Fluvial facies architecture in small-scale river systems in the...

Fluvial facies architecture in small-scale river systems in the Upper Dupi

Tila Formation, northeast Bengal Basin, Bangladesh

M. Royhan Gani*, M. Mustafa Alam

Department of Geology, University of Dhaka, Dhaka 1000, Bangladesh

Received 27 December 2002; revised 15 October 2003; accepted 3 November 2003

Abstract

The late stage basin-fill history of the fluvial Dupi Tila Group (Plio-Pleistocene) is described. These rocks have been deposited in the

Sylhet trough, a sub-basin of the Bengal Basin, in a foreland basin setting. This outcrop study, carried out in Sylhet, Bangladesh, presents the

first detailed facies analysis of the Upper Dupi Tila Formation. Four facies have been identified: trough cross-bedded sandstone (St), ripple

cross-laminated sandstone (Sr), finely laminated mud with ripples (Fl), and massive mud with rootlets (Fm). Facies analysis supplemented

with embedded Markov chain analysis, reveals small-scale fining-upward cycles (average 4.5 m thick). Facies architectural elements include

channel (CH), lateral accretion (LA), sandy bedforms (SB), and overbank fines (OF) with limited vertical and lateral connectivity of the sand

bodies. The average channel depth and width is 5 and 30 m, respectively. Sand body geometry ranges from tabular, to sheet, to shoestring

with a 0.45 net to gross ratio. This study shows that the Upper Dupi Tila Formation is composed of small-scale, mudstone-reach meandering

river deposits. In Bangladesh, the Dupi Tila Formation is the main aquifer presently being utilized. Understanding of facies architecture and

sand body geometry of this Formation is crucial in examining the issue of arsenic and other contaminations of ground water in Bangladesh.

q 2003 Elsevier Ltd. All rights reserved.

Keywords: Bengal basin; Dupi Tila Formation; Fluvial deposits; Facies architecture; Arsenic hazard

1. Introduction

The Bengal Basin (Fig. 1), covering Bangladesh and part

of eastern India, is found within the junction of the

Himalayan Range to the north and Indo-Burman Range to

the east, and preserves the tectono-sedimentary history

(Cretaceous-Holocene) of these two orogenic provinces.

The Bengal Basin is known to develop a thick (20 km)

sedimentary succession (Curray, 1991) that has long been of

interest from the petroleum exploration point of view. Due

to some of the recent studies (Alam et al., 2003; Gani and

Alam, 2003, 1999) the tectono-sedimentary history of the

Bengal Basin is now better understood.

Dupi Tila Formation is the main aquifer bearing strata

for the entire Nation of Bangladesh, except for the

southwest corner. Although Bangladesh is now experien-

cing a public health crisis due to extreme arsenic

concentrations (locally concentrations reach 3.5 mg/l) in

ground water, there are no published accounts on the Dupi

Tila Group that focus on sedimentology and sand body

architecture. The objective of the present study from the

Sylhet Trough, a sub-basin in the northeast of the Bengal

Basin, is to provide the first detailed description of the

fluvial facies architecture of the Upper Dupi Tila

Formation and its possible influence on determining

aquifer behavior and the transport pathways of arsenic.

2. Geologic setting

The geologic evolution of the Bengal Basin (Fig. 1)

began in the Late Mesozoic with the break-up of

Gondwanaland and is on going. Alam et al. (2003) have

presented a revised tectonic and stratigraphic scenario of the

Bengal Basin emphasising three separate geo-tectonic

provinces within the basin. The Sylhet Trough (province

2), mostly underlain by continental crust, has accumulated

more than 18 km thick sedimentary strata. Post-Paleogene

history of the Sylhet Trough has been controlled mainly by

1367-9120/$ - see front matter q 2003 Elsevier Ltd. All rights reserved.

doi:10.1016/j.jseaes.2003.11.003

Journal of Asian Earth Sciences 24 (2004) 225–236

www.elsevier.com/locate/jaes

* Corresponding author. Department of Geosciences, The University of

Texas at Dallas, P.O. Box 830688, FO 21, Richardson, TX 75083-0688,

USA. Tel.: þ1-972-883-2401; fax: þ1-972-883-2537.

E-mail address: [email protected] (M. Royhan Gani).

http://www.elsevier.com/locate/jseaes

two tectonic events—increased movement along Dauki

Fault (upthrust), and westward advancement of Indo-

Burman Range due to the continuing oblique subduction

of Indian plate beneath the Burmese plate. In the late

Pliocene, when the Chittagong Tripura Fold Belt (CTFB)

was uplifted at the eastern margin of the Trough, a huge

volume of clastic sediments have been shed off, resulting in

the deposition of the Plio-Pleistocene Dupi Tila Group in

the resulting foreland basin (Fig. 1).

