Forecasting severe rainfall in the equatorial Southeast Asia

Abstract

Accurate prediction of monsoon heavy rainfall events in the equatorial region has always been a challenge to weather forecasters. In this paper, forecast of a severe precipitation event that occurred over the eastern central coast of Peninsular Malaysia was attempted using the state-of-the-art Florida State University (FSU) Global and Regional Spectral Models. The sensitivity of parameterized convection in these models on precipitation forecast skill is studied using two different parameterization schemes for cumulus convection (the Relaxed Arakawa-Schubert scheme and the modified Kuo scheme). Low precipitation threshold of rainfall less than 2 mm day^–1 was successfully predicted by both versions of the FSU model. However, the convection schemes lacked skill in predicting the correct placement of the area and amount for the high precipitation threshold greater than 40 mm day–1. Further evaluation of the predictive skills showed that the Relaxed Arakawa-Schubert scheme was a consistently better predictor of rainfall due to its low bias and lower root mean square errors (RMSEs) compared to the modified Kuo parameterization scheme.