See discussions, stats, and author profiles for this publication at: [Link]

net/publication/226412914

Three Recent Extreme Floods in Bangladesh: A Hydro-Meteorological Analysis

Article in Natural Hazards · January 2003

DOI: 10.1023/A:1021169731325

CITATIONS READS

152 12,381

1 author:

Monirul Mirza
University of Toronto
48 PUBLICATIONS 4,404 CITATIONS

SEE PROFILE

All content following this page was uploaded by Monirul Mirza on 24 January 2015.

The user has requested enhancement of the downloaded file.

 Natural Hazards 28: 35–64, 2003.
35
© 2003 Kluwer Academic Publishers. Printed in the Netherlands.

Three Recent Extreme Floods in Bangladesh: A

Hydro-Meteorological Analysis

M. MONIRUL QADER MIRZA

Adaptation and Impacts Research Group (AIRG), The Institute for Environmental Studies (IES),
University of Toronto, 33 Willcocks Street, Toronto ON M5S 3E8, Canada (E-mail:
[Link]@[Link] also: mqmirza@[Link])

(Received: 15 December 2001; accepted 4 March 2002)

Abstract. Bangladesh is highly vulnerable to floods due to its geographical location at the deltas
of the Ganges, Brahmaputra and Meghna (GBM) rivers. About 92.5 per cent of the area of three
basins lies outside the boundaries of the country. More than 80 per cent of the annual precipitation of
Bangladesh occurs in the monsoon period between June and September. The hydro-meteorological
characteristics of the three river basins are unique and they often cause large to extremely large
floods in Bangladesh. It is possible that these floods could inundate 70 per cent of the country and
the physical damage could be very serious for the economy of Bangladesh with its low gross domestic
product (GDP). In 1987, 1988 and 1998, Bangladesh experienced three extreme floods, leaving trails
of devastation and human misery. In this article it is demonstrated that these floods differed in terms
of magnitude, extent, depth and duration. The external and internal hydro-meteorological dynamics
were also different.

Key words: extreme floods, Bangladesh, Ganges, Brahmaputra and Meghna, hydro-meteorological
characteristics and ENSO.

1. Introduction
Normal floods are important in the life of the people and economy of Bangladesh.
People in rural and urban areas of Bangladesh have adapted to the changes in-
troduced by normal floods, and are highly resilient. However, this resiliency is in
jeopardy if an extreme flood event occurs. Such events are due to the unusual beha-
viour of the monsoon precipitation regime over the GBM river basins, compounded
by drainage dynamics usually controlled by tidal inflow and backwater effects.
Bangladesh’s location in the GBM basins is unique (Figure 1). The country con-
tains only 7.5 per cent of the three river basins, while the remaining 92.5 per cent is
located in China, Nepal, India and Bhutan (Mirza, 1997). The precipitation regimes
of the three basins are significantly different, varying from low to very high areas
of precipitation (Ahmed and Mirza, 2000). Very heavy precipitation during the
monsoon in the upstream area generates a huge amount of runoff that flows through
Bangladesh and drains to the Bay of Bengal through a narrow passage. Therefore,
over bank spilling is a common occurrence causing floods in Bangladesh, where
36 M. MONIRUL QADER MIRZA

Figure 1. The Ganges, Brahmaputra and Meghna basins. Dates of onset of monsoon are also
shown.

Figure 2. Causes of floods in Bangladesh.

local precipitation within the country aggravates the situation. In addition to in-
tense monsoon precipitation, other factors that may cause floods include: snow
and glacier melt, El Nino Southern Oscillation (ENSO) induced conditions, loss of
drainage capacity due to the siltation of principal distributaries, backwater effect,
unplanned infrastructure development, deforestation and the synchronisation of
flood peaks of the major rivers (Ahmed and Mirza, 2000) (Figure 2).
Extreme floods in 1987, 1988 and 1998 engulfed as much as 70 per cent of
Bangladesh (Figure 3). Economic damage and human misery, especially in rural
areas, was enormous. These floods generated local and international debates on
flood control in Bangladesh (Boyce, 1990). Thirteen donor countries formed a
THREE RECENT EXTREME FLOODS IN BANGLADESH 37

Figure 3. Area (× 1000 [Link]) flooded in various years in Bangladesh.

consortium which funded 27 studies under the ‘Flood Action Plan (FAP)’ (Mirza,
1991). Some analyses of the meteorological aspects of the 1987 and 1988 floods
were carried out (UNDP, 1989; Brammer, 1990; Hofer, 1998) but contributions of
external precipitation regimes were not adequately analysed. Detailed meteorolo-
gical analysis of the recent 1998 flood in Bangladesh did not appear in the scientific
literature, and a comparative analysis of the hydrological aspects of these three
extreme floods did not get proper attention. Therefore, to fill the knowledge gaps,
an attempt is made in this article to examine the external and internal precipitation
regimes associated with the three extreme floods; and the hydrological aspects of
the floods in terms of magnitude, depth and duration.

2. Monsoon Precipitation in the GBM Basins

Monsoons have wide ranging effects on life, society, hydrology and water re-
sources, and the economy of major river basins in South Asia. On a long-term basis,
there is a remarkable degree of regularity in the occurrence of the monsoon, but
there are large variations in the amount of precipitation in time and space (Rogers
et al., 1989). These variations may cause either floods or droughts in the river
basins. The onset of the Indian monsoon in terms of rainfall occurs over the Bay of
Bengal in late May, and then over India about 1 June (Figure 1). The retreat of the
monsoon from India occurs typically during September.
The first spell of the monsoon rain occurs around June 1 in the Meghna basin
over north-eastern India and Bangladesh. Among the three river basins, the average
precipitation (2600–3700 mm) of the Meghna basin is much higher than the other
two, attributed to the topographical effects of the north-eastern part of India. For
example, some of the world’s highest precipitation totals have been recorded at
Cherrapunji on the southern slopes of the Meghalaya plateau, where the moist air
38 M. MONIRUL QADER MIRZA

from the Bay of Bengal suddenly rises to 1,200 m (Rao, 1981). The copious precip-
itation in the Meghna basin in India has remarkable consequences for flash floods
in the north-eastern region of Bangladesh, but local precipitation also contributes
to this process. The monsoon completely retreats from the Meghna basin by 15
October.
The moist air mass of a monsoon arrives in the Brahmaputra basin in India
between 1–5 June in a typical year. Before the onset of the monsoon, there is
considerable thunderstorm activity over the basin in the month of May due to the
incursion of moisture in the region from the neighbouring Bay of Bengal (Dhar and
Nandargi, 2001). The Bangladesh part of the basin experiences monsoon rainfall
about a week later, between 5–10 June. Average precipitation in the basin area
in India and Bangladesh is 2350 mm and 1900 mm, respectively. In Bhutan, the
average monsoon precipitation varies from less than 400 mm in the high Himalayas
to 2000–4000 mm in the southern foothills. The monsoon retreats from the basin
between 1–15 October.
Due to the vastness of the basin, the monsoon arrivals in the Ganges basin
in India range between 5 June–1 July. The first spell of monsoon rain occurs by
5 June around Kolkata in India and in the adjoining area in Bangladesh. In the
northern part of the basin area around Delhi, monsoon rain occurs around 1 July.
The average monsoon precipitation in the Ganges basin in India varies from 360
mm on the western end to about 800 mm in the central area and to 1200–1600 mm
near the delta (Mirza, 1997). In Nepal, monsoon precipitation is about 1500 mm.
Rainfall in the Terai region of Nepal in the south along the Indian border is higher
than the northern region. The monsoon retreats from north to south of the basin
from 1 September–15 October.

