IMPACTS OF GRID SPACING AND CUMULUS SCHEMES ON THE QUALITY OF RAINFALL FORECASTS OVER TANZANIA

The skill of a numerical mesoscale WRF model in forecasting rainfall over Tanzania, and the sensitivity of the results to convective parameterization schemes (CPs) and horizontal grid spacing especially over coastal areas of Indian Ocean during heavy rainfall event, are assessed. The cumulus schemes investigated are Kain-Fritsch (New Eta) (KF), Betts-Miller-Janjic (BMJ), Ensemble Grell ? Devenyi, Ensemle Grell 3d, Tiedkte (TS) and Kain-Fritsch (Old version). The skill of the model was examined for three selected cases from April 2014 rainy season using an Object Oriented Technique and Contingency Table. These cases are; a day of no - event (less than 10 mm/day), a day of moderate rainfall (less than 50 mm/day), and a day of heavy rainfall event (more than 50 mm/day).Verification procedures shows that, all six CPs have high skill to predict spatial rainfall distribution over Tanzania, and the quality of the results increases with the increase of model resolution. KF scheme performs better than other schemes.