest-

ing increased influ ence of over ly ing sur face water at these

sites. To eval u ate this pos si bil ity, a model sim u la tion was

per formed in which the clay layer hydrau lic con duc tiv ity

prox i mal to these wells was increased to 6 £ 10¡7 ms¡1, a

change that sim u lates either a thin ning of the clay or an

increase in the hydrau lic con duc tiv ity at this loca tion (clay

win dow, Figs. 12 and 13). This sim u la tion pro duces results

that improve the fit with the observed data in these three

wells, indi cat ing that spa tial vari a tions in the clay thick-

ness or hydrau lic con duc tiv ity may explain sea sonal trends

in these wells and may be an impor tant fea ture of delta

ground wa ter flow sys tems, a trend con sis tent with Har-

vey et al. (2006). Impor tantly, these sim u la tions indi cate

that the local ized zone of increased recharge influ ences

wells some dis tance away (TC51 is almost 1000 m inland

of the sim u lated high con duc tiv ity clay win dow), with a

greater rel a tive influ ence on inland wells. This sim u la tion

also high lights how pref er en tial recharge can have a dis-

Fixed Head WetlandBoundary


No flow Boundary

No

Flo

w B

ound

ary

No F

low B

oundary60 m

TH11

Clay

Sand

4000 m

TC 11TC 31TC 31

TC 11

Time (min)

0 1000 2000 3000

Time (min)

0 1000 2000 3000

Time (min)

0 1000 2000 3000

Hyd

raul

ic H

ead

(m)

4.1

4.2

4.3

4.4

4.5

Hyd

raul

ic H

ead

(m)

4.1

4.2

4.3

4.4

4.5

Hyd

raul

ic H

ead

(m)

4.1

4.2

4.3

4.4

4.5

MekongObserved WellSimulated Well

TH 11 MekongObserved WellSimulated Well

TC 31 MekongObserved WellSimulated Well

Fig. 9. Model cal i bra tion results to tidal fluc tu a tions at TC11, TH11 and TC31. Each sim u la tion was per formed with the plot ted Me kong data as a tran sient

bound ary con di tion and a fixed head wet land bound ary. For clar ity, all data was nor mal ized to the same ini tial hydrau lic head value of approx i mately

4.3 m.


pro por tion ate influ ence on flow in the sub sur face; much

of the aqui fer water in this sim u la tion ema nates from the

‘clay win dow’. The pres ence of a clay win dow also reduces

aver age aqui fer res i dence times.

3.11. Link ing chem is try and net annual flow

Map ping the sim u lated net annual flow field on As con-

cen tra tions across the aqui fer (Fig. 14) pro vides insights on

the flow-chem is try rela tion ships, although cau tion should

River Boundary7.93 m

Wetland Boundary 3.95 m

No flow

boundary

No flow boundary

No

flow

bou

nda

ry

4 m

7 m6 m

5 m

River Boundary1.51 m

Wetland Boundary 5.41 m

No flow

boundary

No flow boundary

No

flow

bou

ndar

y

2 m3 m

4 m5 m

August 18th

February 16th

Fig. 10. Sim u lated hydrau lic head dis tri bu tion for August 18th, 2005 and Feb ru ary 16th, 2006. Com pare to observed field val ues in Fig. 6.

No flow

boundary

No flow boundary

No

flow

bou

ndar

y

Wetland Boundary (5.5 m MASL) River Boundary(4.0 m MASL)

Sand K=3.5 x 10-4 m sec-1

Clay K=1.0 x 10-8 m sec-1

a b c d e f

Fig. 11. Sim u lated flow field for net annual flow. Sim u lated travel times along each flow line are shown in Table 4.

Table 3

Net annual ver ti cal and hor i zon tal ground wa ter flow veloc i ties

Ver ti cal flow Hor i zon tal flow

Min i mum (K = 10¡8 ms¡1) Max i mum (K = 10¡7 ms¡1) Min i mum (K = 10¡4 ms¡1) Max i mum (K = 10¡3 ms¡1)

Aver age net flow veloc ity

(ma¡1)

0.04 0.43 1.3 13

Stan dard devi a tion (m) 0.003 0.034 0.87 8.8

Num ber of cal cu la tions 3 3 5 5

Table 4

Sim u lated travel times for flow paths in Fig. 11

Flow paths Travel time (a)

Clay Sand Total

a 285 604 889

b 285 431 716

c 281 307 588

d 272 215 487

e 261 136 398

f 248 70 319


be used given the ide al ized nature of the flow sim u la tions.