3. Characteristics of Flood Flows in Bangladesh

The flood flows of Bangladesh rivers are unique in terms of timing of occurrence,
magnitude, duration and severity. They are highly influenced by the precipitation
patterns of the drainage areas outside the border and within Bangladesh.

3.1. TIME OF PEAK FLOW

Rivers in Bangladesh peak in different months during the monsoon in response
to onset dates and precipitation characteristics. The flood hazard in the central
region (around the Baruria Transit) can increase significantly if floods arrive sim-
ultaneously in the Ganges and Brahmaputra rivers. Assessment of the timing of
the river peaks is thus important in order to provide estimates of the magnitudes
of discharge, flooded areas and damage. An analysis of occurrences of floods in
some major rivers and their tributaries was made so that the months with greatest
potential for generating simultaneous floods could be identified. For the analysis,
monthly maximum discharges for various periods were compiled from FEC (1989)
THREE RECENT EXTREME FLOODS IN BANGLADESH 39
Table I. Occurrence of maximum monthly peak discharge (in %)

River/Station June July August September Number of

peak years

The Brahmaputra family Brahmaputra 9 41.0 35.0 15.0 34.0

(Bahadurabad)
Teesta (Kaunia) 0 58.0 37.5 4.5 24.0
Old Brahmaputra 0 12.5 75.0 12.5 8.0
(Mymensingh)
The Ganges family Ganges 0 0 45.0 55.0 53.0
(Hardinge Bridge)
Padma (Baruria) 0 19.0 57.0 24.0 21.0
Brahmaputra family∗ 5 44.0 41.0 10.0
Weighted Average 0 5.0 48.0 47.0
Denotes weighted average (in terms of occurrence) distribution. Indented rivers are tributary and
distributary.

and BWDB (1995). Table I Shows that the peaks in the Brahmaputra family
of
rivers occur in July and August and in the Ganges family of rivers, in August and
September. Maximum discharge in the Brahmaputra occurs 35 per cent of the time
and for the Ganges occurs 45 per cent of the time in August. This indicates a real
likelihood of simultaneous floods in the Ganges and Brahmaputra rivers in August.
The weighted average of these two families of rivers also indicates a high possib-
ility of flooding in August. The similarity in the time of occurrence of peaks in the
Brahmaputra and Teesta rivers shows that they drain areas with similar climatic
patterns.

3.2. DISTRIBUTION OF LAG - TIMES OF ANNUAL PEAKS

An assessment of lag-times between the yearly peaks of the two major rivers helps
determine the possibility of simultaneous flood peaks. In this regard, a 28-year
period common to Bahadurabad and Hardinge Bridge was analysed and presented
in Table II showing that the occurrence of “simultaneous” (assumed to occur within
one day period) peaks in the Ganges and Brahmaputra rivers is not a rare event. On
average, the Brahmaputra peaks occur 29 days earlier than the Ganges. However,
the lag-times are well scattered. The distribution shows one group (0–10 days) that
has occurred 22 per cent of the time over a 28-year period.

FEC (1989) defined the family of rivers according to the peaking time of discharge.
40 M. MONIRUL QADER MIRZA

Table II. Distribution of lag-times between yearly peaks of the Ganges River at Hardinge Bridge
and Brahmaputra River at Bahadurabad

Time-Lag(days) Number (cumulative) Percentage distribution Percentage (Cumulative)

−82 to −72 – 3.57 –

−71 to −24 2 7.14 7.14
−23 to −14 3 3.57 10.71
−13 to −1 4 3.57 14.28
0–10 10 21.43 35.71
11–30 12 7.14 42.85
31–40 18 21.43 64.28
41–50 21 10.72 75.00
51–60 25 14.28 89.28
61–70 26 3.57 92.85
71–88 28 7.15 100.00

A minus sign indicates the Ganges peaks occurred first. The shaded area indicates “period of
simultaneous occurrence of floods.

3.3. CRITICAL PEAK PERIODS FOR THE GANGES AND BRAHMAPUTRA RIVERS
The next important issue is the identification of peak periods for the Ganges
and Brahmaputra rivers. From records of the peaks (a sample of 28 years was
considered), two specific periods have been identified for the Brahmaputra.

3.3.1. Period One

A first flood occurs in the Brahmaputra basin at around day 128 (August 7) of the
water year (1 April to 31 March) with the normal occurrence of the Brahmaputra
peak. In response to the precipitation and basin characteristics, the Brahmaputra
River generates a peak in advance of the Ganges River.

3.3.2. Period Two

The late Brahmaputra peak arrives at around day 153 (August 31) of the water
year, which overlaps with the average Ganges peak that occurs at around day 157
(September 4). In some years, the annual peak occurred during this second period.
During period two, the likelihood of a simultaneous occurrence is very high. Peaks
in the Teesta and Meghna rivers occur simultaneously because they drain catch-
ments with similar characteristics. This indicates the possibility of simultaneous
flooding in all river basins is higher with the late Brahmaputra peak. In a given
year, the Brahmaputra may experience two flood peaks. For example, a second
peak may occur in August, with serious implication if it coincides with the normal
Ganges peak.
THREE RECENT EXTREME FLOODS IN BANGLADESH 41

Figure 4. Relationship between depth of water above danger level and recession time for the
Brahmaputra River at Serajgonj. Hpeak = Peak water level (m + PWD) and Hdanger = Danger
level (m + PWD).

3.4. DURATION OF FLOOD DISCHARGE

The duration of a flood is important in terms of damage and recovery, and is a

function of inflow of discharge, drainage capacity of the river channel, backwater
effect, tidal activity, etc. Although many of these factors have synergetic effects
on flood duration, a simple relationship between depth of water above a critical
level and recession time of various floods are plotted in Figure 4 for the Serajgonj
station on the Brahmaputra River. With the increases in flood level, recession time
increases exponentially. Other major rivers show similar behaviour.

3.5. SEVERITY OF PEAK DISCHARGE

Flood volume alone does not explain the severity of flood per unit area of the basin.
The severity can be assessed by determining the regional coefficient ‘K’, using
the Francou-Rodier relationship (FEC, 1989). The coefficients are calculated using
predicted peak flows of different return periods for several rivers and arranged in
decreasing order Table III.
The K values (Table III) indicate that, per unit area, the severity of flooding in
the Brahmaputra River basin is the highest of all the basins. The basin areas of the
Teesta and Meghna are small, but severity of flooding is substantial because of the
precipitation influence of Sub-Himalayan West Bengal and the Assam ranges. The
K value for the Meghna shows the damping effect of the Meghna depression (FEC,
1989). The Padma is a hybrid of the Ganges and Brahmaputra rivers. Its K value is
strongly influenced by the Brahmaputra River.
 42

Table III. Francou-Rodier’s regional coefficients

River/Station Basin Area (103 [Link]) K (10) K (20) K (25) K (50) K (75) K (100) Remark

Brahmaputra (Bahadurabad) 536.0 5.23 5.38 5.42 5.56 5.63 5.68 1

Padma (Baruria) 1490.0 4.79 4.97 5.02 5.19 5.28 5.33 3
Teesta (Kaunia) 12.3 4.51 4.65 4.69 4.80 4.86 4.90 1
Meghna (Bhairab Bazar) 60.7 4.54 4.65 4.75 4.78 4.82 4.85 1
Ganges (Hardinge Bridge) 907.0 4.14 4.32 4.37 4.52 4.61 4.67 2

1. Brahmaputra Regime. 2. Ganges Regime. 3. Brahmaputra-Ganges Regime. Francou-Rodier’s Coefficient (K) is determined from
the relationship Q/106 = (A/108 ) ∗ Exp (1 − K/10); Q in m3 /sec and A in [Link].
M. MONIRUL QADER MIRZA
THREE RECENT EXTREME FLOODS IN BANGLADESH 43

Figure 5. Return periods of peak discharges of the GBM rivers.