The Dupi Tila Formation was named by Evans (1932)

and interpreted as having been deposited in a fluvial

environment. In seismic stratigraphy, the Dupi Tila

Formation has been broadly subdivided into a lower sandy

unit and an upper argillaceous unit (Table 1) (Hiller and

Elahi, 1988). Together these two units are called Dupi Tila

Group (Alam et al., 2003). Johnson and Alam (1991)

mentioned that the alternating channel and flood plain

deposits of the Dupi Tila Formation indicate fining-upward

cycles of probable meandering river origin. The present

study focuses on the Pleistocene Upper Dupi Tila Formation

(Table 1).

3. Study area and methodology

The Upper Dupi Tila Formation is exposed on several

small hills on the north side of the Sylhet Town,

Bangladesh, in gently tilted strata (dip ,48). In this area,

four hills have been chosen for this study (Fig. 2). Vertical

cliff faces of these four hills are oriented NW-SE, range

in thickness from 8 to 20 m, and in lateral extent from 18

to 30 m.

Three approaches have been taken in interpreting the

sedimentological history of the study area. A vertical facies

analysis (1D) has been done along suitable lines in each of

the cliff faces to establish the basic facies types and to detail

inherent sedimentary features. Embedded Markov chain

analysis has been performed on facies transitions matrix to

Fig. 1. Generalized tectonic map of the Bengal Basin and surroundings (from Gani and Alam, 2003). Hinge zone separates the shallow Indian platform to

the northwest from the deeper Bengal foredeep to the southeast. The study area is indicated as X mark within the Sylhet Trough. (CTFB, Chittagong Tripura

Fold Belt).

M. Royhan Gani, M. Mustafa Alam / Journal of Asian Earth Sciences 24 (2004) 225–236226

determine whether there is a preferential vertical cycle in

these fluvial deposits. Photographs of the entire cliff faces

were made and facies architecture analysis (2D) was

conducted by tracing bounding surfaces from these

photographs. This type of analysis shows degrees of lateral

and vertical connectivity of sand bodies and helps to

reconstruct the paleohydraulics.

Dominant paleocurrent direction in the study sections is

towards southwest indicating that the cliff faces are oriented

roughly perpendicular to the paleoflow (Fig. 2). Sediments

were delivered to the study area from both north and east.

The northern source area includes the preexisting Shillong

plateau and its fringe sedimentary rocks, whereas the

eastern source includes the newly uplifted CTFB (Fig. 1).

Table 1

Stratigraphic succession of the Sylhet trough (revised from Hiller and Elahi, 1988). Seismic markers refer to the continuous and reasonably traceable seismic

reflectors in the seismic sections of the Sylhet trough

Age (approx.) Group Formation Seismic marker Thickness (max.) (m)

Holocene Alluvium

Pleistocene Dupi Tila Upper Dupi Tila Yellow 3350

Late Pliocene Lower Dupi Tila

Mid-Pliocene Tipam Girujan Clay 3500

Tipam Sandstone Brown

Early Plio to Miocene Surma Upper Surma Red 3900

Lower Surma Violet

Oligocene Barail (Undifferentiated) 7200

Kopili Shale Blue

Eocene to Paleocene Jaintia Sylhet Limestone

Tura Sandstone

Pre-Paleocene Undifferentiated sedimentary rocks (with some volcanics?) on the

continental basement complex

No data

Fig. 2. Map of Sylhet town, Bangladesh showing the locations of the studied hills. Note that dominant paleocurrent direction is roughly perpendicular to the

cliff faces.

M. Royhan Gani, M. Mustafa Alam / Journal of Asian Earth Sciences 24 (2004) 225–236 227

4. Results

4.1. Vertical facies analysis (1D)

Facies code of Miall (1978) has been used for the present

investigation because only four basic facies, which can be

readily codified applying this scheme, have been found for

the entire study area with only few exceptions. These four

facies are trough cross-bedded medium sandstone (St),

ripple cross-laminated very fine to fine sandstone (Sr), finely

laminated mud with ripple cross-lamination (Fl), and

massive mud with rootlets (Fm). The description and

interpretation of each of these facies (Fig. 3) are presented

in Table 2.

For the purpose of detailed characterization of facies

variability in the study sections, vertical lithologs were

generated for each of the hill sites (Fig. 4). The vertical

arrangement of four facies types is indicative of repetitive,

fining-upward cycles of fluvial origin (Fig. 4, Table 2). Each

of these upward fining cycles is marked, from oldest to

youngest, in the lithologs. The lateral variability and the

order of bounding surfaces of the facies in terms of

architectural element are discussed in a later section.