Along nearly all sim u lated flow paths there is an increase

in As con cen tra tions with dis tance (age). The steep est gra-

di ents in As con cen tra tions are, how ever, observed along

the ver ti cal flow lines down wards across the sur fi cial clay

sed i ments (con cen tra tions increase from <1 to 500 lg L¡1

over a dis tance of 10 m), and the high est As con cen tra tions

within the shal low pore water appear to ema nate from

aban doned chan nel pond wet lands. Detailed pro files fur-

ther dem on strat ing gra di ents in As con cen tra tions across

the first 10 m of the flow path are pro vided and dis cussed in

detail in Ko car et al. (2008). Increases in As con cen tra tions

along ver ti cal pro files within the clay can be inter preted

to reflect increases in As con cen tra tions along a flow path.

In con trast, within the aqui fer, where flow is pri mar ily

hor i zon tal, ver ti cal sam pling pro files will cap ture waters

trav el ing along dis pa rate flow paths with poten tially very

dif fer ent ori gins and geo chem i cal his to ries.

Given the esti mated aqui fer res i dence times, the deeper

portion of the aqui fer con tains water that may be up to

1 ka and may reflect pro cesses that occurred at some dis-

tal loca tion under con di tions that may no longer be oper a-

TC 11

Time (d)50 150 250 350 450

Time (d)50 150 250 350 450

Time (d)50 150 250 350 450

Time (d)50 150 250 350 450

Time (d)50 150 250 350 450

ObservedSimulation ASimulation B

TC 21


TH 71


TC 45


TC 51


Hyd

raul

ic H

ead

(m)

-2

0

2

4

6

8

10

Hyd

raul

ic H

ead

(m)

-2

0

2

4

6

8

10

Hyd

raul

ic H

ead

(m)

-2

0

2

4

6

8

10

Hyd

raul

ic H

ead

(m)

-2

0

2

4

6

8

10H

ydra

ulic

Hea

d (m

)

-2

0

2

4

6

8

10

Fig. 12. Sim u la tions with ‘clay win dow’ (dashed) and with out (solid line) com pared to observed trends (dia mond sym bols).

No flow

boundary

No flow boundary

No

flow

bou

ndar

y

Wetland Boundary (5.5 m MASL) River Boundary(4.0 m MASL)

Sand K=3.5 x 10-4 m sec-1

Clay K=1.0 x 10-8 m sec-1

Fig. 13. Sim u lated flow field for net annual flow with ‘clay win dow’ (ver ti cal hashed area, K = 6 £ 10¡7 ms¡1). Diag o nal hashed area rep re sents water enter-

ing the aqui fer through the ‘clay win dow’.


tional today. There fore, appli ca tion of the model to trends

in the deeper (and older) flow sys tem evokes sig nifi cant

uncer tainty. For exam ple, at the base of the aqui fer near

the mid dle of the pro file, water con tain ing highly ele vated

As con cen tra tions (>1000 lg L¡1) is observed. Accord ing

to the con cep tual model, this water entered that aqui fer

as recharge in the wet lands 1000 s of meters inland and

100 s of years ago and release from the sed i ments may

have occurred at any loca tion along that flow path. These

obser va tions sug gest that the great est cer tainty regard ing

the link ages between flow paths and pro cesses lead ing to

As release from sed i ments exists in the youn gest waters

flow ing down wards through the sur fi cial clay sed i ments.

Tem po ral mon i tor ing of ground wa ter As con cen tra-

tions indi cates that sea sonal var i a tion is only observed in

the zone of net recharge (shal low wells in the clay) and the

zone of net dis charge (at the bound ary with the river). In

the zone of recharge, the high est As con cen tra tions were

observed dur ing strong down ward flow, while near the

river, As con cen tra tions decline when river lev els are ris-

ing and river water is flow ing into the aqui fer (Po lizz ot to

et al., 2008). These trends are con sis tent with the pro posed

flow model and are indic a tive of rel a tively con ser va tive As

trans port within the aqui fer.

