I. Bangladesh and Flood

Flood and Nature

Bangladesh has been much in the news of late as a dramatic illustration of the devastating long term threat posed by global warming and a consequent rise in sea level. Consider the following elevation map of the region courtesy of the Center for International Earth Science Information Network (CIESIN) at Columbia University:

Elevation Map of Bangladesh

Map of Bangladesh elevations and population densities

Bangladesh and adjacent parts of India share two important geographical qualities with the US state of Florida: they are adjacent to the sea and remarkably flat. As can be seen here, most of Bangladesh is within 20 meters of sea level, and a great deal of it is within 10 meters. Bangladesh’s extreme flatness is the consequence of the majority of the land surface being laid down as an enormous series of overlapping river deltas. The descendants of these rivers are still there. The Brahmaputra/Jamuna, Ganges and Megna drain 1,600,000 km/sq—an area more than ten times the size of Bangladesh. This drainage includes a significant part of the Himalayas and parts of India, China, Tibet and of course Bangladesh. Coursing through a flat land without river valleys or bedrock to constrain them, the power and force of these rivers is at times simply astonishing. So flat is the majority of Bangladesh that rises in sea level brought on by future global warming may some day submerge much of the southern part of the country and expose much of the rest to devastating storm surges.

In the meantime, and as far back as people have lived on the land, what is now Bangladesh has been one of the most flood prone in the world. It is a pure function of geography that Bangladesh would experience ‘normal’ annual floods. Bangladesh rivers have very low gradients (the measure of the fall of a river over distance)—in the case of the Jamuna only 0.000077—and as as rivers increase their flows in summertime, already wide channels fill and then extend out locally from there. It is hard to know how large normal monsoon flooding was in the distant past, but in recent times it has covered as much as 20% of the country. Statistically, geography and climate also dictate occasional ‘abnormal’ huge floods: abnormal only in the sense that they come rarely. Some of these like the ones in 1988 and 1998 cover more than half the country (68% in the case of 1998) and stay for weeks.

Flood and People

That such normal and abnormal floods are sometimes locally and nationally catastrophic is a function of changing human adaptations to these geographical facts. Three hundred years ago and before, population densities in what is now Bangladesh were much lower than today, and much of the country was still bush. A hundred and fifty years ago tigers roamed within 50 km of Dhaka:

Dhaka gravestone of a man killed by a tiger in 1836

The gravestone of John Demetrius Ellias, killed by a tiger 25 miles from Dhaka, 1836

Gravestone located at the TSC, University of Dhaka

Lower population densities allowed the agriculturally-dependant people of the region to adapt to flood in many ways. Land availability was less constrained than today, and this allowed farmers to produce consistent crops and a stored reserve surplus, at least when the latter was not appropriated by rapacious landlords, bandits and rulers. The core crop was rice (as it remains), which provided high yields per hectare, and where relevant, people locally developed a range of flood adapted varieties. As elsewhere, localized flooding also added to the land's fertility. Land availability was such as to allow homesteads to be raised up enough to at least lessen the disruptions of local flooding and to provide a livable safe haven during high water.

Traditional Bangladesh homestead during flood

A traditional homestead during flood. Note how the land around the house is raised more than a meter

Traditional homestead and boats, Bangladesh

Traditionally, homesteads all had boats to use during monsoon season, as well as for general transportation

During times of flood people also could turn their efforts to fishing:

Fishing during flood time, Kazipur, Bangladesh

Even today there are many fishing techniques employed in Kazipur (see our captioned fishing photo essay).

This one depends on channeling fish during flood and is carried out much as it was in 1900:


Traditional homestead and boats, Bangladesh

Fish Traps in Bengal, from L.S.S. O'Malley, Bengal, Bihar and Orissa, Sikkim (1917)

Moreover, there is every reason to believe that ‘abnormal’ flooding was less in earlier times, both in areal extent and length. Even fifty years ago much of the lower slopes of the Himalayas that were ancestrally forest were still so, and this lessened runoff and reduced the incidence of sudden releases of water into downstream rivers. Before 1900 almost nowhere were riverbanks constrained by dykes (called embankments in Bangladesh), as the Mississippi is today. The historical lack of embankments increased the likelihood of local floods but decreased the chance of catastrophic ones. Bush and swampland soaked up localized flood waters like sponges, lessening the probability of violent increases in water level downstream.