4.2. Embedded Markov chain analysis

In sedimentology, Markov chain analysis is used to

establish the prevalent pattern of vertical facies change in

Fig. 3. Photographs of four facies identified in the studied succession (see Table 2 for description and interpretation of the above facies types). (a) Trough cross-

stratified sandstone facies (St) showing poorly defined set of trough cross-stratification. The face of the exposure is oriented perpendicular to the cliff face of

hill-1. (b) Ripple cross-laminated sandstone facies (Sr) showing sets of ripple cross-lamination. (c) Laminated mudstone facies (Fl) showing finely laminated

mud with very fine sand interlaminae. Note the black root traces (arrowed). (d) Massive mudstone facies (Fm) showing structureless gray mudstone with well-

developed network of large roots (arrowed) preserved as concentric iron precipitation around a central hollow. (e) Facies Fm containing two tiny channels

(arrowed) filled with fine sandstone, probably originated from short-lived storm runoff on flood plains.


a stratigraphic succession. The technique filters out

significant facies transitions (i.e. signal) from randomly

occurring facies transitions (i.e. noise). The four measured

lithologs of this study give the opportunity to test whether

there is any preferred facies transition in the Upper Dupi

Tila Formation. Table 3a gives the one-step vertical facies

transitions matrix (facies in rows are overlain by facies in

columns) obtained from Fig. 4 with a total of 72 transitions.

Because there is a diagonal structural zero in the facies

transition matrix, a rigorous statistical method called

‘embedded Markov chain analysis’ (Powers and Easterling,

1982; Carr, 1982) has been applied. Expected transition

frequencies of Table 3b have been calculated by quasi-

independence method of Powers and Easterling (1982).

Table 3c gives the probability matrix (normalized differ-

ences of Table 3a and b) of the facies transitions. The

transitions with higher positive values in Table 3c indicate

preferred facies transitions with high probability of

occurrence.

Combining both statistically significant transitions (from

Table 3c) and geologically meaningful transitions, which

may not be statistically significant, a path diagram of facies

transitions was constructed (Fig. 5a). Facies St are most

commonly [normalized probability (NP) ¼ þ1.69] overlain

by facies Sr, and are less commonly (NP ¼ þ0.27) overlain

by facies Fl, which, in tern, are most commonly

(NP ¼ þ1.572) overlain by facies St. This facies transitions

path gives rise to a ‘modal cycle’ of decreasing flow energy

originating probably from repetitive chute cut-off of an

active channel (Fig. 5). When facies Fl is overlain by facies

Fm (NP ¼ þ0.192) it generates bilateral facies transitions

between facies Fm and Sr, which is interpreted as crevasse

splay/sheet-flood cycle of overbank deposits (Fig. 5). Path

diagram of Fig. 5a leads to an ideal fining-upward

succession (Fig. 5b) of typical of fluvial strata.

4.3. Facies architecture analysis (2D)

One of the most powerful and widely used techniques

in analyzing fluvial strata is architectural element

analysis. This technique emphasizes facies distribution

and associated bounding surfaces beyond the level of 1D

vertical lithologs to deduce the depositional scenario.

Miall (1985, 1996) has standardized basic architectural

element types and their bounding surface hierarchy in

fluvial succession. However, as most of the natural

processes are continuous, it may not be always easy to

apply Miall’s scheme strictly (Bridge, 1993). Some of

the recent publications on fluvial facies architecture (e.g.

Halfar et al., 1998; Holbrook, 2001; Jo and Chough,

2001) show various degrees of application of the Miall’s

scheme.

The objective of facies architecture analysis in the

present investigation is to apply the central theme of Miall

(1985, 1996) in interpreting depositional history and

depicting facies heterogeneity in the Upper Dupi Tila

Formation. Various architectural elements and the order of

bounding surfaces as applied in this study follow Miall’s

scheme: first-order surfaces bind bedforms (dunes and

ripples) cross sets, second-order surfaces bind bedforms

cosets (not used in this study), third-order surfaces define

macroform (e.g. bar) growth increments, fourth-order

surfaces define minor channel bases or tops of macroform

units, fifth-order surfaces represent bases of major channels.