Het er o ge ne ity in the dis tri bu tion of the sur fi cial clay

would have a poten tially com plex influ ence on As dis tri bu-

tion in the aqui fer. Assum ing As release is pri mar ily from

near-sur face sed i ments, and if the release of As to the pore

water is rate-lim ited rel a tive to the change in infil tra tion

veloc ity, a clay win dow could act to dilute aqui fer waters

(as pro posed by van Geen et al., 2006). Alter na tively, if the

infil trat ing waters also pro mote higher rates of As release

(e.g. by pro vid ing a source of labile organic C), As con cen-

tra tions in the infil trat ing waters from the clay win dow

would remain high and could be the dom i nant source of

As to the aqui fer.

3.12. Com par i son with obser va tions at other sites

Locally unique char ac ter is tics indi cate that care

should be taken when attempt ing to com pare or con trast

results from dif fer ent research sites across the del tas

of Asia. How ever, As-rich del taic ground wa ters of Asia

share sim i lar geo logic his tory, topog ra phy, gross stra-

tig ra phy and dom i nant nat u ral hydro logic driv ers. All of

these aqui fers are located on mas sive del taic sequences

depos ited over the last 10 ka. As del taic sys tems, aqui fer

mate ri als were sim i larly depos ited and are com posed of

sim i lar grain size dis tri bu tions. At most sites through out

the del tas of Asia the aqui fer mate ri als are char ac ter-

ized as fine to medium grained sands with inter-bed ded

clays; most aqui fers are capped by fine-grained silts and

clays of vary ing thick ness. All these del taic aqui fer sys-

tems are located beneath sub dued, and often inverted,

topog ra phy. Levees adja cent to the riv ers are typ i cally the

ele va tion high point with inland wet land basins extend-

ing outward from the river levee sys tem. The sub dued

topog ra phy, cou pled with the low gra di ent of the river

sys tem, pro duces very low regional hydrau lic gra di ents

dic tated by the ele va tion of the land sur face rel a tive to

the dis tant ocean. In con trast, all the riv ers of the Asia

del tas undergo large sea sonal vari a tions in water lev els

(5–10 m) and, in the absence of sig nifi cant ground wa ter

extrac tion, these sea sonal fluc tu a tions strongly influ-

ence ground wa ter flow.

Direct quan ti ta tive com par i sons of hydrol ogy between

del tas are lim ited. How ever, hydrau lic con duc tiv ity val ues

for both the sand and clay at this site are remark ably sim i-

lar to those observed in Ban gla desh (BGS and DPHE, 2001).

The ide al ized flow-mod eled veloc ity esti mates for pre-

pump ing con di tions in Ban gla desh, are nearly iden ti cal to

those obtained for the pres ent field area (BGS and DPHE,

2001; Har vey et al., 2006). Fur ther more, recent obser va-

tions on the Red River, where the over ly ing clay layer is

also thick, sim i larly exhibit increas ing As con cen tra tions

over the clay pro file (Berg et al., 2008).

Given these sim i lar i ties, the first-order hydro logic

char ac ter is tics observed on the upper Me kong Delta may

pro vide a basis for under stand ing con di tions on other del-

tas of Asia prior to anthro po genic dis rup tion of the flow

sys tems. In par tic u lar, the ini tial con di tion of ground wa-

ter flow from wet lands towards the river may prove to

be a use ful assump tion when assess ing the cur rent com-

plex, irri ga tion-influ enced, flow sys tems observed at

many sites.

An impor tant caveat is that, while the hydro logic

driv ers of ris ing and fall ing riv ers cou pled with sea sonal

flood ing of wet lands are sim i lar, the amount of sur fi cial

clay across the Asian del tas var ies dra mat i cally, with some

sites in Ban gla desh report ing <1 m of clay com pared to

m 0m 0001m 0002m 0003

0 m

-50 m

10 m

-20 m

-30 m

-40 m

0 m

-50 m

10 m

-20 m

-30 m

-40 m

-50 m

10 m

-20 m

0 m

-30 m

-40 m

-50 m

10 m

-20 m

0 m

-30 m

-40 m

-50 m

10 m

-20 m

0 m

-30 m

-40 m

wetlands ponds Mekong(0-100)

(100-400)

(400-850) (>850)

Fig. 14. Sim u lated flow field for net annual flow (with out clay win dow) over lain with As con cen tra tions. Arsenic con cen tra tions are shown as shaded con-

tours, and val ues in paren the ses indi cate range of inter vals in lg L¡1.