3.6. RETURN PERIODS OF PEAK DISCHARGE AND WATER LEVEL

The return period of an extreme event is inversely related to its frequency of oc-
currence, and very severe events occur less frequently than more moderate events.
A frequency analysis was carried out to relate the magnitude of extreme events to
their frequency of occurrence through the use of probability distributions. The ana-
lysis demonstrates how frequently these three events may occur. This analysis has
substantial implications for design of water control structures and their economic
values, delineation of flood plains and determining the effect of encroachments on
the flood plains.
For flood frequency analyses, the Gumbel Type I distribution (EV1) has been
recommended for the major rivers in Bangladesh (FAP, 1992). Accordingly, the
EV1 distribution was applied for estimating return periods of peak discharge and
water level for the Ganges, Brahmaputra and Meghna rivers and presented in Figure
5 and Figure 6. The EV1 probability distribution function is:

 
x−u
F (x) = exp − exp − −∞≤x ≤∞ (1)
α

where F (x) is the probability of an annual maximum Q ≤ x and α and u are loc-
ation parameter and scale parameter, respectively. Defining s as standard deviation
and x̄ as mean, the expressions for u and α are as follows:
√
6s
α= (2)
π
44 M. MONIRUL QADER MIRZA

Figure 6. Return periods of peak water levels of the GBM rivers.

u = x̄ − 0.5772α. (3)

With a finite sample, the mean and standard deviation can be estimated from the
moments of the data sample. The probability model can be obtained by substituting
the value of α and u in Equation (1). Figure 5 and Figure 6 present the magnitude
of discharges and water levels of the Ganges, Brahmaputra and Meghna rivers
for various return periods. Due to the relatively high coefficients of variations of
discharge and water level for the Ganges River, the magnitude of uncertainty for
return periods is higher than that of the Brahmaputra and Meghna rivers.

4. The Extreme Floods of 1987, 1988 and 1998: Hydro-Meteorological

Analysis
Three extreme floods devastated Bangladesh in the year 1987, 1988 and 1998,
when 36, 63 and 69% area of the country was engulfed (Figure 3). The floods
occurred in different months in response to high monsoon rainfall in Bangladesh
(rainfall stations are shown in Figure 7) and cross-border basin areas. Floods in
1987 (Figure 8) began in late June when flash floods occurred in the north-eastern
districts of Sylhet and Sunamgonj (BWDB, 1987) as a result of high rainfall. In
late July, the Brahmaputra and Ganges exceeded their danger levels at Chilmari
and Goalundo, respectively (Figure 7). The Meghna exceeded its danger level at
Brahrab Bazar in early August. The floods of 1988 (Figure 9) began when with
some of the rivers in the South-eastern hill basins crossing their danger levels in
early May. The Brahmaputra River remained in flood stage from early July and
had two other peaks prior to the catastrophic flood of August–September. The late
THREE RECENT EXTREME FLOODS IN BANGLADESH 45

extreme peak in the Brahmaputra coincided with the moderately high peak in the
Ganges River with a time difference of only three days. The Meghna flood reached
the highest level at the end of first week of September. The flood in 1998 (Figure
10) started in the late second week of July and continued till the middle of Septem-
ber and in some places the whole of September (Chowdhury, 2000). Districts in
the northern and central parts and a part of South-west Bangladesh were inundated
mainly due to the onrush of flood waters from India. The flood situation worsened
in the three river basins in the first week of September. Magnitudes of the peak
discharge of the major rivers were different during the three flood years. Flood
depths were also different and floods remained differing periods in the upstream
and downstream areas.

4.1. THE PRECIPITATION REGIMES

Occurrence of rainfall in the Indian subcontinent is correlated with El Nino South-
ern Oscillation (ENSO) (Miyakoda et al., 2000). A study of the Indian monsoon
has shown that there is a good correlation between the strong Southern Oscillation
Index (SOI) (cold phase) or “La Nina” and good monsoon years (Parthasarathy and
Pant, 1985). Similar research on Bangladesh monsoon rainfall (Ahmed, 1999) has
shown that in general there is a decrease in rainfall in El Nino years (warm phase)
in all three seasons - the pre-monsoon, monsoon and post- monsoon. The SOI
is a measure of difference in pressure anomaly between Darwin in Australia and
Tahiti Island situated in the eastern Pacific. When this index is positive, it reflects
a cold phase of sea surface temperature in the southern Pacific Ocean. During an
El Nino event (warm phase) this index is negative. During an El Nino period there
is an abrupt rise of temperature in the eastern Pacific Ocean and as a result the
atmospheric circulation in the region changes. The easterly wind from the Pacific
to the Bay of Bengal, which brings a lot of moisture, reverses or weakens. As a
result, Bangladesh and other parts of the Indian subcontinent receive less rain dur-
ing an El Nino year. On the other hand, when the temperature in the eastern Pacific
decreases, the easterly wind becomes strong, the flow of moisture is abundant and
there is increased rainfall in Bangladesh and neighboring regions, resulting heavy
flooding.

4.1.1. 1987 Precipitation Regime

The year 1987 was an El Nino year (Figure 11). Krishnamurti et al. (1987) sum-
marized the monthly meteorological conditions and sea surface temperature from
April to August of 1987 on a global scale. They concluded that the planetary scale
divergent circulation at 200 mb shifted eastward and equatorward, due to the El
Nino related warm sea surface temperature over the equatorial Pacific, and that
this eastward shift of the divergent circulation in the upper troposphere caused
severe drought over most of the Indian subcontinent, and flooding in northeast India
and Bangladesh. In India, the monsoon season’s rainfall departure was −19% and
46 M. MONIRUL QADER MIRZA

Figure 7. Flood monitoring and rainfall stations of the BWDB.

the area under deficient rainfall was 64% (Kulshrestha, 1997). Overall monsoon
rainfall across Bangladesh was significantly above the long-term mean (Table IV
and Figure 7).
In 1987, severe floods occurred in the Brahmaputra basin in eastern India
and Ganges basin in Bihar due to a break monsoon,2 one of the components of
2 Rao (1976) defined the break monsoon as follows: the monsoon trough is located close to the
foot of Himalayas which leads to a striking decrease of rainfall over most of India but an increase
along the Himalayas and parts of northeast India and southern Peninsula. Dhar et al. (1984) showed
that the increase in rainfall during break monsoon periods is at maximum at the southern foothill of
Himalayas between longitudes 85◦ and 87◦ E around 27◦ N.
THREE RECENT EXTREME FLOODS IN BANGLADESH 47