At 154 million, Bangladesh’s current population is only a bit less than half that of the United States and is comparable to that of Russia. All these people live on a land base of 134,000 km/sq or 52,000 miles/sq. That is roughly the same size as Greece or the US states of Alabama and Iowa. The average population density for the whole country is an astonishing 1,000 per km/sq or 2,600 per mi/sq. This compares with roughly 140 per km/sq in China, 350 per km/sq in India and 340 per km/sq in Japan.

With respect to the consequences of flood, this changes everything. Almost two thirds of Bangladeshis remain dependant on agriculture, and roughly 80% are rural. Farming areas in many parts of the country today have higher population densities than do typical North American residential suburbs.

Remarkably, through enormous effort and with little outside assistance beyond the critical introduction of high yield rice (in most places also introduced by local people) the rural people of Bangladesh have made the country self sufficient in food today—but just barely, and for one third of the people, inadequately. This has been achieved through a radical transformation of agriculture and with it, rural life and the ecology. There has been an enormous expansion in farmed land, such that more than one half of the whole country is now farmed. It is probable that virtually all easily exploitable land was already being farmed a hundred years ago. This expansion has come at the expense of forest and bush, which have disappeared over wide swaths of the countryside, along with all the attendant biological diversity. A combination of local demand and regional marketing opportunities have decimated riverine fish. The intensity of agriculture has also radically increased. Paddy is sown on over 80%  of the arable land, often with multiple cropping. Indeed, the country as a whole is approaching an average of two crops per year on farmland. As fields are subdivided again and again through inheritance, high yield varieties of rice developed internationally have of necessity supplanted lower yielding, but locally adapted ones. An ever growing percentage of the farmers have too little land to support themselves through agriculture alone, and send family members out into an increasingly swollen population of farm laborers and clothing factory workers. Everywhere, a significant minority of rural people—perhaps 40% of the rural population nationally—are entirely or functionally landless.

All this has changed settlement patterns. More people both farm and live on radically flood prone land. Of necessity, houses are located were possible rather than where optimal.

Today, even highly localized flooding can be economically destabilizing to many. Large ones lead to incalculable misery. Data summarized by Tony Beck for the World Bank show that the 1988 flood affected 45,000,000 people residing in 7,200,000 houses, and resulted in two million tons of lost rice production, over a billion dollar loss to the economy, and 2,400 deaths. Infrastructure such as roads, embankments, bridges and culverts were destroyed in their thousands. The flood of 1998 wrought a similar dose of misery and destruction. The country was hit again by massive flooding in 2004.

II. Riverbank Erosion

Many people outside Bangladesh are at least dimly aware of the significance of flood there. What is not so generally known is that millions of people in Bangladesh suffer from another consequence of the country’s unique mix of environmental and social factors: riverbank erosion. The flat delta lands of Bangladesh offer little resistance to the hydraulic forces of its rivers, particularly during periods of high flow. As a consequence, in many regions rivers such as the Jamuna run wide rather than deep: easily 10 km wide at places during high water, yet rarely in channels with sides deeper than two or three meters.

An eroding riverbank that threatens part of a village with destruction, Bangladesh

An eroding riverbank that threatens part of a village with destruction.

See our captioned photo essay on erosion.

Rivers are often “multiply braided”, which is to say that they divide and subdivide into a number of parallel channels separated by riverine islands (called chars).