up to 20 m on the Me kong Delta. Given the sen si tiv ity of

the flow sys tem to the per me abil ity (or thick ness) of the

clay aqui tard, this may be a key param e ter that dif fer en-

ti ates flow behav ior at sites across the del tas of Asia. Clay

thick ness has a par tic u larly strong influ ence on aqui fer res-

i dence time and the degree of ver ti cal flow. In addi tion,

spa tial het er o ge ne ity in clay thick ness, likely to be more

dra matic when the clay unit is thin ner, will pro duce pref-

er en tial infil tra tion, a phe nom e non doc u mented to impact

As dis tri bu tion in Ban gla desh where the upper clay layer

is some times absent (van Geen et al., 2006; Stute et al.

2007).

4. Sum mary and con clu sions

The pre sented regional-scale hydro logic model has

been cal i brated against an exten sive field-col lected data-

set and pro vides an inter nally-con sis tent model of water

flow within the field area. The model con strains the net

annual flow direc tion and indi cates water flows down-

wards from wet lands through the clay aqui tard and hor i-

zon tally to dis charge at the river. Impor tantly, this model

pro vides a phys i cal frame work with which to inter pret

spa tially dis trib uted ground wa ter chem i cal data in the

con text of evolv ing com po si tion over time and dis tance.

Accord ingly, observed increases in As along flow paths at

this site are great est within the ver ti cal zone of recharge,

indi cat ing the impor tance of the shal low sed i ments as a

source of As to the aqui fer.

In a com pan ion pub li ca tion (Ko car et al., 2008), it is pro-

posed that aban doned chan nel ponds may be par tic u larly

impor tant sources of bio geo chem i cal As release. Indeed,

As con cen tra tion pro files align with mod eled flow lines

and the high est As con cen tra tions through out the aqui fer

are found along flow lines ema nat ing from the ponds. The

hydrau lic behav ior of wells TC45, TC51 and TH71, all prox-

i mal to but inland from the aban doned chan nel ponds,

sug gests the ponds may also pro duce zones of pref er en tial

recharge, poten tially due to the pres ence of sur fi cial depos-

its that are thin ner or more con duc tive. Such a unique

com bi na tion of bio geo chem i cal and phys i cal-hydro logic

prop er ties of aban doned chan nel ponds may con spire to

pro duce both high rates of As release and high rates of

recharge, poten tially deliv er ing much of the As observed

in the sub sur face.

Acknowl edge ments

This work was sup ported by the EVP pro gram of Stan-

ford’s Woods Insti tute for the Envi ron ment, two US. EPA

STAR grad u ate fel low ships, the Stan ford NSF Envi ron men-

tal Molec u lar Sci ences Insti tute (NSF-CHE-0431425), and

the Boise State Uni ver sity New Fac ulty Devel op ment

Fund. We appre ci ate the assis tance from Tom Clemo

(mod el ing), Monty Bus bee (sed i ment cor ing), Molly

Meyer (sur vey ing), Rebecca Neu mann (field assis tance),

Amna Aziz (hydrau lic con duc tiv ity deter mi na tion) and

the RDI staff (data col lec tion). We would like to extend

a thank you to our edi tor (David Polya) and review ers

(John Tel lam, Wil liam Bur gess, anon y mous) for their

help ful com ments.

Ref er ences

Berg, M., Sten gel, C., Trang, P.T.K., Viet, P.H., Samp son, M.L., Leng, M., Sam-reth, S., Fred er icks, D., 2007. Mag ni tude of arsenic pol lu tion in the Me kong and Red River Del tas – Cam bo dia and Viet nam. Sci. Total Envi-ron. 372, 413–425.

Berg, M., Trang, P.T.K., Sten gel, C., Bus ch mann, J., Viet, P.H., Van Dan, N., Giger, W., Stü ben, D., 2008. Hydro log i cal and sed i men tary con trols lead ing to arsenic con tam i na tion of ground wa ter in the Ha noi area, Viet nam: the impact of iron–arsenic ratios, peat, river bank depos its, and exces sive ground wa ter abstrac tion. Chem. Geol. 249, 91–112.

Brit ish Geo logic Sur vey and Depart ment of Pub lic Health Engi neer ing (BGS and DPHE), 2001. Arsenic Con tam i na tion of Ground wa ter in Ban gla desh, Key worth, UK.