Table IV. Monthly rainfall (mm) in Bangladesh during 1987 flood

Basin/Station July August September

Normal 1987 Deviation Normal 1987 Deviation Normal 1987 Deviation

Ganges Basin
Panchagarh 836 1149 313 568 1238 670 276 809 533
Dinajpur 455 947 492 318 843 525 283 415 132
Rajshahi 334 405 71 241 424 183 221 235 14
Pabna 281 332 51 278 491 213 218 169 −49
Khustia 311 306 −5 292 491 199 200 115 −85
Jessore 329 533 204 309 751 442 239 159 −80
Khulna 370 570 200 301 666 365 249 168 −81
Faridpur 337 424 87 300 375 75 244 433 189
Barisal 481 433 −48 384 798 414 332 237 −95
Brahmaputra Basin
Kurigram 455 870 415 297 607 310 296 311 15
Kaunia 504 1555 1051 318 644 326 341 270 −71
Rangpur 443 1378 935 329 525 196 280 396 116
Bogra 401 582 181 321 734 413 237 214 −23
Dewangonj 460 1062 602 318 535 217 244 425 181
Jamalpur 440 865 425 363 702 339 253 604 351
Mymensingh 460 408 −52 390 517 127 310 407 97
Tangail 283 336 53 286 356 70 211 341 130
Dhaka 360 541 181 333 464 131 234 340 106
Meghna Basin
Sylhet 721 1290 569 599 735 136 439 953 514
Sunamgonj 1410 1629 219 1079 690 −389 596 1701 1105
Sheola 762 793 31 552 454 −98 413 781 368
Moulvi Bazar 452 584 132 391 490 99 269 209 −60
Habigonj 402 548 146 393 443 50 255 276 21
Durgapur 852 941 89 608 659 51 526 629 103
Bhairab Bazar 432 305 −127 325 474 149 266 485 219
Comilla 492 305 −187 369 445 76 263 357 94
Chandpur 380 771.4 391.4 320 527 207 235 251 16
South Eastern Hill Basin
Parshuram 557 768 211 414 595 181 227 294 67
Noakhali 629 1287 658 595 836 241 419 696 277
Rangamati 573 783 210 465 902 437 260 425 165
Chittagong 680 1303 623 568 616 48 294 515 221
Panchpukuria 645 1189 544 462 668 206 234 487 253
Bandarban 612 647 35 424 796 372 270 353 83
Lama 777 1263 486 533 926 393 315 397 82
48 M. MONIRUL QADER MIRZA

Figure 8. Extent of flood in 1987. Upazila denotes the lowest administrative district.

interseasonal variation of monsoon activity. These two regions are close to the
Bangladesh border and parts of the shared river basins. In Assam in eastern India
the area engulfed was the second highest since 1953 (3.08 million ha) (CSE, 1992).
Precipitation in the bordering meteorological subdivisions of South Assam and Sub
Himalayan West Bengal in the Bramhaputra basin was notably higher than normal.
A similar precipitation pattern was noticed in the Gangetic West Bengal and Bihar
plain meteorological subdivisions in the Ganges Basin (Table V and Figure 12).
The cross-border flood waters generated from the excess precipitation eventually
moved into Bangladesh.
THREE RECENT EXTREME FLOODS IN BANGLADESH 49

Figure 9. Extent of flood in 1988. Upazila denotes the lowest administrative district.

Before the peak time of flow of the main rivers in 1987, precipitation across
the basin areas in Bangladesh filled all water bodies and low-lying areas. For
example, in July rainfall in the four basins within Bangladesh was much higher
than the normal (Table IV). Most noticeable was the 100% higher rainfall in the
Brahmaputra basin than the long-term mean. The Ghagot and Teesta sub-basins of
the Brahmaputra basin also experienced huge rainfall. The Ganges basin (down-
stream of the Hardinge Bridge) in Bangladesh experienced torrential rainfall in
August, which was twice the normal (Table IV). In 1987, the Ganges attained two
flood peaks – one in August and another in September. The Brahmaputra discharge
peaked in August. The Ganges as well as the Brahmaputra peaks were not cor-
50 M. MONIRUL QADER MIRZA

Figure 10. Extent of flood in 1998.

related with the rainfall processes in Bangladesh, thus only the rainfall situation
outside Bangladesh can explain the situation (Hofer, 1998). The extraordinary peak
of the Teesta from 10–16 August coincides with the peak of the Brahmaputra that
caused floods in Northwestern Bangladesh. But the rainfall that occurred in the
region was not sufficient to explain this situation. The Meghna reached its peak in
the middle of August. Overall, local rainfall in Bangladesh throughout the monsoon
months aggravated the flood situation in the four basins (Table IV).
THREE RECENT EXTREME FLOODS IN BANGLADESH 51

Figure 11. The southern oscillation index (SOI).

Figure 12. Deviations (%) of rainfall from normal in some meteorological subdivisions in the
GBM basins in India.

4.1.2. 1988 Precipitation Regime

In 1988, abundant rains occurred in the GBM basins area with the “La Niña” event.
As seen from the SOI index in Figure 11 that from September 1987 onwards, the
central equatorial pacific began retreating from the warming phase to the colder
phase and entered into the colder phase in March 1988. By July the “La Niña”
phase began and continued until December 1988.
52 M. MONIRUL QADER MIRZA

Table V. Monsoon rainfall (mm) in various meteorological subdivisions of the GBM rivers in India

Meteorological subdivision Normal 1987 Deviation 1988 Deviation 1998 Deviation

Brahmaputra and Meghna basins

North Assam 1969 1738 −231 1546 −423 2553 584
South Assam 1247 1507 260 1787 540 1057 −190
Sub Himalayan West Bengal 2038 2665 627 2448 410 2017 −21
Ganges Basin
Gangetic West Bengal 1097 1272 175 1196 99 958 −139
Bihar Plateau 1078 1067 −11 963 −115 1167 89
Bihar Plain 1005 1585 580 1029 24 1164 159
East Uttar Pradesh 898 607 −291 903 5 1019 121
West Uttar Pradesh 774 447 −327 935 161 929 155
Haryana 513 142 −371 786 273 729 216
West Rajasthan 243 98 −145 272 29 221 −22
East Rajasthan 630 361 −269 590 −40 628 −2
West Madhaya Pradesh 922 684 −238 863 −59 885 −37
East Madhaya Pradesh 1145 945 −200 945 −200 854 −291

Source: IMD (2001).

There were two waves of floods in 1988 in Bangladesh, which were mainly
caused by runoff generated in the cross-border basin areas. During the first wave
of flood in July a storm crossed the Meghalaya, entered the Assam Valley and
caused extensive rainfall. During July 1–9, severe rainfall (2633 mm) was recor-
ded in Cherrapunjee (Hofer, 1998). The Teesta, Dharala and Brahmaputra rivers in
Bangladesh all exceeded crossed their danger levels. The Brahmaputra remained
in flood stage from early July and had two flood peaks on 10 July and 30 July.
However, the flood in the first part of July in the Meghna basin is clearly related to
the rainy phase (mid June-mid July) that was recorded at almost all stations across
the basin (Table VI). The annual peak occurred at Bhairab Bazar station on 11 July.
The second wave of flood hit Bangladesh in late August and early September.
Some rainfall occurred in the northern and northwestern part of the country, but
the rains were not adequate to cause such a huge flood (Hofer, 1988). Therefore,
an important cross border input has to be assumed in addition to the high rainfall
in some parts of Bangladesh itself. Very high rainfall was reported in the Sub
Himalayan West Bengal and Gangetic West Bengal meteorological subdivisions
in the latter part of August (GOI and GOB, 1990), leading to the peaking of the
flow in the Ganges River at Hardinge Bridge on 3 September. Rainfall in the basin
area in Bangladesh was also substantially higher (Table VI). In the 1988 flood,
the Brahmaputra River at Bahadurabad reached its peak on August 30. During the
week August 18–24 and August 25–31, precipitation levels in Aurunachal Pradesh,
Assam and Meghalaya were on average 336% and 76% higher than normal (GOB
THREE RECENT EXTREME FLOODS IN BANGLADESH 53

and GOI, 1990). Precipitation in Sub-Himalayan West Bengal was 176% above
normal, which contributed to the peak of the Teesta River (26 August, 1988).
Therefore, in 1988, excess precipitation in the Brahmaputra basin in India contrib-
uted significantly to the flood peak at Bahadurabad, Bangladesh. Some contribution
from Bangladesh was also observed. In August, rainfall in Kaunia and Rangpur
was 130 and 84% higher than normal. Downstream of Bahadurabad, rainfall in
Dewangonj was 30% and Jamalpur 36% higher than normal which contributed to
flooding in the areas downstream of Bahadurabad. In August, high rainfall was
recorded in the stations upstream of Bhairab Bazar in the Meghna basin (Table
VI and Figure 7). In five out of eight BWDB stations used for flood forecasting,
rainfall was 44 –132% higher than average (Table VI). However, downstream of
Bhairab Bazar, rainfall was slightly less than the normal. In September, rainfall in
three upstream stations was 29–83% higher.

