Search

During the last decades, the African continent has been experiencing population dynamics at rates which are considered to be among the strongest in human history. While Africa's population increased rapidly to currently 1.1 billion of people, it is expected to double until the year 2050 [1]. This human population increase goes in line with a strong urban growth, and urbanization rates of some African cities already rank among the highest in the world. In many African countries, this growth is additionally boosted by considerable migration from rural areas, and is concentrated on only few major cities, which results in uncontrolled proliferation of slums and urban poverty.

Quantifying the regional effects of land cover change is imperative to improve future hydrological budget estimates within large river basins. In this study we aim to utilize binary logistic regressions models within a cellular automation (CA) modeling environment to find causalities for satellite remote sensing measured land cover change (LCC). We used 30-meter Landsat and 250-meter MODIS time-series observations to map LCC for different time trajectories in two large study areas in Burkina Faso and Ghana. We used the FAO land cover classification system (LCCS) legend to map LCC processes from the satellite trajectories. Socio-economic data on population density, distances to roads, and biophysical data sets were processed in the CA model. The neighborhood effect of the change predictors were accounted for by using an enrichment factor. The relationship between the satellite derived LCC and the major biophysical and socio-economic drivers showed that population density, and the increase of cropland areas are responsible for the conversion of forests and woodlands. This was observed for both study areas.

Land cover information is an essential parameter for many earth system models and studies. However, especially in Africa, rigorous, regional standardized and geometrical spatially explicit land cover data on medium scale is missing, or often inconsistent and outdated. In this study we employed well corrected 250-meter MODIS time-series observations from 2006 and 2007 to map land cover rigorously over the Volta Basin in West Africa, with focus on the Burkina Faso and Ghana respectively. We used the FAO land cover classification system (LCCS) standard [8] for the legend codes and 15-30 ASTER imagery as reference data to train the classification tree algorithm. Primary aim is to contribute to a continental African and standardized medium scale land over data base, with improved mapping accuracies. Therefore, the dataset presented here can be deemed interoperable that is between local scale studies, at finer resolutions, and 1-kilometer global land cover products.