Statistical Assessment of High-Resolution Climate Model Rainfall Data in the Ciliwung Watershed, Indonesia

Abstract

The impact of climate change on hydrometeorological hazards pointed out the necessity for information on rainfall data. Using Climate Hazard Group InfraRed Precipitation with Station (CHIRPS) data could solve the problem of the scarcity of observed rainfall data at a finer spatial resolution. This paper examines the performance of high-resolution rainfall climate model data called CORDEX SEA and NEXGDPP in the Ciliwung watershed, Indonesia. We used CHIRPS data as observed data, which was separately divided for calibration (1981-2005) and validation (2006-2020) of the climate models. Totally 14 climate models were used, comprised of 4 CORDEX and 10 NEXGDPP. The models accuracy was assessed based on three statistical indicators: bias, mean absolute percentage error (MAPE), and mean square error (MSE). We determined the best model based on Taylor Diagram. The results showed that the bias value in the dry season was smaller than in the wet and transitional seasons. All models performed well as shown by the low bias values except for the ACCESS1-0 RCP8.5 model. The findings revealed that MRI-CGCM was the best model for calibration, whereas EC-Earth was the best model in the validation period for both RCP4.5 and RCP8.5 scenarios. Further, the choice of climate model may influence water resource management over watershed scale.