Evaluating Weather Research and Forecasting Model in Simulating March-May Rainfall in Tanzania: Implications of Selecting Parameterization Schemes

Abstract

The current study evaluates the performance of the Weather Research and Forecasting (WRF) model in simulating seasonal rains of March-May (MAM) in Tanzania, based on the implication of selecting parameterization scheme combinations. The model was configured into two domains with horizontal resolutions of 36 km and 12 km and the initial and lateral boundary conditions were provided by the Climate Forecast System Version 2 at 00 UTC. However, only the inner domain of 12 km was used for analysis, which has a fine horizontal resolution that accounts for small scale features such as terrains. Twelve simulations have been performed using four Cumulus and three Microphysics schemes to determine the best scheme combination for MAM seasonal rainfall. The model outputs were compared with the Climate Hazards Group InfraRed Precipitation with Station and gauged rainfall data. The performance of the model in simulating MAM seasonal rainfall was analyzed using standard statistical measures and ranking transformation analysis. The results indicated that Grell-Freitas (GFE) and Betts-Miller-Janjic (BMJ) cumulus schemes when combined with the WRF Double Moment 6 (WDM6) class microphysics scheme performed reasonably better in simulating MAM seasonal rainfall in Tanzania. Moreover, the combination of New Tiedtke (TDK) cumulus and Kessler (KSS) microphysics was found to be the less accurate combination among all. Therefore, in improving operational seasonal prediction in Tanzania and increase the confidence of the forecast, the study recommends that GFE-WDM6 and BMJ-WDM6 scheme combinations should be used for operational forecasting of MAM seasonal rainfall.