Runoff generation processes: Results and analysis of field data collected at the east central supersite of the HAPEX-Sahel experiment

Within the scope of the HAPEX-Sahel experiment, the hydrological functioning of two small nested catchments was studied at two different scales: the plot scale (of the order of 100 m(2)) and the catchment scale (0.2 km(2)). At local scale, four runoff plots were set-up on the typical soil surface conditions observed on the catchments (plateau bare soil, two plots on fallow grassland) and an additional one was installed on a millet field. Soil moisture investigations at the plot scale have shown that infiltration was limited between 0.6 to 2 m deep on three sites, but was deeper than 3.4 m on the most pervious site (millet). The maximum water storage on all the sites was found to be reached at the maximum activity of the rainy season (late August), and not at the end of the season. During the dry months, the soil was fully dried off by evapotranspiration, resulting in the absence of inter-annual soil water storage. No influence of vegetation cover on runoff was observed on the fallow sites, but runoff generation was found to be very sensitive to tillage on the millet field. The parameter P-u, calculated from a rainfall hyetograph and defined as the rainfall depth that can actually produce runoff, is shown to be relevant to compute runoff on untilled soils, as it explains more than 87% of the variance in runoff depth. On tilled soils, it is necessary to take into account additionally the temporal evolution of the soil surface, especially the days after weeding operations. Simple linear relationships were derived to compute runoff depth from P-u on the plots for the most typical soil moisture conditions observed, and modified SCS equations have been derived for the catchments. Using the linear equations derived at the plot scale in a simple, empirical, semi-distributed model lead to formulate the assumption that the partial source area concept applied on the catchments. Analysis of discharge data at the catchment scale highlights that seepage through the bottom of a gully between two gauging stations leads to the abstraction of non negligible volumes of water. Moreover, the water totally infiltrates in a spreading zone downstream from the outlet of the largest catchment showing that discontinuities occur in the surface water transmission within a catena. Such discontinuities constitute a major problem for the concern of aggregation of hydrologic precesses. (C) 1997 Elsevier Science B.V.