However, as admitted previously, strict application of the

scheme is neither feasible nor desirable. Bedding diagrams

Table 2

Description and interpretation of identified facies in the present study

Facies types Description Interpretation

Trough cross-bedded

sandstone (St) (Fig. 3a)

Yellowish brown to pinkish red medium sandstone; mud clasts

(pebble- sized), mostly occurring at the basal part, sometimes

scattered throughout; poorly defined trough cross-stratification

with an average 25 cm thick set; invariably starting on an

erosion surface

Migration of sandy dunes on an erosion surface

of variable hierarchies; representing both

channel filling and bar developing

Rippled sandstone

(Sr) (Fig. 3b)

Fine to very fine yellowish sandstone; ripple cross-lamination,

mostly trough shaped with a few cm thick set; mostly with

organic-material/mud drapes along trough; asymmetric when

ripple forms preserved

Migration of current ripples; fluctuating currents

indicated by drapes of fine materials

Laminated mud with

ripples (Fl) (Fig. 3c)

Finely (a few mm) and parallel laminated mud with some very

fine sand interlaminae containing micro ripples; color variation

of white, red, pink, and yellow in mud laminae; occasional

rootlets, very occasional burrows disrupting laminae;

sometimes laterally discontinuous facies

Suspension settling with very weak current

mostly on bar top, and subordinately on flood

plain; color of the mud indicating oxidized and

well drained condition; small plant growth and

burrowing activity at the time of low stage of

channel flow

Massive mud with

rootlets (Fm) (Fig. 3d and e)

Dark grey to bluish grey mud; mostly structureless in

appearance; well established network of large root systems

preserved as concentric iron precipitation with a central hollow;

very occasional leaf impressions and coalified tree stems

Suspension settling over flood plains; prolonged

waterlogged condition with stable plant

development


Fig. 4. Vertical lithologs of the four hills. The locations of the measured lines on cliff sections are shown in Figs. 6–8. (a) Hill-1 with seven fining-upward

cycles. (b) Hill-2 with six fining-upward cycles. (c) Hill-3 with two fining-upward cycles. (d) Hill-4 with two fining-upward cycles.


of the studied hills (except hill-2) have been produced,

which, in combination with the corresponding vertical

lithologs (Fig. 4), are suitable to perform facies architecture

analysis.

4.3.1. Hill-1

The base of C2 (cycle-2) in hill-1 (Figs. 4a and 6) rests on

a relatively thick (about 2 m) facies Fm, and is a prominent

erosion surface with abundant coal and mudstone clasts.

The bases of C3 through C6 are less prominent erosion

surfaces. Facies St found at the base of each cycle consists

of both single and multiple sets of trough cross-bedding.

Boundaries between facies St and Sr are mostly gradational.

The base of C7 is again a prominent erosion surface with

high erosional relief at the cut-bank side. Large slump

blocks (up to 1 m long) and lateral accretion surfaces are

also observed on the cut-bank side (Fig. 6b).

Identified architectural elements and their order of

bounding surfaces in hill-1 are shown in Fig. 6b. First-

order surfaces define the set boundaries of cross-bedded

facies (St). Boundaries between sandy (St and/or Sr) facies

and muddy (Fl and/or Fm) facies within each of the fining-

upward cycles are interpreted as fourth-order surfaces, as

they possibly indicate the end of a macroform (e.g. a bar)

development within a channel (Miall, 1994). Also, the

minor erosion surfaces (bases of C3 to C6) are clearly

fourth-order surfaces (Miall, 1994). The two prominent

erosion surfaces (bases of C2 and C7) thought to represent

bases of major channels are assigned as fifth-order surfaces

(Holbrook, 2001; Halfar et al., 1998). Identified architec-

tural elements for hill-1 are shown in Fig. 6b. In this case,

C2 to C6 represent repetitive development of element SB

(sandy bedforms, consisting of dunes and ripples), and the

lower half of C7 indicates lateral juxtaposition of elements

CH (channel) and LA (lateral accretion). OF (overbank

fines) elements are also identifiable in Fig. 6.

4.3.2. Hill-3

Following the same principles and criteria as described

under hill-1, architectural elements and order of bounding

surfaces in hill-3 are depicted in Fig. 7. Internal bounding

surface of the lower sandy portion of C1 is not observable.

The base of C2 represents a fifth-order surface. The sandy

portion of C2 develops lateral wings (elements SB) into

overlying OF element, and shows internal lateral accretion

surfaces (third-order surface) with average dip of 158 in a

direction roughly perpendicular to the paleoflow.

4.3.3. Hill-4

Although the internal bounding surfaces of the sandy

portion of C1 is not observable, its base probably indicates a

fifth-order surface. Two minor channels bounded by fourth-

order surfaces (Halfar et al., 1998) are characteristically

developed within thick OF element. SB element at the upper

part consisting of facies Sr is continuous across the

exposure.

