Evaluation of soil moisture ensemble runs to estimate precipitation variability in convection-permitting model simulations for West Africa

Abstract

In this investigation, the impact of soil moisture on the precipitation amount and distribution in model simulations for West Africa was assessed. This included the examination of the dependence of precipitation on soil moisture, boundary layer parameters or convective conditions. For this purpose, different soil moisture fields from the rainy season of the year 2006 in West Africa were collected. These included three analysis fields, satellite measurements, and model output from the African Monsoon Multidisciplinary Analysis (AMMA) Land Surface Model Intercomparison Project. The different soil moisture fields were analyzed and compared first; large differences in the mean values, the north–south gradients, and the heterogeneity were revealed. The distributions were taken to initialize five convection-permitting model runs performed with the limited-area model COSMO (Consortium for Small-scale Modeling), forced by reanalyses from the European Centre for Medium-Range Weather Forecasts. A case study of a period with intense convective activity in the Sudanian and Sahelian zone was selected for the simulations. The simulated precipitation was compared with data from two satellite-based precipitation analyses and with forecasts from two global models of lower resolution. The model spread encompassed the precipitation analyses in the southern subdomain, while the latter lay far above the model results in the northern subdomain. The different soil moisture fields also influenced the turbulent fluxes at the surface, the boundary layer state and height as well as convective instability. Standard deviation of the simulations was highest for convective available potential energy, while the impact on convective inhibition was less pronounced. Only a weak negative relation of precipitation to soil moisture or conditional instability could be discerned. In this case, dry soil favored a higher boundary layer with better developed convective cells contributing to larger precipitation sums. The spread between the five precipitation forecasts suggests that a high-resolution limited-area ensemble, built by applying atmospheric variations as well as variations of the initial land surface conditions, will be a proper method to improve the prediction of mesoscale convective systems and convective precipitation for West Africa.