Reduced complexity model for assessing patterns of rainfall erosivity in Africa

Multivariate geostatistical modeling can be used to generate spatial patterns of hydro-climate data over ungauged regions, but these models may be unsuitable when the hydro-climatological data available over large and remote areas are sparse and show different scales of resolution. In these cases, reduced complexity modeling can be better used in order to increase understanding of hydrological extremes at spatial scales and over time periods not covered by rainfall records. In this study we present and evaluate the African Rainfall Erosivity Subregional Empirical Downscaling (ARESED) model which has been developed based on hydro-climatological and geotopographic data from 46 stations across Africa with very varied climates and elevations. We spatially downscale the 85th percentile of monthly precipitation, based on several decades of data, from 50 km to 10 km grid squares in order to predict values of rainfall erosivity across Africa. This yields inputs comparable to values based on the standard Revised Universal Soil Loss Equation (RUSLE). The 46 African stations were chosen for model development because they are also sites for which there are RUSLE-based erosivity values. Once parameterized to capture mean rainfall erosivity over several decades, the ARESED model was run as a validation tool, comparing the output with actual erosivity data. On a continental scale and over decadal time scales, the ARESED model captures most of the important processes within the hydro-climatological system. Its reduced complexity structure also makes it suitable for application to regional management and environmental planning.