5. Discussion

5.1. Depositional pattern

Results of the analyses presented above suggest that the

Upper Dupi Tila Formation in the Sylhet Trough was

deposited by single-thread, meandering river systems. The

presence of repetitive upward fining cycles with facies

transitions of decreasing flow energy, simple bank-attached

bar development with lateral accretion surfaces, and

channel confinement within thick flood plain deposits attest

to this interpretation.

Hill-1 represents thickest and best quality exposure

comparing to the other hills, and can be used as the basis for

section comparison. Two fifth-order surfaces of this hill

divide the entire succession into three units X, Y and Z,

from oldest to youngest (Fig. 6b), with each unit having its

own style of sedimentation. As hills-1, -2, and -3 are closely

Table 3a

One-step embedded matrix of observed facies transitions (facies in rows are

overlain by facies in columns). Data has been obtained from Fig. 4,

combining hill-1 (30 transitions), hill-2 (17 transitions), hill-3 (9

transitions), and hill-4 (16 transitions)

St Sr Fl Fm Total

St – 9 7 1 17

Sr 1 – 7 10 18

Fl 9 3 – 8 20

Fm 4 6 7 – 17

Total 14 18 21 19 72

Table 3c

Probability matrix of facies transitions. Calculation has been made from

Table 3a and b by applying normalized-difference formula of Power and

Easterling (1982)

St Sr Fl Fm

St – þ1.690 þ0.207 21.886

Sr 21.627 – 20.154 þ1.547

Fl þ1.572 21.560 – þ0.192

Fm 20.109 þ0.141 20.032 –

Table 3b

Estimated expected transition frequencies of Table 3a. Quasi-independence

method of Power and Easterling (1982) has been applied to calculate

transition frequencies

St Sr Fl Fm Total

St – 5.16 6.47 5.37 17.00

Sr 4.42 – 7.42 6.16 18.00

Fl 5.36 7.18 – 7.47 20.01

Fm 4.23 5.66 7.10 – 16.99

Total 14.01 18.00 20.99 19.00 72.00


spaced (Fig. 2) these three units may be co-relatable within

these hills; and for hill-4 they may be comparable.

Unit X, exposed only at Hill-1, contains C1 (cycle 1)

(Fig. 6b), which is a well developed fining-upward cycle

with about 2 m thick Fm facies (Fig. 4a). This dark grey

overbank facies Fm contains abundant organic matter with

conspicuous and well-developed networks of large roots

(Table 2 and Fig. 3d). These criteria suggest water-logged

reducing condition (Collinson, 1996) and flood plain

stabilization for a prolonged period.

The erosional base of unit Y represents channel avulsion

on flood plain mud with the initiation of new channel cycles

(Fig. 6b). Based on the similarities of sedimentation pattern

and resulting sand body distribution, the entire succession of

hill-2 (Fig. 4b), and basal part of hill-3 (Fig. 7) are co-

relatable to this unit at hill-1. The unit represents repetitive

fining-upward cycles. Each cycle starts with element SB on

an erosion surface of fourth-order, indicating within-channel

development of sandy bedforms like dunes and ripples, and

ends up with thin deposition of facies Fl. Repetitive

development of element SB with intervening thin Fl facies

(Figs. 4a, b and 6b) probably indicates periodic chute cut-off

(Fig. 5a) of a meandering channel during major flood events.

Facies Fl shows poor lateral continuity in hill-2. The top of

unit Y is characterized by a thick development of element

OF similar in character as that of unit X.

Unit Z is well exposed in hill-1, -3, and -4 (Figs. 6b, 7

and 8). The base of this unit also represents a new cycle of

channel avulsion on flood plain mudstones. Channels with

distinct cut-banks are preserved in hill-1 and -4. Deformed

large mudstone blocks found close to the cut-bank in hill-1

suggests slumping of the cut-bank (BØe, 1988). In hill-3,

the channel sand body of unit Z shows evidence of lateral

migration from northwest to southeast. The successive

positions of channel migration are depicted with lateral

wings (Fig. 7, element SB) of levee deposits (Friend, 1983).

The stacking pattern of these wings strongly suggests that an

initial laterally migrating channel had turned into a rapidly

aggrading channel. The thick flood plain deposits of unit Z

in hill-4 (Fig. 8) contain two small anastomosed-channel

Fig. 5. (a) Path diagram of the facies transitions combining both objective (statistically significant, Table 3c) and subjective (geologically meaningful)

judgment. Double-lined arrows indicate high positive values (.þ1.5) of normalized probabilities (Table 3c). (b) Idealized fining-upward cycle (based on Fig.

