Forecasting storms over Lake Victoria using a high resolution model

Lake Victoria in East Africa is one of the world's largest freshwater lakes and is used on a daily basis by thousands of fishermen. Each year, severe storms on the lake cause multiple boating accidents which often result in fatalities. Recent initiatives have seen an effort to reduce accidents by issuing storm warnings when severe weather is expected. Here the Met Office global Unified Model is evaluated along with a 4 km limited-area model which has been set up to assist forecasters in the region to issue these warnings. Findings indicate the 4 km model is capable of producing more realistic strong wind speeds and rain rates than the global model. Case studies relating to fatal boating accidents on 1 March and 4 March 2012, showed improved warning signals of severe storms in the 4 km model compared to the global model. Objective comparisons between model and observations were conducted on 2 months of data. An objective method was used to determine 'storm'/'no storm' in the model forecasts. These were then compared against cloud top temperature from IR satellite and lightning data from the arrival time difference (ATD) radio ground network to determine whether each model was successful at forecasting storm/calm events. The 4 km model was able to capture more storm hits (thus had fewer storm misses), but also gained more false alarm events. Overall, the objective analysis showed that both models had some predictive skill and both were an improvement on a persistence forecast.