Jump to Main Content
Using land cover changes and demographic data to improve hydrological modeling in the Sahel
- Paturel, Jean Emmanuel, Mahé, Gil, Diello, Pierre, Barbier, Bruno, Dezetter, Alain, Dieulin, Claudine, Karambiri, Harouna, Yacouba, Hamma, Maiga, Amadou
- Hydrological processes 2017 v.31 no.4 pp. 811-824
- Sahel, climatic factors, cropland, drought, evaporation, hydrologic models, land cover, rain, runoff, rural society, simulation models, soil water, soil water retention, surface water, time series analysis, vegetation, water holding capacity, water resources
- At the beginning of the drought in the Sahel in the 1970s and 1980s, rainfall decreased markedly, but runoff coefficients and in some cases, absolute runoff increased. This situation was due to the conversion of the land cover from natural vegetation with a low annual runoff coefficient, to cropland and bare soils, whose runoff coefficients are higher. Unless they are adapted, hydrological conceptual models, such as GR2M, are unable to reproduce this increase in runoff. Despite the varying environmental and climatic conditions of the West African Sahel, we show that it is possible to increase the performance of the GR2M model simulations by elaborating a time‐varying soil water holding capacity and to incorporate this value in the annual maximum amount of water to be stored in reservoir A of the model. We looked for interactions between climate, rural society, and the environment. These interactions drive land‐cover changes in the Sahel, which in turn drive the distribution of rainfall between infiltration, evaporation, and runoff and hence the water resources, which are vital in this region. We elaborated several time series of key indicators linked to these interactions. We then integrated these changes in the runoff conditions of the GR2M model through the maximum value of the reservoir capacity. We calculated annual values of water holding capacity using the annual values of four classes of land cover, natural vegetation, cultivated area, bare soil, and surface water. We then used the hydrological model with and without this time‐varying soil value of A and compared the performances of the model under the two scenarios. Whatever the calibration period used, the Nash–Sutcliffe index was always greater in the case of the time‐varying A time series.