Study of Biascorrection Methods for Seasonal Rainfall Forecast over Vietnam

Abstract: This study presents some results about biascorrectionseasonal rainfall forecast from the regional spectral model (RSM), following two methods are quantile-quantile with an approximate gamma function (QM-G), and Bayesian joint probability (BJP). RSM ran forecast for the period 1982-2014, with data input from global model CFS,and lead time up to five months. The results show that the BJP made the correlationbetween rainfall forecastand observation increased significantly, the coefficient correlation after corrected is about 0.77 in all three lead times. The bias and error after did correctly by BJP were reduced away clearly, the differencesare almost notin all of three lead times, the error in months from April to October is the smallest and about 20-50%, therein the Northwest climate gives the smallest error. The correction with QM-G did not improve the correlation and bias, which is also made the model losing systematic ofthe error. Keywords: Rain correction, seasonal forecast, Vietnam region, RSM