Predictability of convective precipitation for West Africa: Does the land surface influence ensemble variability as much as the atmosphere?

• A convection-permitting ensemble was set up for West Africa Perturbations were developed to cover known uncertainties of NWP models at the land surface
• The validation of the convective precipitation shows sufficient forecast skill
• The ensemble variability caused by atmospheric and land-surface perturbations was analyzed We conclude that both factors have an equally strong influence on the predictability of precipitation in the Sahel region

In recent studies, the importance of the influence of the land surface and especially of soil-moisture heterogeneities on convective systems and convection initiation in the Sahel was established. This investigation aims at comparing the land-surface part of the influence on convection with that of the atmosphere. For this reason, realistic land-surface perturbations were generated to set up an ensemble of convection-permitting simulations that contains atmospheric as well as land-surface perturbations. The simulation of precipitation by the ensemble proved to be sufficiently realistic. By comparing precipitation forecasts of individual members, it was found that the effectiveness of soil perturbations in generating ensemble variability is as large as the effectiveness of the atmosphere. This means that the representation of the land surface, reflected by parameters such as the soil-type distribution and absolute soil moisture as well as its heterogeneities, is as important for the predictability of convective precipitation in the Sahel region as atmospheric conditions. However, soil perturbations do not determine the day on which larger convective systems occur. This rather depends on larger-scale factors such as African easterly waves, the strength of the monsoon flow as well as the location and intensity of the heat low. In each case, it is a combination of different processes determining the occurrence of convection and convective precipitation.