4.1.3. 1998 Precipitation Regime

In 1998, a “La Nina” event similar to that of 1988 brought huge rains to the region
and was the main cause of floods in Bangladesh (Chowdhury, 2000a, b) (Figure
11). In 1998, part of the Ganges basin in India experienced higher rainfall than
normal in Bihar Plateau, Bihar Plain, East Uttar Pradesh and West Uttar Pradesh
from late August to the middle of September (Table V and Figure 12). The external
contribution to the flood discharge of the Ganges River in Bangladesh came from
Bihar and Uttar Pradesh. Upstream of the Brahmaputra Basin in India, high rainfall
was concentrated in the North Assam meteorological subdivision and a significant
departure is noticeable (Figure 12). In 1998, Dhar and Nandargi (2001) reported 14
flooding events at Dhubri gauging site on the Brahmaputra River, which eventually
reached downstream to Bangladesh.
In the main flooding period (late August–middle of September), rainfall in the
Ganges Basin was generally normal except for three stations (Table VII). Three
stations on the left bank of the Ganges River experienced lower than normal rain-
fall. Therefore, local contribution to flooding in the Ganges basin was negligible in
1998. In August, rainfall in the Brahmaputra basin in Bangladesh was also signific-
ant. In the Teesta basin, rainfall was 32–122% higher than normal, contributing to
the peak discharge of the Teesta River, which was already high due to the external
contribution from India. Peak discharge of the Teesta River merged with the inflow
of the Brahmaputra River from India and generated 102,534 m3 /sec flood peak
at Bahadurabad, the highest ever in the recorded history. Stations downstream of
Bahadurabad also experienced high rainfall (42–97% higher the normal) contrib-
uting to flooding in the central region of Bangladesh (Table VII). The external
contribution to the flood discharge of the Meghna River was not significant in
1998. In the main flooding period, local rainfall in the Meghna basin contributed
significantly to flooding. The basin area upstream of Bhairab Bazar experienced
rainfall 15–71% higher than normal (Table VII), raising the flood peak of the river
at Bhairab Bazar.
54 M. MONIRUL QADER MIRZA

Table VI. Monthly rainfall (mm) in Bangladesh during 1988 flood

Basin/Station July August September

Normal 1988 Deviation Normal 1988 Deviation Normal 1988 Deviation

Ganges Basin
Panchagarh 836 823 −13 568 1341 773 276 481 205
Dinajpur 455 489 34 318 893 575 283 198 −85
Rajshahi 334 343 9 241 363 122 221 84 −137
Pabna 281 320 39 278 181 −97 218 180 −38
Khustia 311 300 −11 292 274 −18 200 128 −72
Jessore 329 277 −52 309 422 113 239 94 −145
Khulna 370 343 −27 301 402 101 249 56 −193
Faridpur 337 227 −110 300 264 −36 244 190 −54
Barisal 481 429 −52 384 492 108 332 194 −138
Bramaputra Basin
Kurigram 455 666 211 297 723 426 296 333 37
Kaunia 504 438 −66 318 730 412 341 391 50
Rangpur 443 474 31 329 608 279 280 523 243
Bogra 401 586 185 321 272 −49 237 281 44
Dewangonj 460 538 78 318 407 89 244 317 73
Jamalpur 440 376 −64 363 494 131 253 267 14
Mymensingh 460 657 197 390 399 9 310 406 96
Tangail 283 404 121 286 292 6 211 ∗ ∗

Dhaka 360 253 −107 333 168 −165 234 187 −47
Meghna Basin
Sylhet 721 723 2 599 1152 553 439 604 165
Sunamgonj 1410 1726 316 1079 2514 1435 596 1093 497
Sheola 762 505 −257 552 809 257 413 621 208
Moulvi Bazar 452 386 −66 391 667 276 269 207 −62
Habigonj 402 584 182 393 567 174 255 331 76
Durgapur 852 920 68 608 937 329 526 502 −24
Bhairab Bazar 432 439 7 325 334 9 266 291 25
Comilla 492 640 148 369 346 −23 263 239 −24
Chandpur 380 332 −48 320 276 −44 235 156 −79
South Eastern Hill Basin
Parshuram 557 907 350 414 471 57 227 344 117
Noakhali 629 526 −103 595 894 299 419 631 212
Rangamati 573 286 −287 465 521 56 260 ∗ ∗

Chittagong 680 798 118 568 414 −154 294 224 −70
Panchpukuria 645 632 −13 462 520 58 234 448 214
Bandarban 612 579 −33 424 420 −4 270 430 160
Lama 777 472 −305 533 438 −95 315 349 34

∗ Not available.
THREE RECENT EXTREME FLOODS IN BANGLADESH 55

Table VII. Monthly rainfall (mm) in Bangladesh during 1998 flood

Basin/Station July August

Normal 1998 Deviation Normal 1998 Deviation

Ganges Basin
Panchagarh 836 1154 318 568 903 335
Dinajpur 455 686 231 318 555 237
Rajshahi 334 363 29 241 261 20
Pabna 281 422 141 278 275 −3
Khustia 311 429 118 292 241 −51
Jessore 329 213 −116 309 279 −30
Khulna 370 373 3 301 309 8
Faridpur 337 312 −25 300 351 51
Barisal 481 661 180 384 524 140
Brahmaputra Basin
Kurigram 455 533 78 297 661 364
Kaunia 504 612 108 318 420 102
Rangpur 443 477 34 329 616 287
Bogra 401 745 344 321 504 183
Dewangonj 460 684 224 318 617 299
Jamalpur 440 439 −1 363 517 154
Mymensingh 460 796 336 390 767 377
Dhaka 360 515 155 333 562 229
Meghna Basin
Sylhet 721 1347 626 599 779 180
Sunamgonj 1410 1591 181 1079 1841 762
Sheola 762 2143 1381 552 780 228
Moulvi Bazar 452 622 170 391 236 −155
Habigonj 402 577 175 393 453 60
Durgapur 852 978 126 608 751 143
Bhairab Bazar 432 613 181 325 292 −33
Comilla 492 813 321 369 328 −41
Chandpur 380 736 356 320 538 218
South Eastern Hill Basin
Parshuram 557 831 274 414 577 163
Chittagong 680 1247 567 568 1306 738
Panchpukuria 645 1203 558 462 1046 584
Bandarban 612 880 268 424 690 266
56 M. MONIRUL QADER MIRZA

4.2. THE DISCHARGE , PERIOD OF OCCURRENCE , WATER LEVEL AND

DURATION REGIMES

Peak discharge values and corresponding water levels in the major rivers are the
major contributors to riverine flooding in Bangladesh. However, when a peak dis-
charge value is assessed in terms of external and internal runoff contributions,
location of the gauging station must be taken into account. The stations at Hardinge
Bridge (Ganges River) and Bahadurabad (Brahmaputra River) are close to the
Bangladesh-India border (Figure 1). Basin areas above these two stations within
Bangladesh are small and it is unlikely that these small parts of basins contribute
significantly to generate peak discharges of the Ganges and Brahmaputra rivers.
However, the Bhairab Bazar station on the Meghna River is located far from the
border and receives significant amounts of runoff from the area within Bangladesh.

