Towards water stress indicators in sub-Saharan West-Africa using high resolution optical satellite data

Abstract

Sub-Saharan Africa, particularly the Sahel region, has been found to be vulnerable to climate change and variability, and has suffered severe droughts in the mid-1970s, mid-1980s and the last ten years, among them in 2012. Precipitation falls almost exclusively during the rainy season between May and October. Particularly in drought years, water stress situations (here considered as critical lack of available water for livelihoods) are likely to occur towards the end of the dry season, where pastoralists, farmers, and villagers depend on water availability in surface water bodies and water points. Increasing population growth and climate change effects could foster future severe or more frequent water stress events. Remotely sensed imagery allows observations of surface changes over larger areas and at different points in time, using high resolution data or time series of medium resolution imagery, and can be used as a tool for studying water stress phenomena. Satellite data provide consistent, objective measurements at periodical time intervals and over large areas, therefore they can bring added value to the use of ground-based data. In the region analysed in this study, so far only a few studies have been carried out using high resolution satellite data, and the presented work contributes to a better understanding of surface processes as well as their connection to water scarcity. In this paper the suitability of satellite imagery to derive water stress indicators is evaluated. Semi-arid West-African regions in Burkina Faso and Niger with livelihoods mainly based on rain-fed agriculture and pastoralism are chosen as example areas for this study. On the one hand, possible indicators in the domain of Land Use change are studied, and on the other hand changing wetlands are explored. Strong seasonal differences between rainy and dry season have to be considered. Thus, the main dataset used in this study are two Landsat time series (1984-2011), one acquired at the end of the rainy season (Oct/Nov) where wetlands reach their maximum water level and annual vegetation is still present, and one at the end of the dry season (Mar/Apr) where the water level in wetlands (some completely dry out during the dry season) determines the water availability for pastoralism, agriculture and the local population since no rainfall occurs throughout the dry season until about May. A feasibility study is conducted where the suitability of detecting changes of land use and wetland parameters is tested with available high resolution satellite data from archives, based on the Landsat archive and extended by other high resolution image datasets. Challenges are discussed related to spatial and temporal scaling issues of different water stress phenomena and applicable remote sensing data. Furthermore, it is presented how satellite imagery may be used to build-up a framework of remote sensing based and land surface feature related water stress indicators.