A char (riverine island) village at low water, Bangladesh

A char (riverine island) village at low water

The term ‘riverbank erosion’ does not adequately convey the processes at work when Bangladesh’s rivers swell with water and attack their banks. It is not simply a nuisance issue of small, localized soil loss in the sense one might think when guided by typical Western examples. Episodically, the whole bank of a river can shift in one direction or another for a number of kilometers—and sometimes, it can shift quickly. We have personally seen the most westerly channel of the Jamuna move a quarter of a mile further west in little more than a year. To get a good visual image of riverbank erosion, watch rainwater trickling down a window. Often you will see these little, apparently stable streams quickly shift position, moving right or left on the window pane without any apparent reason; sometimes they will split or recombine just as easily. Bangladesh’s mighty rivers can do much the same thing, but on a massive scale. Like the water running down a window pane, the force and direction of riverbank shifting can be capricious and relentless. It is also unstoppable, given the resources of Bangladesh people and government. Unlike the window pane example, river shifting is both a destroyer and creator: destroying all before it, but often accreting new land elsewhere.

At least five million people in 20% of Bangladesh’s upazilas (local administrative regions) are at risk of losing land—and perhaps home and village—from riverbank erosion at any one time. Among them, at least 100,000 are forcibly displaced each year. An unknown but probably comparable number ‘voluntarily’ shift residence to try to make use of newly accreted land.

Map of areas most affected by erosion, Bangladesh

A Bangladesh Water Development Board map of the regions of Bangladesh most severely affected by riverbank erosion


Kazipur and Riverbank Erosion

Kazipur Upazila, Sirajganj District attracted our attention as social anthropologists in the late 1980s. At the time, we were both based in the Department of Anthropology of the University of Lethbridge, in Lethbridge, Alberta, Canada. By then we had much prior research experience with international migration to North America (chiefly by people from South and Southeast Asia): the key topical areas we concentrated on were settlement and community structure, race relations, the migration of political refugees, institutional responses to social issues, and the gendered analysis of migration. 

Our research in Kazipur was directly inspired and greatly facilitated by pioneering work done there on riverbank erosion done by teams from the University of Manitoba (Winnipeg, Canada) and Jahangirnagar University (Savar, Bangladesh). Their research was done in the mid-1980s as part of the IDRC‑funded Riverbank Erosion Impact Study (REIS); see Elahi, Haque, Rogge, Weist, Zaman, REIS, etc. in References & Links. Kazipur Upazila was selected because it was then facing very extensive riverbank erosion, and had been for some time. While most of the REIS studies were geographical (many, specifically hydrological) in focus, REIS participants carried out an extensive household survey of displaced people living on a local flood control embankment, and Mohammad Zaman carried out significant anthropological fieldwork on local politics as part of his doctoral research.

Because the REIS provided a wealth of baseline information on a subject then scarcely touched upon elsewhere in Bangladesh, and because REIS contributors in both Canada and Bangladesh were so helpful and supportive, we decided to build on their work. We initiated this with a rigorous fieldwork-based study of environmentally displaced people in Kazipur Upazila during 1989-90.

Map of Sirajganj, Bangladesh

A map of Sirajsanj District showing Kazipur. Much of the land eastwards of the site marked "Kazipur"