4.2.1. Peak Discharge

The peak discharge of the Ganges River in 1987 at Hardinge Bridge was
the highest ever at that station. The computed peak discharge at this station
was 76,000 m3 /sec, which exceeded the previously recorded highest flow of
73,200 m3 /sec in 1961. The peak flow for the Brahmaputra River at Bahadurabad in
1998 was 102,534 m3 /sec, the highest ever, slightly more than the previous record
of 98,600 m3 /sec in 1988. Note that while the Ganges River drains a basin area
twice the size of the Brahmaputra basin, the magnitude of peak discharge of the
Brahmaputra River is greater than the Ganges River because average monsoon
rainfall in the later is twice that of the former. The highest peak discharge of the
Meghna River at Bhairab Bazar occurred in 1988 (19,800 m3 /sec), slightly higher
than the previous recorded highest flow of 19,500 m3 /sec in 1974. Magnitudes of
peak discharge and water levels and their return periods are tabulated in Figure 5
and Figure 6 which shows that the Brahmaputra River produces the most extreme
floods followed by the Ganges River.

4.2.2. Period of Occurrence

One important aspect of the 1987 flood was that the peak flows of the Ganges
and Brahmaputra did not coincide. For the Brahmaputra River, it was about period
one as described in Section 3. While the Ganges peak occurred 15 days later (19
September) than the average peaking date September 4 (Figure 13). Peak discharge
in the Ganges and Brahmaputra occurred 34 days apart and the likelihood of such
occurrence is 21.5% (Table II). The pattern was similar to the floods in 1954 and
1974 (Matin, 1987). The Meghna peak occurred on 8 August, just 4 days prior to
the Brahmaputra peak. This implies similar rainfall patterns in the two river basins
in 1987.
In 1988, the peak discharges of the three rivers occurred about the same time
(Figure 13). For the Brahmaputra River, the date of occurrence of peak discharge
falls around period two (see Section 3). As indicated before, during period two, the
THREE RECENT EXTREME FLOODS IN BANGLADESH 57

Figure 13. Date of occurrence of flood peaks.

likelihood of a simultaneous occurrence is very high. The Ganges peak occurred

just two days earlier than the average peaking date (Figure 13). Peak discharge in
the Ganges and Brahmaputra occurred only three days apart. The combined flow
of the two rivers slowed the outflow from the Meghna River at Chandpur. The high
floods on the Ganges and Brahmaputra rivers, their similar times and the backed
up flood water in the Meghna basin caused a severe flood in Bangladesh.
In 1998, the times of occurrence of the peak discharges of the Ganges,
Brahmaputra and Meghna rivers (Figure 13) had two important aspects. First,
they occurred in the same month in September. Second, flood peaks in the Ganges
and Brahmaputra rivers occurred only a day apart, an occurrence, which had not
happened before in recorded history (Figure 13). The time of occurrence is unusual
for the Brahmaputra, as only 15% of the flood peaks occurred in September (Table
I). The Meghna flood peak also occurred on 7 September. The simultaneous oc-
currence of flood peaks in the three rivers caused the severe flood, which engulfed
69% of the country.

4.2.3. Water Levels

In 1987, all the monitoring stations in the Ganges basin reported exceedence of
danger levels. This implies severe flooding in the Ganges basin. In the Brahmaputra
basin, water levels were above the danger level in 7 stations out of 10. In the
Meghna basin the situation was as serious as the Ganges basin. In the South Eastern
Hill Basin, water levels exceeded danger levels in all stations. Overall, in 92%
of the monitoring stations (31 out of 34) of the four basins, flood level exceeded
danger levels (BWDB, 1987).
58 M. MONIRUL QADER MIRZA

In 1988, the water level in all but one monitoring station on the Ganges River
crossed the danger level. Although peak discharge was slightly lower than in 1987,
the monitoring stations Hardinge Bridge, Goalundo, Bhagyakul, Gorai Railway
Bridge and Faridpur registered higher water levels than in the previous year, per-
haps caused by deposition of sediments from the high floods of 1987. Due to very
high flood flows in the Brahmaputra River and in its tributaries, water levels ex-
ceeded respective danger levels in all monitoring stations as did the Meghna basin
also registered water levels above danger levels. A similar situation was reported
on the South Eastern Hill Basin. In total, water levels in 97% of the monitoring
stations exceeded danger levels, demonstrating the severity of flood in 1988 in
Bangladesh.
The flood situation of 1998 became worse on July 28 when 23 stations monit-
oring several rivers were reported flows above their respective danger levels (Islam
and Chowdhury, 1999). Steady or little improvement was observed between the
end of July and the middle of August. The flood situation deteriorated again with
all the stations in the Ganges basin exceeding their danger levels and became severe
on August 30, when 24 stations in all basins were above their danger levels. After a
slight improvement for about a week, the situation worsened again due to the syn-
chronisation of the flood peaks through the upstream of Brahmaputra and Ganges
rivers, and became extremely severe on September 7 when 25 monitoring stations
had flows above the danger level (Islam and Chowdhury, 1999).

4.2.4. Depth of Floods

Analysis of depth of flooding reveals some interesting features for the major rivers.
The depths of flood for the Ganges and Brahmaputra rivers were almost equal
in 1987. The maximum depth of flood water at Hardinge Bridge on the Ganges
was 1.3 meters (calculated by subtracting ground level3 from the water level).
At Bahadurabad, the maximum depth of flood was 1.28 m, while flood depth for
the Meghna was only 0.4 m at Bhairab Bazar. However, higher flood depths were
observed at various upstream stations in the Meghna Basin (BWDB, 1987).
In 1988, maximum depth of flooding in the Ganges River at Hardinge Bridge
was estimated to be 1.37 m. This was 7 cm higher than the flood depth of 1987
although the peak discharge was smaller. A staggering 2.22 m flood depth was
observed for the Brahmaputra River at Bahadurabad. The depth of flooding demon-
strates the severity of flooding in 1988 in the Brahmaputra basin. Depth of flooding
for the Meghna at Bhairab Bazar was approximately 1.16 m, which was 0.8 m
higher than in 1987, caused by the high peak discharge of the Meghna River. The
Meghna River at Chandpur registered a maximum 1.86 m flood depth.
Although in 1998, the magnitude of flood peak of the Ganges River at Hardinge
Bridge was smaller than that of 1988, the recorded water level was higher Figure 5.
3 Ground level of the Ganges and Brahmaputra at Hardinge Bridge and Bahadurabad is 13.5 +
PWD and 18.4 m + PWD, respectively.
THREE RECENT EXTREME FLOODS IN BANGLADESH 59

This was perhaps caused by aggradation/change in the channel cross-section of the

river. Therefore, depth of flooding increased to 1.69 m compared to 1988. On the
other hand, despite the highest ever discharge in 1998, flood depth in Brahmaputra
at Bahadurabad was 1.97 m, which was 25 cm smaller than in 1988. Degradation
in the river caused smaller flood depth, otherwise the flood situation would have
been much worse. These phenomena demonstrate that morphological conditions
also control the flooding process in the large rivers in Bangladesh. Flood depth in
the Meghna River at Bhairab Bazar was 1.08 m, against a lower peak discharge
than 1988.

