Use of MERIS data to detect the impact of flood inundation on land cover changes in the Lake Chad Basin

Abstract

Lake Chad Basin is a vital source of water to human, livestock and wildlife communities. Over the last forty years, the region has experienced a series of devastating droughts and the lake has shrunk to less than a tenth of its former size leading to environmental degradation and food insecurity. Consequently, stake holders have adopted the strategy of cultivating flood recessed lands as a livelihood resource, to cope with the variability of the lake. One of the defining characteristics of the lake is the seasonal inundation that occurs every year. The inundation of the lake is an important economic event because it provides irrigation water, moisture retentive soils for recessional farming and a breeding environment for fish. A major negative impact is the harvest uncertainties experienced on seasonally inundated land and lake floor whereby unpredictable flooding can cause loss of crops and the destruction of infrastructure thus necessitating the need for data collection and monitoring to provide early warning systems. Consistent and accurate data acquisition using conventional methods has proved to be a major difficulty because of the remoteness of the region. As such, there is no current inventory of cultivated or cultivable land and little emphasis is directed to the small scale recessional farming activities that occur in and around the lake. Remote sensing as a tool offers a lot of scope for environmental assessments and has been used successfully in studies, to monitor changes in the lake environment using various satellite platforms. In this study, the potential of MERIS images for broad regional studies of Lake Chad was evaluated by investigating multi temporal patterns of change which could be used to identify farmlands in the lake basin. The MERIS data were validated with ASTER while the suitability of ASTER data for validating MERIS data was ascertained by visual comparison with high resolution IKONOS images and fieldwork. Spectral Mixture Analysis was performed on MERIS data. Three endmember fractions, soil, vegetation and water were generated. Post classification threshold change detection analysis was performed using the endmembers as indicators of land cover changes to assess temporal changes over the lake basin. To establish a preliminary basis for an early warning system, the relationship between altimetric lake level data and areal extent of surface inundation were examined. This was done by estimation of the spatial distribution and areal extent of annual flooding of the lake, from MERIS water fraction abundance. The areal extents were then related to lake height level data derived from (i) TOPEX POSEIDON altimeter data and (ii) HYDROM generated lake level data. Results showed that temporal change patterns could be detected using MERIS thus making it suitable for inventorying and monitoring resources on the lake basin. However, due to the coarse spatial resolution of MERIS, farmlands smaller than the nominal pixel resolution of MERIS were not detected when analyzing the change patterns. The relationship between altimeter and areal extent of surface inundation showed that external factors such as the farming practices and the physical characteristics of the lake influence the surface inundation and therefore modify this relationship.