Improving global rainfall forecasting with a weather type approach in Japan Jean-Francois
By Vuillaume and Srikantha Herath
An automated weather classification system was used to analyze daily weather conditions in Japan. This system used data from two sources: mean sea-level pressure data from the ECMWF Re-Analysis dataset and daily forecast data from the TIGGE dataset.
The classification identified 11 weather types, including anticyclones, cyclones, hybrids, and various wind directions. The study showed that cyclones, hybrids, westerly, and northwest winds were the main contributors to total rainfall.
By applying a gamma-based bias correction, the accuracy of global rainfall forecasts improved by 10%. When applying specific weather type bias corrections, the overall error in rainfall forecasts was reduced by up to 20%, with a 5-10% reduction in root mean square error. Both global and weather type bias corrections enhanced extreme rainfall predictions, particularly for rainfall intensities exceeding 100 mm/d.
