Characterizing and relating variability in satellite images of the West African Sudano-Sahel to desertification and food security

Abstract

At the 7.6 km spatial scale in which remotely-sensed satellite imagery is used in many studies of subcontinental-scale vegetation vigor and dynamics, the information acquired has yet to be fully understood and integrated with ground-level reality. This dissertation reports results and analysis from ground-truth-sampling in the arid lands of West Africa's Sudano-Sahelian zones. The geographical locations of the transects investigated were obtained from areas exhibiting steep gradients in the interannual (1980-1994) coefficients of variation (CoV) of the mean annual monthly maximum composite of the Global Area Coverage's (GAC) Normalized Difference Vegetation Index (NDV1) from the Advanced Very High Resolution Radiometer (AVHRR) series of satellites. I begin this work by disaggregating the term "food security" into its various components, then continue by exploring what is generally understood by the concept of "desertification" and what this actually translates to in terms of land degradation. I then discuss how an error in NASA's method for calculating interannual NDVI CoVs impacted both my own work and our concepts of the Sahel's boundaries. Field data I gathered in the central and northern Sahel indicate that cogent, simple explanations of latitudinal variations in CoV do exist, albeit not everywhere. The Gourma region of Mali provides an excellent example of how complexity confounds any neat quantization of information. For the more southerly agropastoral zone, high CoV variability flags rapid, dynamic desertification processes. Results of village- and household-level profiles along a transect in the heart of Hausaland confirm that rapid, dynamic land degradation corresponds with a high interannual CoV. Climate, especially rainfall and potential evaporation, form the basis of an analysis the outcome of which explains how and why the Malian Gourma shows a nonlinear, "anomalous" NDVI response to rainfall. I also explore the strong correlation between rainfall and NDVI in the southern Sahel, but conclude that if there is a link between NDVI and crop yields, it is very weak indeed. Finally, my research highlights several policy measures that may retard desertification and enhance food security.