Bus ch mann, J., Berg, M., Sten gel, C., Samp son, M.L., 2007. Arsenic and man-ga nese con tam i na tion of drink ing water resources in Cam bo dia: coin-ci dence of risk areas with low relief topog ra phy. Envi ron. Sci. Tech nol. 41, 2146–2152.

Dome nic o, P.A., Sch wartz, F.W., 1998. Phys i cal and Chem i cal Hy drog e ol-o gy. John Wiley, New York.

Flec ken stein, J.H., Nis won ger, R.G., Fogg, G.E., 2006. River-aqui fer inter-

ac tions, geo logic het er o ge ne ity, and low-flow man age ment. Ground Water 44, 837–852.

Har vey, C.F., Swartz, C.H., Bad ruzz aman, A.B.M., Keon-Blute, N., Yu, W., Ali, M.A., Jay, J., Bec kie, J.R., Nie dan, V., Brab an der, D., Oates, P.M., Ashf aque, K.N., Islam, S., He mond, H.F., Ahmed, M.F., 2002. Arsenic mobil ity and ground wa ter extrac tion in Ban gla desh. Sci ence 298, 1602–1606.

Har vey, C.F., Ashf aque, K.N., Yu, W., Bad ruzz aman, A.B.M., Ali, M.A., Oates, P.M., Michael, H.A., Neu mann, R.B., Bec kie, R., Islam, S., Ahmed, M.F., 2006. Ground wa ter dynam ics and arsenic con tam i na tion in Ban gla-desh. Chem. Geol. 228, 112–136.

Hvors lev, M.J., 1951. Time Lag and Soil Per me abil ity in Ground wa ter Obser va tions. US Army Corps of Engi neers Water way Exper i men tal Sta tion, Vicks burg, Miss.

Islam, F.S., Ga ult, A.G., Booth man, C., Polya, D.A., Char nock, J.M., Chat ter-jee, D., Lloyd, J.R., 2004. Role of metal-reduc ing bac te ria in arsenic release from Ben gal delta sed i ments. Nature 430, 68–71.

JICA, 2002. Assess ment of Ground wa ter Resources in Cam bo dia, Koku sai Ko gyo Co., Ltd.

Klump, S., Kip fer, R., Cir pka, O.A., Har vey, C.F., Brenn wald, M.S., Ashf aque, K.N., Bad ruzz aman, A.B.M., Hug, S.J., Im bo den, D.M., 2006. Ground wa-ter dynam ics and arsenic mobi li za tion in Ban gla desh assessed using noble gases and tri tium. Envi ron. Sci. Tech nol. 40, 243–250.

Ko car, B.D., Po lizz ot to, M.L., Ben ner, S.G., Samp son, M., Fen dorf, S., 2008. Bio geo chem i cal and depo si tional con trols on arsenic mobil ity within sed i ments of the Me kong Delta. Appl. Geo chem. (This vol ume).

Koni kow, L.F., Neu zil, C.E., 2007. A method to esti mate ground wa ter deple tion from con fin ing lay ers. Water Re sour. Res. 43, doi:10.1029/2006WR00559.

Kre sic, N., 1997. Quan ti ta tive Solu tions in Hy drog e ol o gy and Ground wa-ter Mod el ing. CRC, Boca Raton.

Mast er son, J.P., Ga rab e dian, S.P., 2007. Effects of sea-level rise on ground-wa ter flow in a coastal aqui fer sys tem. Ground Water 45, 209–217.

McDon ald, M.G., Harb augh, A.W., 1988. A mod u lar three-dimen sional finite-dif fer ence ground wa ter flow model. US Geol. Surv. Tech. Water-Re sour. Invest. 34 (Book 6), 586, (Chap ter A1).

McAr thur, J.M., Baner jee, D.M., Hud son-Edwards, K.A., Mish ra, R., Puro hit, R., Rav ens croft, P., Cro nin, A., Ho warth, R.J., Chat ter jee, A., Taluk der, T., Lowry, D., Hough ton, S., Cha dha, D.K., 2004. Nat u ral organic mat ter in sed i men tary basins and its rela tion to arsenic in anoxic ground wa ter: the exam ple of West Ben gal and its world wide impli ca tions. Appl. Geo chem. 19, 1255–1293.

