| dc.description.abstract | Accra, the capital of Ghana, is located in a tropical climate where flash floods are a common occurrence, especially during the rain season. A project by multiple companies has been set up which aims to prevent casualties of these flash flood by building a warning system. Warnings will be based on a hydraulic model, which needs precipitation forecast data as input. The Weather Research \& Forecast (WRF) model will be used as an operational numerical weather model to try and accurately forecast precipitation at a high resolution. The focus of this research is to find a good model setup for Ghana. This includes designing a domain, deciding what meteorological and geographic input data to use and by which schemes physics will be described. Schemes responsible for precipitation are assumed to be most important. Therefore, the performance of various combinations of cumulus parameterizations and microphysics schemes have been tested in detail by doing two case studies. Verification of the modeled precipitation is limited to satellite observations due to a lack of radar data or surface observations. Results vary a lot for the tested combinations, indicating the model is sensitive to the choice of physics schemes. Overall, the forecast skill is low. However, Thompson microphysics combined with a multi-scale Kain-Fritsch cumulus parameterization scheme consistently performed the best in both cases studied. Most nocturnal convective systems are missed by the model and the too early triggering of convection raises concerns as well. Identifying forecast errors by verifying atmospheric properties such as temperature, wind and moisture could lead the way to improving the model setup. | |