5a) of the studied Upper Dupi Tila Formation with depositional interpretation. See text for discussion.


Fig. 6. Vertical cliff face of hill-1. Beds are gently tilted towards left (northwest). (a) Photograph of the cliff face. Note the standing man (circled) for scale.

(b) Overlay tracing of hill-1 showing architectural elements and their bounding surface hierarchy. Vertical line indicating cycle numbers is the measured

section of Fig. 4a. Paleocurrent direction is roughly perpendicular to the cliff face and towards the viewer (southwest).

Fig. 7. Overlay tracing of hill-3 showing architectural elements and their bounding surface hierarchy. Beds are gently tilted towards left (northwest). Vertical line

indicating cycle numbers is the measured section of Fig. 4c. Paleocurrent direction is roughly perpendicular to the cliff face and towards the viewer (southwest).


deposits probably indicating ephemeral channels of storm

origin (e. g. Olsen, 1987). Element SB at the top part of Fig.

8 is indicative of sheet-flood deposits.

The exposure window of the present study is not wide

enough to go further and study the interrelationship of

channel aggradation, avulsion, and lateral migration.

However, facies architecture and stacking behavior of

sand bodies of units Y and Z are distinctly different from

each other. The concept of LAB model (Allen, 1978; Bridge

and Leeder, 1979), though it has been criticized (Heller and

Paola, 1996), can be applied in this connection. Due to

repeated chute cut-off cycles, unit Y shows restricted lateral

migration probably under low accommodation relative to

sediment supply. Whereas, the channel stacking pattern of

unit z, particularly thick OF element separating CH element,

indicates high accommodation relative to sediment supply.

This increase in accommodation may be due to an increase

in subsidence rate of the foreland basin as a result of

continuous loading of the adjacent fold-thrust belt (Fig. 1).

5.2. Paleohydraulics and sand body geometry

Several empirical formulae have been derived to

determine the depth and width of paleochannels from

sedimentary structure data. According to Leclair and Bridge

(2001), paleoflow depth of a fluvial channel can be

calculated from the set thickness of cross-bedding. The

formulae used are: Dune height ¼ 3 £ mean cross set

thickness; Flow depth ¼ (6 to 10) £ dune height. As the

average set thickness of trough cross-bedding of facies St is

25 cm, the average paleoflow depth is 5 m. Moreover,

according to Leeder (1973): w ¼ 1:5 £ h=tan u; where, w;

width of the channel, h; depth of the channel, u; dip angle of

lateral accretion surface. As the average u in the present

study is 158, the average channel width is 30 m. Therefore,

the average width to depth ratio of the meandering channels

that deposited the Upper Dupi Tila Formation is 6.

The terminology related to sand body geometry in fluvial

strata mostly depends on the scale of observation and the

scale of depositing channel. Considering the spatial scale of

present investigation, three different types of sand body

have been observed. The two small channels of unit Z at

hill-4 (Fig. 8) can be termed as shoe-string sand bodies;

element SB at the top part of unit Z at hill-4 represents a

blanket-type sheet sand body; and the sand bodies of unit Y

and at the lower part of unit Z probably indicate a tabular

geometry.

5.3. Implications for the arsenic hazard in Bangladesh

Upper Dupi Tila Formation serves as the main aquifer of

Bangladesh except for the south-west part of the country.

Arsenic is now a serious environmental hazard for Nepal,

West Bengal (India) and Bangladesh due to its presence in

the aquifer at high concentration. Out of 64 districts in

Bangladesh 25 districts have arsenic concentration above

the recommended level (0.05 mg/l) of WHO for drinking

water (locally concentration reaches 3.5 mg/l) affecting

more than 25 million people. However, sand body geometry

and facies architecture of these aquifer bearing strata are so

far very poorly understood. Due to the lack of any published

facies model, hydrogeologists working in Bangladesh are

mostly using a layer-cake model in constructing aquifer

panel diagrams.

Fig. 8. Overlay tracing of hill-4 showing architectural elements and their bounding surface hierarchy. Vertical line indicating cycle numbers is the measured

section of Fig. 4d. Paleocurrent direction is roughly perpendicular to the cliff face and towards the viewer.


