Verification of Rainfall Forecasts for the South Central Climate Region of Vietnam

This study aims to investigate the performance of the Weather Research and Forecasting (WRF) model for rainfall forecasts in the South Central climate region of Vietnam. The investigation was carried out by analyzing the accuracy of the model outputs at station sites and the spatial structure of rain events for different rainfall thresholds over the whole year and in the flood and dry seasons. The traditional (standard) method was utilized to analyze the accuracy of the WRF model in predicting precipitation point-by-point, whereas the Contiguous Rain Area (CRA) method was applied to analyze the spatial structure of rain events. The results showed that rainfall forecasts by the WRF model for the South Central region had certain limitations because the model scores and measured error criteria were not close to their perfect values. The proportion of hit forecasts decreased from 30 % with the traditional verification method to 10% with the spatial structure verification method. The pattern error was a main contributor to the total error at 53%, followed by the intensity error at 34%. The location error accounted for the lowest percentage contribution to the total error, at only 13%. The performance of this model could lead to substantial errors in weather and streamflow predictions for the south-central region and may lead to a lack of forecast effectiveness for mitigating the damage from natural disasters. Thus, improvements in the performance of the Numerical Weather Prediction (NWP) model for the studied area are necessary.