Visiting the char people - September 2 - 4, 2003

One of the objectives of our project is to develop a method to automatically distinguish the bankline of a river on a satellite image. This is useful for several purposes: a time-series of the bankline of a particular stretch of the river can show the progression of bank erosion, and automatic analysis might be devised to predict that progression based on the remotely-sensed geometry of the river banks and islands (chars). The project orders image acquisition, say, from the Quick-Look or Radarsat satellite, by specifying the date, the location and other parameters, then the satellite operator programs the acquisition. We had ordered a September 1 image of one of our study areas, about 50 kilometers of the Jamuna River including a village called Shubhagachha which has been almost entirely consumed by erosion. To analyze the image, features in it have to be associated with actual features on the ground, in terrestrial coordinates like latitude and longtitude. Someone has to go there with a Geographical Positioning System (GPS) instrument, find the features and measure their positions. This should be done more or less at the same time as the image is acquired because in the flood season, with the river level changing, the feature shapes can change significantly and rapidly. So I went with a couple of GIS analysists, Imam and Saif, to survey parts of the bankline and chars.

We drove up to the Jamuna bridge then had to go way west almost to Bogra to approach the study area from a direction that has roads passable in the flood. Bangladesh is a big floodplain so roads are levees built up with material that doesn't resist erosion well. We hired a boat and set off up the river.

Left to right: the crew man steering, Imam, Saif and the boatman Totashek (listening to a cricket game, I guess). The yellow thing in the foreground is the GPS instrument.

Planting aman rice on a char.

Char kids gathering firewood.

Vestiges of an embankment. See how the river has entered behind it and covered the cultivations, depositing silt. This is good in the long run but if the silt is coarse, the land will be infertile for several years. The settlement in the foreground will have been consumed by the river in the next week or so.

Zooming in on the settlement in the previous picture.

With the fields gone, people resort to fishing.

The boatman.

We got off on a char and had the boatman walk around the perimeter with the GPS set to log his path.

The boatman.

Imam and Saif.

This guy might be fishing. The guy behind him seems to be in despair, or having a dump.


These guys are optimists, irrigating a cultivation that's meters away from being eroded. The bankline can move that far in a day.

More fishing.

Going to the jute market.

Preparing to land at the jute market to get more batteries for the GPS.

The jute market is on a char.

Who's the guy in the dress? (It's a lungi.) Imam took this, during a rain. He named the file "ghatok sid". When I asked him (in email) what "ghatok" means, he just said "matchmaker". I asked my office neighbor and colleague Mustafa why, and he laughed and explained that the ghatok is a cultural figure in the countryside that has taken on the caricature of a scheming avaricious person who strides under a ratty black umbrella among the villages, looking for opportunities to be the middleman in the arrangement of marriages.

Evidently these people weren't too keen on the ghatok coming by. Note that the flood rose into their house and fell back again. They're living on the edge.

At this point my camera ran out of power so I took no more photographs. From here we went about ten kilometers down the river to Shubhagachha, a village that Pepa and I visited on June 2, where she took this picture:

We had had 15 copies of the picture printed in Dhaka, and we wanted to give them to the women if/when we ever got back there. So now, in September, I was back, but the place was unrecognizable under a couple of meters of water. The embankment was gone, allowing the river to flood over everything and seriously erode the bankline to at least two hundred meters west of where we'd seen it in June. I know this because a landmark I'd hoped to recognize, a schoolhouse of concrete and brick two stories high then about 150 meters from the river, was gone. We motored around trying to find it, using a spar (a 100 meter-long concrete pier sticking staight out from the bank into the river, intended to protect the downstream bank from erosion) as a reference.

The spar was now isolated from the bank, forlorn and useless several hundred meters out in the river. At a likely spot we landed and showed some people the picture, expecting that they would recognize some of the faces. Someone did, saying, she is gone, ie, left the area because her home and land are lost. Erosion is a cruel phenomenon. We left the photos with a local constable who gets around the area a lot, and who said he'd try to distribute them.

If the erosion prediction component of our project is successful, in future the authorities will be able to influence settlement on land that has a high probability of erosion. Land prices will reflect the risk and people will be able to make informed decisions in their land purchases. The government could institute an erosion loss insurance system that compensates people displaced by erosion, on the basis of the bankline monitoring tools we're making. The social cost of these displacements is high because such people are financially ruined and often migrate to the cities to seek better opportunities.

They start at the very bottom of the social ladder, camping on the roadsides, are exploited by the canny city people, eke out livings as beggars, rickshaw drivers, garbage recyclers, brick-breakers and goat herders, find a spot in a shanty area, get evicted or burned out by arson, etc. The cities are bursting with people and garbage, approaching the limits of growth, over their capacity to provide water, power and transportation, maintain order and provide justice. Life seems to go on in any case, though.