According to a recent study by McArthur et al. (2001)

the distribution of organic matter, which is responsible

for reduction of FeOOH and release of sorbed arsenic in

the aquifer sediments, is the primary control on arsenic

pollution in Bangladesh. If this is the case then the

vulnerability of arsenic contamination in these fluvial

aquifers can be predicted by knowing the relative

distribution of overbank (aquiclude) and channel (aquifer)

deposits that controlled the sites of organic matter

accumulation in the fluvial basin. Shallow fluvial

stratigraphy may also control some of the redox cycling

of arsenic as well as movement of arsenic-contaminated

groundwater in Bangladesh (S. Goodbred pers. Comm.,

2002). Considering the above views, the present study

may contribute significantly in controlling arsenic

problem in Bangladesh as it gives the first detailed

description of bedding geometry of the Upper Dupi Tila

Formation deposited in mudstone-rich meandering fluvial

systems. This may suggests a far more compartmenta-

lized aquifer than has been previously understood or

modeled. Although dealing with only a portion of the

Upper Dupi Tila Formation in some limited outcrops,

this study constrains the dimension, spatial distribution,

and small-scale heterogeneity of the sand bodies of this

formation. The knowledge of this kind is an essential

pre-requisite for any meaningful groundwater flow

modeling designed to mitigate arsenic as well as other

groundwater hazard in Bangladesh. Moreover, as the

studied deposits are from a relatively small channel

system, they may have greater relevance to the shallow

aquifer than studies of the main Ganges or Brahmaputra

braid belts (S. Goodbred pers. Comm., 2002).

The facies architecture presented here can also be used as

analog for reservoir heterogeneity of small-scale mean-

dering river deposits. This architectural model can be

improved by further research incorporating log data from

boreholes done for groundwater withdrawal in this region.

6. Conclusions

Present study presents first detailed facies architectural

analysis of the Upper Dupi Tila Formation of the Sylhet

Trough, Bengal Basin. The following conclusions can be

made from this study:

1. Upper Dupi Tila Formation shows repetitive develop-

ment of fining-upward cycles (average 4.5 m thick)

containing four facies: trough cross-stratified sandstone

(St), ripple cross-laminated sandstone (Sr), finely

laminated mud with ripples (Fl), and massive mud with

rootlets (Fm).

2. Facies architecture analysis from 2D outcrops reveals

elements CH (channel), SB (sandy bedform), LA (lateral

accretion), and OF (overbank fines) with four different

orders of bounding surfaces.

3. Paleohydraulic reconstructions show that average depth

and width (at the bankfull stage) of the paleochannel was

5 and 30 m, respectively. Sand body geometry ranges

from tabular, to sheet, to shoestring.

4. The studied succession has been deposited by small-

scale, mudstone-rich meandering river systems with the

dominance of single-channel fluvial style characterized

by simple bank-attached bars.

5. The complex facies architecture of Upper Dupi Tila

Formation, which is the main aquifer bearing strata in

Bangladesh, indicates a less connected and highly

compartmentalized aquifer geometry. This type of

knowledge is essential for meaningful modeling of

groundwater flow to mitigate the arsenic as well as

other groundwater pollutions in Bangladesh.

Acknowledgements

The authors are very grateful to D. Nahid Sultana, the

wife of first author, for drafting the diagrams and editing the

manuscript. Thanks are due to D. Zafrul Hasan, the then

principal of School of Forestry, Sylhet, for giving logistic

support during the field work. Discussion with Janok

Bhattacharya has improved the quality of the paper. An

encouraging review from Steven Goodbred on an earlier

version of the manuscript is acknowledged. Finally, the

authors are grateful to the journal’s reviewers Mead

A. Allison and Brian J. Willis for their constructive

criticism on the manuscript.

References

Alam, M., Alam, M.M., Curray, J.R., Chowdhury, M.L.R., Gani, M.R.,

2003. An overview of the sedimentary geology of the Bengal Basin in

relation to the regional tectonic framework and basin-fill history.

Sedimentary Geology 155, 179–208.

Allen, J.R.L., 1978. Studies in fluviatile sedimentation: an exploratory

quantitative model for the architecture of avulsion-controlled alluvial

suites. Sedimentary Geology 21, 129–147.

Bridge, J.S., 1993. Description and interpretation of fluvial deposits: a

critical perspective. Sedimentology 40, 801–810.

Bridge, J.S., Leeder, M.R., 1979. A simulation model of alluvial

stratigraphy. Sedimentology 26, 617–644.

BØe, R., 1988. Alluvial channel bank-collapse phenomena in the Old Red

Sandstone Hitra Group, Western Central Norway. Geological Magazine

125, 51–56.

Carr, T.R., 1982. Log-linear models, Markov chains and cyclic sedimen-

tation. Journal of Sedimentary Petrology 52, 905–912.

Collinson, J.D., 1996. Alluvial sediments. In: Reading, H.G., (Ed.),

Sedimentary environments: process, facies, and stratigraphy, Black-

well, Oxford, pp. 37–82.