Ngu yen, V.L., Ta, T.K.O., Tatei shi, M., 2000. Late holo cene depo si tional envi ron ments and coastal evo lu tion of the Me kong River Delta, South-ern Viet nam. J. Asian Earth Sci. 18, 427–439.

Nick son, R.T., McAr thur, J.M., Rav ens croft, P., Bur gess, W.G., Ahmed, K.M., 2000. Mech a nism of arsenic release to ground wa ter, Ban gla desh and West Ben gal. Appl. Geo chem. 15, 403–413.

Po lizz ot to, M.L., Har vey, C.F., Sut ton, S.R., Fen dorf, S., 2005. Pro cesses con-du cive to the release and trans port of arsenic into aqui fers of Ban gla-desh. Proc. Nat. Acad. Sci. USA 102, 18819–18823.

Po lizz ot to, M.L., Ko car, B.D., Ben ner, S.G., Samp son, M., Fen dorf, S., 2008. Near-sur face wet land sed i ments as a source of ar sen ic release to ground water in Asia. Nature 454, 505–508.

Polya, D.A., Ga ult, A.G., Di eb e, N., Feld man, P., Ro sen boom, J.W., Gil li gan, E., Fred er icks, D., Mil ton, A.H., Samp son, M., Row land, H.A.L., Lyth goe, P.R., Jones, J.C., Mid dle ton, C., Cooke, D.A., 2005. Arsenic haz ard in shal low Cam bo dian ground wa ters. Miner. Mag. 69, 807–823.


Post ma, D., Lar sen, F., Hue, N.T.M., Duc, M.T., Viet, P.H., Nhan, P.Q., Jes sen, S., 2007. Arsenic in ground wa ter of the Red River flood plain, Viet nam: con trol ling geo chem i cal pro cesses and reac tive trans port mod el ing. Geo chim. Cos mo chim. Acta 71, 5054–5071.

Rav ens cort, P., Bur gess, W.G., Ahmed, K.M., Bur ren, M., Per rin, J., 2005. Arsenic in ground wa ter of the Ben gal Basin, Ban gla desh: dis tri bu tion, field rela tions, and hydro log i cal set ting. Hydrol. J. 13, 727–751.

Stute, M., Zheng, Y., Sch los ser, P., Horn eman, A., Dhar, R.K., Ho que, M.A., Sed di que, A.A., Sham sudduha, M., Ahmed, K.M., van Geen, A., 2007. Hydro log i cal con trol of As con cen tra tions in Ban gla desh ground wa ter. Water Re sour. Res. 43, doi:10.1029/2005WR00449 W09417.

Ta, T.K.O., Ngu yen, V.L., Tatei shi, M., Ko bay ash i, I., Tan a be, S., Sa i to, Y., 2002. Holo cene delta evo lu tion and sed i ment dis charge of the Me kong River south ern Viet nam. Qua tern. Sci. Rev. 21, 1807–1819.

Tam ura, T., Sa i to, Y., Si eng, S., Ben, B., Kong, M., Choup, S., Tsuk awaki, S., 2007. Depo si tional facies and radio car bon ages of a drill core from the Me kong River lowland near Phnom Penh, Cam bo dia: evi dence for tidal sed i men ta tion at the time of holo cene max i mum flood ing. J. Asian Earth Sci. 29, 585–592.

Thang ara jan, M., Linn, F., Uhl, V., Ba kaya, T.B., Gab aake, G.G., 1999. Mod el ling an inland delta aqui fer sys tem to evolve pre-devel op-ment man age ment schemes: a case study upper Tha mal a kane River val ley, Botswana, south ern Africa. Envi ron. Geol. 38, 285–295.

van Geen, A., Aziz, Z., Horn eman, A., Wein man, B., Dhar, P.K., Zheng, Y., Good bred, S., Ver steeg, R., Sed di que, A.A., Ho que, M.A., Ahmed, K.M., 2006. Preliminary evi dence of a link between sur face soil prop er ties and the arsenic con tent of shal low ground wa ter in Ban gla desh. J. Geo-chem. Explor. 88, 157–161.

Groundwater flow in an arsenic-contaminated aquifer, Mekong Delta, Cambodia

Documents

Transcript of Groundwater flow in an arsenic-contaminated aquifer, Mekong Delta, Cambodia