has been eroded away since the data for this map was collected

Map of Kazipur Upazila, Bangladesh showing flood control embankments

A map of Kazipur Upazila. The orange lines are flood control embankments

More specifically, most of this initial phase of research was carried out in and around Kazipur Union. Of Kazipur Upazila’s eleven unions, or administrative subdivisions, Kazipur Union (official 1981 population: 27,735; fewer than 10,000 actually in residence in 1990) was then among the worst erosion‑affected.(1) Once this was not so. On regional cartographic surveys done in the early 1920s the most westward branch of the multiply‑braided Jamuna was an insignificant one four miles east of its present course. Our informants claimed that the first major channel was then a five hour walk away. Kazipur was a stable, orderly, village-based mainland area (khiar), contrasting strongly with the rapidly shifting depositional char islands to the east and erosion-threatened (bir) lands immediately alongside the river. Between then and 1990, the nearest branch of the river grew and shifted westward at an annual rate of over 200 meters a year [Mafizuddin 1991:89 in References & Links], such that long ago it destroyed the town of Kazipur, the region’s major marketing and administrative center. The Union lost about 40 per cent of its agricultural land base (estimated loss, 2,000 acres), the Thana administrative headquarters and four villages during 1968‑81 alone [Haque 1988:189, ditto].(2) Floods during 1984 and 1988 resulted in further land loss and population displacement.(3) During our initial fieldwork (December 1989‑August 1990) erosion coincident with a flood covering nearly all the arable land in the Union took away several hundred additional acres and important schools and government buildings in the Union’s sole remaining town of Meghai.(4) Thereafter severe erosion continued during the 1991 and 1992 rainy seasons, destroying more of Meghai town and parts of two more villages. Although the 125 mile long Brahmaputra Right Bank Flood Control Embankment passes through the Union, its perhaps misguided purpose [Boyce 1990. ditto] is to stop annual flooding, not erosion. Breached repeatedly by the river, periodic embankment reconstruction each time destroyed and marginalized still more local land.

Flood control embankment during dry season, Bangladesh

A section of a local embankment during the dry season (December 1989). This part of the embankment had not

yet been occupied by people displaced by riverbank erosion. See our photo essays on abandoning land threatened

by erosion, and on resettling elsewhere.


Although the Union formally encompasses roughly eleven square miles, its total land base by the mid-1990s was less than a third of this – a rectangle two and a half miles north to south by a mile and a half wide. By then, more than half of the union’s people had been uprooted due to riverbank erosion during the past fifteen years, many several times.(5) On this, see our three captioned photo essays on the connected processes of abandoning homes, land and ways of life, shifting to another place, and resettling.


Google Earth map of Kazipur, Bangladesh

A Google Earth map of the area around Kazipur as it was a few years ago


As can be seen in the Google Earth image of Kazipur above (24.644’ N, 89.655’ E), the Jamuna locally is multiply braided and has created many chars. If you zoom in on the image above you might be able to see three spurs projecting into the river adjacent to Kazipur. These have been added fairly recently to control riverbank erosion. For reference, each of these is about 4-500 meters long.

Deposition and erosion around Kazipur continues to change the character of the land today. The most recent Google Earth images of the region (taken in 2009) show dramatic differences in just a few years:


Google Earth map of Kazipur, Bangladesh

A Google Earth map of the area around Kazipur in 2009

You can click on the title link of the map immediately above to open Google Earth and see the area displayed in this map in context.



Google Earth map of Kazipur, Bangladesh

A Google Earth closeup of Kazipur in December 2010

This is a schematized map of Kazipur and environs around 1990:

Schematic map of Kazipur Upazila, Bangladesh


The Kazipur Union area around 1990

Doreen Indra’s first fieldwork stay in Bangladesh was from September 1989 to August 1990, while Norman Buchignani was there during September-December 1989 and April-August 1990. During this period we were aided by two to three graduate research assistants. Further survey research and mapping was carried out by research assistants in 1991 and 1992, and Doreen Indra stayed for a short time to facilitate RA work during 1993.

During this initial phase of fieldwork our chief interests were specifically in the local social impact of environmentally forced migration, and on how forced migration impacted on women and women-headed households in particular. We also concentrated on landless migrants, especially on the hundreds of households resident for varying periods on local flood control embankments.

A recently settled part of the embankments, with the 'roadway' on top being used to dry fodder, Kazipur, Bangladesh


A recently settled part of the embankments, with the 'roadway' on top being used to dry fodder


This field research produced a great deal of raw data: over 3 MB of field notes, a number of topical surveys, a complete household survey of embankment residents, many maps, videos, and roughly 2,400 color slides, color negatives and black and white negatives.