Curray, J.R., 1991. Geological history of the Bengal geosyncline. Journal of

Association of Exploration Geophysicist XII, 209–219.

Evans, P., 1932. Tertiary succession in Assam. Transaction of Miner-

alogical and Geological Institute of India 27, 155–260.

Friend, P.F., 1983. Towards the field classification of alluvial architecture

or sequence. In: Collinson, J.D., Lewin, J. (Eds.), Modern and ancient


fluvial systems, International Association of Sedimentologists, Special

Publication, 6, pp. 345–354.

Gani, M.R., Alam, M.M., 1999. Trench-slope controlled deep-sea clastics

in the exposed lower Surma Group in the southeastern fold belt of the

Bengal Basin, Bangladesh. Sedimentary Geology 127, 221–236.

Gani, M.R., Alam, M.M., 2003. Sedimentation and basin-fill history of the

Neogene clastic succession exposed in the south eastern fold belt of the

Bengal Basin, Bangladesh: a high resolution sequence stratigraphic

approach. Sedimentary Geology 155, 227–270.

Halfar, J., Walter, R., Walther, H., 1998. Facies architecture and

sedimentology of a meandering fluvial system: a Palaeogene example

from the Weisselster Basin, Germany. Sedimentology 45, 1–17.

Heller, P.L., Paola, C., 1996. Downstream changes in alluvial architecture:

an exploration of controls on channel-stacking patterns. Journal of

Sedimentary Research B66 (2), 297–306.

Hiller, K., Elahi, M., 1988. Structural growth and hydrocarbon entrapment

in the Surma basin, Bangladesh. In: Wagner, H.C., (Ed.), Petroleum

Resources of China and Related Subjects, Circum-Pacific Council

Energy Mineral Resources Earth Science Series 10, pp. 657–669.

Holbrook, J., 2001. Origin, genetic interrelationships, and stratigraphy over

the continuum of fluvial channel-form bounding surfaces: an illustration

from middle Cretaceous strata, southeastern Colorado. Sedimentary

Geology 144, 179–222.

Jo, H.R., Chough, S.K., 2001. Architectural analysis of fluvial sequences in

the northwestern part of Kyongsang Basin (Early Cretaceous), SE

Korea. Sedimentary Geology 144, 307–334.

Johnson, Y.J., Alam, A.M.N., 1991. Sedimentation and tectonics of the

Sylhet Trough, Bangladesh. Geological Society of America Bulletin

103, 1513–1527.

Leclair, S.F., Bridge, J.S., 2001. Quantitative interpretation of sedimentary

structures formed by river dunes. Journal of Sedimentary Research 71,

713–716.

Leeder, M.R., 1973. Fluviatile fining-upward cycles and the magnitude of

palaeochannels. Geological Magazine 110, 265–276.

McArthur, J.M., Ravenscroft, P., Safiullah, S., Thirlwall, M.F., 2001.

Arsenic in groundwater: testing pollution mechanisms for sedimen-

tary aquifers in Bangladesh. Water Resources Research 37 (1),

109–117.

Miall, A.D., 1978. Facies types and vertical profile models in braided river

deposits: a summery. In: Miall, A.D., (Ed.), Fluvial Sedimentology,

Canadian Society for Petroleum Geologist, Memoir, 5, pp. 597–604.

Miall, A.D., 1985. Architectural-element analysis: a new method of

facies analysis applied to fluvial deposits. Earth Science Review 22,

261–308.

Miall, A.D., 1994. Reconstructing fluvial macroform architecture from

two-dimensional outcrops: examples from the Castlegate Sandstone,

Book Cliffs, Utah. Journal of Sedimentary Research B64, 146–158.

Miall, A.D., 1996. The Geology of Fluvial Deposits: Sedimentary Facies,

Basin Analysis, and Petroleum Geology. Springer, Berlin, 582 p.

Olsen, H., 1987. Ancient ephemeral stream deposits: a local terminal fan

model from the Bunter Sandstone Formation (L. Triassic) in the

Tronder-3, -4 and -5 wells, Denmark. In: Frostick, L.E., Reid, I. (Eds.),

Desert Sediments: Ancient and Modern, Geological Society of London,

Special Publication, 35, pp. 69–86.

Powers, D.W., Easterling, R.G., 1982. Improved methodology for using

embedded Markov chains to describe cyclical sediments. Journal of

Sedimentary Petrology 52, 913–923.


Fluvial facies architecture in small-scale river systems in the...

Documents

Transcript of Fluvial facies architecture in small-scale river systems in the...