4.2.5. Flood Duration

The duration of floods is important with regard to damage and recovery. In 1987,
the duration of water levels above the danger levels in all stations of the Ganges
River exceeded previous records mainly due to the highest ever peak discharge.
One spectacular aspect is that water levels at Hardinge Bridge, Goalundo and
Bhagyakul stayed above respective danger levels for equal periods of time (Table
VIII). This was probably caused by a constant rate of drainage downstream. Fig-
ure 3 illustrates the relationship between depth of water above danger level and
recession time. As the water levels were significantly high, it took a long time for
them to recede to below the danger level. For example, the danger level at Chilmari
for the Brahmaputra stayed for 24 days. This was caused by the backwater effect at
the confluence of the Brahmaputra and the Teesta rivers. A similar backwater effect
at the confluence of the Ganges and Brahmaputra rivers caused the water level to
remain above danger level at Serajgonj for 41 days. The Kushiyara (upstream of
the Meghna) at Sheola and the Meghna at Bhairab Bazar remained above danger
levels for almost an equal period of time probably caused by a constant rate of
drainage.
In 1988, the duration of floods above danger levels at various monitoring
stations in the Ganges basin was lower than that of 1987 due to a smaller peak dis-
charge (Table VIII). Once the flood peak was reached, recession occurred promptly.
However, in the downstream stations, a slow rate of recession was observed. For ex-
ample, while the flood level was only seven days above the danger level at Hardinge
Bridge in September, it stayed 17 and 18 days at Goalundo and Bhagyakul, respect-
ively. The duration of floods was almost equal for the Brahmaputra at Bahadurabad
for the months of August and September (9 and 11 days, respectively). In total
at Bahadurabad, the Brahmaputra remained above the danger level for 27 days
(Table VIII and Figure 14). The duration of flooding for the Meghna River was
spectacular. At Bhairab Bazar it stayed above the danger level for 68 days, which
was significantly higher than the other two rivers (Table VIII and Figure 14). The
floods ended in the Ganges and Brahmaputra rivers by the middle of September,
while flooding in the Meghna basin continued until the end of September. This was
caused by backwater effect and build up of water in the downstream reaches of the
combined rivers. A spring tide in conjunction with a solar eclipse on 11 September
60 M. MONIRUL QADER MIRZA

Table VIII. Duration (in days) of floods above the danger

levels in 1987, 1988 and 1998

River Station 1987 1988 1998

Ganges Rajshahi 29 24 28
Hardinge Bridge 55 23 27
Goalundo 54 41 68
Bhagyakul 56 47 72
Gorai Gorai Railway Bridge 43 25 25
Brahmaputra Chilmari 24 15 22
Bahadurabad 13 27 68
Serajgonj 41 44 48
Aricha – 31 68
Teesta Kaunia 42 38 –
Meghna Bhairab Bazar 30 68 68
Surma Kanairghat – 75 73
Kushiyara Sheola 27 80 37

considerably delayed the recession of flood water from the southern part of the
country (Hofer, 1998).
The most significant feature of the 1998 flood was the long duration of floods
above danger levels in all the major rivers. In terms of duration, the floods of
1998 surpassed all past records (Ahmed and Mirza, 2000). (Table VIII and Figure
14). From the middle of August to the middle of September, flood levels were
consistently above danger level on the Ganges at Hardinge Bridge. The case of
the Brahmaputra and Meghna was the worst ever. For most of July and from the
middle of August to the middle of September, a total of 68 days, flood levels of the
Brahmaputra were above danger levels at Bahadurabad (Table VIII and Figure 14),
while for the Meghna, it was 68 days (from the middle to July to late September).
The duration of the flood in one upstream station of the Meghna River was a week
longer than Bhairab Bazar.
What caused such a long duration of floods in the three rivers? The flood peaks
of the Ganges and Brahmaputra rivers occurred only 24 hours apart. The com-
bined flow had difficulty passing through the narrow cross-section of the Padma
River downstream of Goalundo, which often causes a backwater effect that can
be observed about 30 kilometers upstream of the confluence of the Ganges and
Brahmaputra rivers (RPT-NEDECO-BCL, 1993). Lunar position (full moon) and
its strong influence often cause high tides and a subsequent backwater effect. The
spring tide during the monsoon of 1998 was much higher than usual, causing
tremendous backwater effect and eventually inhibiting drainage. Sea level in the
northern Bay of Bengal had played a role in flooding. In 1998, sea-level height
THREE RECENT EXTREME FLOODS IN BANGLADESH 61

Figure 14. Water level and flood duration in the GBM rivers in Bangladesh for 1988 and 1998.
62 M. MONIRUL QADER MIRZA

in the northern Bay of Bengal was 0.33 m higher than in 1997 (Peter J. Web-
ster, University of Colorado, personal communications). The compounding effect
was ‘stagnation’ of water between the Ganges and Meghna confluence points.
Consequently, a vast area in the central and north-eastern parts of Bangladesh
was submerged for an unusually long time, resulting in the deluge of the century
(Ahmed and Mirza, 2000).

5. The Three Extreme Floods: Conclusion

Basinwide monsoon rainfall patterns control flood peaks of the Ganges,

Brahmaputra and Meghna rivers. The progression of monsoon air mass is from the
southeast part of the GBM basin to the northwest. Therefore, the Brahmaputra and
Meghna basins experience their first flood peaks in July, while the Ganges basin
usually receives intense rainfall in August- September that leads to its flood peak.
However, the Brahmaputra and Meghna rivers may also experience high rainfall at
the same time of the Ganges River in the upstream and downstream areas.
Along with local rainfall, rainfall in the basin areas adjacent Bangladesh
generally control flood peaks and flooding process. Local rainfall generally ag-
gravates the flooding situation in the downstream areas subject to timing of rainfall
occurrences.
Rainfall in a drier basin generally has higher variability than in a wetter basin.
Among the three river basins, the variability of rainfall in the Ganges basin is
higher than in the Brahmaputra and Meghna basins. This variability causes high
uncertainty in the peak discharge of the Ganges basin than the other two basins.
This has implications for flood damage and mitigation planning in the Ganges
basin.
Due to rainfall patterns in the basins and the time of occurrence of flood peaks,
synchronisation of flood peaks may happen in Bangladesh although not very often.
However, the flood peaks of the Ganges and Brahmaputra rivers coincided twice
in the case of the three extreme floods analysed in this paper, and led to extreme
flooding in Bangladesh in 1988 and 1998.
The narrow cross-section of the Padma River (the combined the Ganges and
Brahmaputra rivers) cannot quickly drain the combined flood volume, which may
be as high as 180,000 m3/sec. Observations show that the drainage of flood flow
is further slowed due to higher sea-level in the northern Bay of Bengal, and the
high tide caused by lunar activity. This is particularly important for flooding in the
central part and Meghna basin in north-eastern Bangladesh.
High flood levels in the main rivers indicate a longer flooding period in
Bangladesh. However, many other factors such as constant supply of flood flow,
coincidence of flood peaks, tidal activity, sea-level, etc. control flood duration. The
synergic effects of these factors on flood duration are difficult to estimate, and call
for a detailed analysis.
THREE RECENT EXTREME FLOODS IN BANGLADESH 63

Acknowledgements
I appreciate comments from David Etkin and Marie Sanderson of the Adaptation
and Impacts Research Group (AIRG) and two anonymous reviewers on an earlier
draft of the paper. Thanks to Farzana Abdulhusein of the AIRG for redrawing most
of the figures.