III. Why do this now?

It has been a long time since this data was collected, and one might reasonably ask why we believe there is value in presenting some of it here, and in this form. Here are some of our reasons:

  1. Ethnography is not a kind of slow journalism with a predefined shelf life. Relevance depends on whether ethnographic data and analysis usefully contribute to a deeper understanding of people, places and processes, not on literal temporal currency. Ethnographic data collection done last year is not, in this sense, necessarily any more valuable than that carried out ten years ago.
  2. Despite its everyday impact, the sociocultural dimensions of riverbank erosion in Bangladesh continue to receive little academic or popular attention. We know of no example of long term anthropological fieldwork done specifically on these issues in the intervening years.
  3. There is every indication that many of the core challenges of rural people faced with riverbank erosion today have been around for a long time, and that people's responses to these challenges in Kazipur in 1990 can help to understand how people deal with these forces today.
  4. We feel an obligation to people in and around Kazipur to tell others about their stories, an obligation that we do not believe that we have fulfilled through our academic writing.
  5. We also feel a social and historical obligation to document places now long lost to riverbank erosion.
  6. We are not trying to do conventional ethnography on the Internet rather than in books. Neither are we specifically aiming at an academic audience. Instead, we hope to use a mix of academically-grounded information, people's own words and photographs to give a more wide-ranging audience some insight into riverbank erosion and rural Bangladesh.

IV. A Note on Specificity and Anonymity

One of anthropology's core ethical mandates mirrors that of physicians: whatever else, do no harm. This, as well as other key research requirements like informed consent have long been encoded in ethical guidelines for research created by anthropological organizations like the American Anthropological Association, universities and granting bodies. While we do not consider the content of this Web site to be anthropological research in the conventional sense, its content is based on anthropological research and we have therefore tried to conform to the spirit the discipline's ethical guidelines here.

As such, while we have made the decision to name the towns, villages, chars and other localities that we portray, pseudonyms are used to refer to all individuals except where we indicate otherwise. Where necessary we have also changed unimportant descriptive facts so that a person's identity cannot be inferred from what is said about them or from their own words.

Appropriately using photographs has presented a difficult ethical challenge to modern day anthropologists. Unlike text describing individuals, photographs of people cannot be rendered anonymous by something analogous to the use of pseudonyms. Also, while both texts and photographs can be read very differently by one person than another, the range of possible readings of photographs seems to be far greater than of text. It therefore is far more difficult to know up front whether a photograph has the potential to do harm. At the same time, many anthropologists (including us) feel an ethical requirement to advocate for individuals and groups, and photographic presentations can be powerful tools for advocacy.

We believe that the separation of nearly twenty years between when these photographs were taken and today lowers the risk that their presentation now will do unintended harm. Nevertheless, this potential does still exist, and we have decided not to use many otherwise informative photographs because of it. In addition, even though we know the names of many of the individuals portrayed we will not use them here. Finally, when individuals are portrayed in a given photograph we will refrain from telling any person-specific stories about the individuals portrayed, bur rather will use them to make more general points.



1. For a combination of political and legal reasons this official population estimate from the Bangladesh Bureau of Statistics [1988: Table 50] follows national census practice in including villages long ago destroyed by the Jamuna, as well as people who have ‘temporarily’ moved away to gain work. Using this method of accounting, Upazila officials estimated the 1987 Union population to be 31,734.

2. Over 18,000 people were displaced from affected unions in the upazila during 1974-81 [Elahi, 1991: 106].

3. In many erosion‑prone unions loss of land is at least partially compensated for by the accretion of new (albeit typically infertile and flood‑prone) land elsewhere. The mainland part of Kazipur Union gained virtually no such land during 1980-93, though lower rates of erosion in 1994 and 1995 were to signal a phase of increased land deposition.

4. Our analysis of Bangladesh Water Development Board maps shows that at some points the bank line shifted westward a quarter mile during 1989‑90, and over a full mile during 1979‑93.

5. Ahmed [1991: 144] reports that 67 per cent of Kazipur residents interviewed in 1985 had been displaced, 40 per cent more than three times. Haque's findings for the same period also indicate about two thirds had been displaced; similar information is presented online by Haque here.