Seasonal Precipitation Bias Correction of GCM Outputs for Thailand: Rayong Province Region

Abstract

Global Climate Models (GCMs) are sources of future simulated climate grid data. Their coarse resolution outputs are restricted to use for climate change impact analysis, though they require downscaling before application at fine scale. Using dynamical downscaling for coarse resolution GCM outputs requires large scale computation that may not be practicable for the daily climate data and spatial grid data of various GCM models; however the effectiveness of downscaling can be improved by statistical downscaling of GCM at fine scale. This study performed bias correction based on seasonal climate characteristics using statistical downscaling of daily precipitation data of six GCM models for Rayong Province in eastern coastal Thailand, selected from a total of 23 GCM models, and compared to GPCP past data covers Thailand region from 1981 to 2000 using score criteria of spatial correlation and RMSE method.  Since the climate of Thailand is directly affected by Southern Local Wind, Tropical Monsoon, and Southern West Monsoon, the climate of Rayong Province can be divided into 3 seasons: dry season (Dec to Feb), first rainy season (Mar to Jul) and second rainy season (Aug to Nov) respectively. Seasonal bias correction was performed on monthly precipitation data classified by daily precipitation data based on seasonal types, i.e. rainless day, normal rainy day, and extreme rainy day. The seasonal bias correction of monthly downscaled precipitation data from the selected GCM outputs was expressed as best fit for the realistic parameter of monthly observed rainfall data at the considered station. The results of this study reveal that seasonal bias correction can effectively compensate for unrealistic precipitation data and increase the reliability of precipitation data from GCM outputs to accurately reflect the precipitation characteristics of Thailand.