Project Summary

This project proposes to use remote sensing images of 20m, 10m, and 5m resolution from 1982-1998 to document urban change in six cities located in and regions of Africa: Marrakech (Morocco), Dakar (Senegal), Bamako (Mali), Niamey (Niger), Dodoma (Tanzania), Gaborone (Botswana). The urban components of the imagery will be used both to compare urban change with land use/cover change in the hinterland and to create a set of classes defining types of urban places. These classes are to be understood as classifications of image pixels (based on their mathematical similarity on multiple wavelengths) which are then modified by ground truthing (people actually going and visually checking what is at the place to which the "classes" of pixels correspond). The computer generated classes are, finally, further modified by additional human inputs (e.g. physically delimiting a shanty town's border on
the image). The resulting classes correspond to our perceived reality but are readily quantifiable because the computer can count all pixels in them. Hence rates of change over time and change in location can easily be calculated and displayed (with a diachronic set of images such as recommended in this proposal).

The project team will use these classes to stratify each city into types of residential, commercial and public spaces. They will further stratify using changes in the size and location of classes between the 1980s and 1998 to define new transitional areas and old areas of each class. Using these criteria for residential areas makes it possible to weight the sample heavily toward the poorer sectors of each city. A grid system will be overlayed on the city image. Interview teams composed of an African supervisor and five African interviewers will number the thirty households closest to the center point of each selected grid cell (inside or outside the cell) and then select at random from that list ten households in which they will interview all adults. The project will do 200 household interviews per city oriented toward three general areas: The first focus will be the livelihood strategy histories of the urban population. A second focus will be on the relationships between natural resource use between 1982 and 1998 and changes in urban structure. The third focus will be the differential impact on the poor, of government multilateral agency, and NGO projects during the study period. For reasons of time, money, and moral urgency as well as the interests of the senior researchers the team have decided to emphasize the poorer segments of each city in the sampling strategy.

A final phase will be to create a GIS database incorporating all the processed imagery and the results of the household interviews which will be made available in French or English to all the African institutions associated with the project and placed on the internet for other researchers to use. The GIs is viewed primarily as a means to easily present the research which will, itself, contribute significantly to understanding relationships between land use/cover change in and environments and urban change as well as answer a host of hypotheses dealing with recent urbanization in Africa. There is reason to believe that the last few decades in and regions of Africa have produced a new poor who defy many of the old preconceptions about poverty in Africa and that recent urbanism in Africa may have significant breaks with the past. Donor agencies now have a presence few would have imagined thirty years ago and human adaptations to poverty have in considerable measure been shaped more by collective use of aid (in all its forms) than by individual zing behavior. Use of extended kin networks involve both need on one side and cultural generosity on the other. To the extent that these new radically impoverished urban dwellers do adapt differently to poverty, policy may need to be reoriented from a focus on individual nuclear households to larger social networks.