Long‐term monitoring of land cover changes based on Landsat imagery to improve hydrological modelling in West Africa

Abstract

The spatial and temporal variability of land cover changes is a fundamental parameter to integrate when modelling water resources in order to reproduce the relations between rainfall and surface flow more precisely. This is particularly important in West Africa, where the land cover has been changing for more than 40 years under the combined impact of climatic effects and human activities. In this study, we evaluated the potential of Landsat imagery to monitor the vegetation cover in the upper Niger watershed (120 000 km²) using archive images from MSS, TM and ETM+ sensors covering three periods of time around 1975, 1985, and 2000. Because of the heterogeneity of the acquisition dates, the spatial and spectral resolution of the images, and the scale of analysis, we chose a simple system of classification. Pretreatments were applied to reduce variations between the images. Vegetation indices (NDVI) were then calculated and subsequently thresholded using the same land‐cover classification system. The thresholds were then optimized by automated recursive calculations of confusion matrices and control parcels. Our results revealed that although the accuracy was not perfect, it was nevertheless possible to estimate changes using an unconventional spatio‐temporal scale. The resulting changes were characterized by a moderate trend to deforestation with a corresponding increase in bare soils, soils with sparse vegetation, and shrublands. The spatial layers produced were then combined with a soil map to incorporate changes in surface conditions in the hydrological modelling of the Niger River.

Acknowledgements

This work was based on the use of a number of Landsat images supplied free of charge by the Global Land Cover Facility (http://www.landcover.org) established by the USGS and NASA and hosted by the University of Maryland. It was funded in the framework of the ECCO‐PNRH programme ‘Assimilation de la variabilité spatio‐temporelle des pluies et des états de surface dans la modélisation des ressources en eau du fleuve Niger’.