Assessing the skill of precipitation forecasts on seasonal time scales over East Africa from a Climate Forecast System model

It is becoming increasingly important to be able to verify the skill of precipitation forecasts, especially with the advent of high-resolution numerical weather prediction models. This study focused on assessing the skill of climate forecast system (CFS) model in predicting rainfall on seasonal time scales over East Africa region for the period January 1981 to December 2009. The rainfall seasons considered were March to May (MAM) and October to December (OND). The data used in the study included the observed seasonal rainfall totals from January 1981 to December 2009 and CFS model forecast data for the same period. The model had 15 Runs. The measure of skill employed was the categorical skill scores and included Heidke skill scores, bias, probability of detection and false alarm ratio. The results from the categorical skill scores confirmed relatively higher skills during OND season as compared to MAM. When compared with individual Runs, the mean of all the 15 Runs depicted relatively higher accuracy during OND season. Some individual Runs – 1, 7, 9 and 10 – also performed better during OND season. During MAM season, the mean of all the 15 Runs showed relatively lower accuracy in predicting rainfall. Some individual Runs – 5, 10, 12 and 14 – performed better than the mean of all the 15 Runs. The prediction of seasonal rainfall over East Africa region using CFS model depends on the season considered. During MAM, the prediction of seasonal rainfall is better as Runs are fewer, which showed relatively higher averaged skills; on the other hand, during OND the prediction of seasonal rainfall is better when using the mean of all the 15 Runs.