References
Ahmed, A. U. and Mirza, M. M. Q.: 2000, Review of causes and dimensions of floods with particular
reference to flood’98: National perspectives. In: Q. K. Ahmad, A. K. A. Chowdhury, S. H. Imam
and M. Sarker (eds.), Perspectives on Flood 1998, University Press Ltd., Dhaka.
Ahmed, R.: 1999, Monsoon prediction in Bangladesh using early pre-monsoon interior-to-coast
temperature gradient index (TGI) and southern oscillation index (SOI). Paper presented at
the Conference on Numerical Weather Prediction (NWP), Jadavpur University, Kolkata, India,
December 30, 1999.
BWDB (Bangladesh Water Development Board): 1987, Flood in Bangladesh – 1987, BWDB,
Dhaka.
Boyce, J. K.: 1990, Birth of a megaproject: Political economy of flood control in Bangladesh,
Environmental Management 14(4), 419–428.
BWDB (Bangladesh Water Development Board): 1995, Discharge Data for Various Rivers in
Bangladesh, BWDB, Dhaka.
Brammer, H.: 1990, Floods in Bangladesh I. Geographical background of the 1987 and 1988 floods,
The Geographical Journal 156(1), 12–22.
Chowdhury, A. M.: 2000a, Flood, 98: Oceanic perspective, In: Perspectives on Flood 1998, In: Q. K.
Ahmad, A. K. A. Chowdhury, S. H. Imam and M. Sarker (eds.), University Press Ltd., Dhaka.
Chowdhury, M. R.: 2000b, An assessment of flood forecasting in Bangladesh: The experience of the
1998 flood, Natural Hazards 22, 139–163.
CSE (Centre for Science and Environment): 1992, Floods, Flood Plains and Environmental Myths,
CSE, New Delhi.
Dhar, O. N. and Kamte, P. P.: 1973, Probable maximum precipitation over the Brahmaputra basin in
Assam, Irrigation and Power Journal 30(3), 237–241.
Dhar O. N. and Nandargi, S.: 2001, A comparative flood frequency study of Ganga and Brahmaputra
river systems of north India-a brief appraisal, Water Policy 3, 101–107.
Dhar, O. N., Soman, M. K., and Mulye, S. S.: 1984, Rainfall over the southern slopes of the Hi-
malayas and the adjoining plains during ‘breaks’ in the monsoon, Journal of Climatology 4,
671–676.
FEC (French Engineering Consortium): 1989, Prefeasibility Study for Flood Control in Bangladesh
– Vol. 2: Present Conditions, FEC, Paris.
FFWC (Flood Forecasting and Warning Centre): 1998, Danger Levels and Flood Duration
([Link]
FAP (Flood Action Plan) 25: 1992, Flood Hydrology Study, Flood Plan Corrdination Organisation
(FPCO), Dhaka.
GOI (Government of India) and Government of Bangladesh (GOB): 1990, Indo-Bangladesh Task
Force on Flood Management Report, GOI and GOB, New Delhi and Dhaka.
Hofer, T.: 1998, Floods in Bangladesh: A highland-lowland interaction? Institute of Geography,
University of Berne, Switzerland.
IITM (Indian Institute of Tropical Meteorology): 1989, Probable Maximum Precipitation Atlas,
IITM, Pune, India.
 64 M. MONIRUL QADER MIRZA

IMD (Indian Meteorological Department): 2001, Subdivisionwide Monsoon Precipitation, IMD,

New Delhi.
Islam, S. M. and Chowdhury, M. R.: 1999, 1998 flood in Bangladesh-Obscure and eventualities
of nature, Proceedings of IDNDR Symposium on Mitigation of Water Related Disasters held at
Nagoya, Japan, February 16–18.
Islam, S. R. and Dhar, S. C.: 2000, Bangladesh Floods of 1998: Role of Flood Forecasting & Warning
Centre, BWDB, Dhaka.
Krishnamurti, T. N., Bedi, H. S., and Subramaniam, M.: 1987, On the Summer Monsoon of 1987,
Tallahassee, Florida.
Kulshrestha, S. M.: 1997, Drought Management in India and Potential Contribution of Climate Pre-
diction, Center for Ocean-Land-Atmosphere Studies and Center for the Application of Research
on the Environment, Maryland, USA.
Matin, M. A.: 1987, Some aspects of 1987 flood. Presented in Seminar Floods in Bangladesh, held
in the Institution of Engineers, Bangladesh (IEB), Dhaka, Bangladesh.
Matin, M. A. and Husain, M. A.: 1988, Hydrological aspect of 1988 flood, Presented in Sem-
inar Floods in Bangladesh, held in the Institution of Engineers, Bangladesh (IEB), Dhaka,
Bangladesh.
Matsumoto, J.: 1988, Synoptic features of heavy monsoon rainfall in 1987 related to the severe flood
in Bangladesh, Bulletin of the Department of Geography 20, 43–56.
Mirza, M. M. Q.: 1991, Flood action plan of Bangladesh – The Embankment Issue, Water Nepal
2(2/3), 25–28.
Mirza, M. M. Q. and Paul, S.: 1992, Pakritik Durjog O Bangladesher Paribesh (in Bengali), Centre
for Environmental Studies and Research (CESR), Dhaka.
Mirza, M. M. Q.: 1997, Modelling the Effects of Climate Change on Flooding in Bangladesh, Un-
published PhD. Thesis, International Global Change Institute (IGCI), University of Waikato,
Hamilton, New Zealand.
Miyakoda, K., Kinter, J., and Yang, S.: 2000, Analysis of the Connection from the South Asian
Monsoon to ENSO by Using Precipitation and Circulation Indices, COLA Technical Report
90. Center for Ocean-Land-Atmosphere Studies, Calverton, USA.
Parthasarathy, B. and Pant, G.B.: 1985, Seasonal relationships between Indian summer monsoon
rainfall and the southern oscillation, J. Climatol. 5, 369–378.
Popeleswski, C. F.: 1988, The global climate for June-August 1988: A swing to the positive phase
of the southern oscillation, drought in the United States, and abundant rain in monsoon areas, J.
Climate 1(11), 1153–1174.
Rao, Y. P.: 1976, Southwest Monsoon, New Delhi, India.
Rao, Y. P.: 1981, The climate of the Indian subcontinent, In: K. Takahashi and H. Arakawa
(eds.), Climate of Southern and Western Asia. World Survey of Climatology, Vol. 9, Elsevier,
Amsterdam.
Relief Web: 2001, India-Floods and Landslides OCHA Situation: Report No. 3 ([Link]
[Link]/websites/[Link]/86d6cf0398b9c2fc125650a00512b.
Rogers, P., Lydon, P., and Seckler, D.: 1989, Eastern Waters Study: Strategies to Manage Flood and
Drought in the Ganges-Brahmaputra Basin, ISPAN, Virginia, USA.
RPT-NEDECO-BCL: 1993, Jamuna Bridge Study, Phase II Study, Feasibility Report, Annex B: River
Morphology, Jamuna Multipurpose Bridge Authority, Dhaka.
Schneider, B.: 1996, Drei Fallstudien von Niederschlagsereignissen der Uberschwemmungsjahre
1987/88 in Bangladesh, Eine Analyse von Wolkenstrukturen auf NOAA-Satellitenbilder und
Klimatologisschen Daten, Unpublished Masters Thesis, Department of Geography, University
of Berne.
UNDP (United Nations Development Programme): 1989, Bangladesh Flood Policy Study-Final
Report, UNDP, Dhaka.

View publication stats