Verification of surface runoff volume prediction with statistical downscaling method by using climate forecast system version 2 (CFSv2) output

Surface runoff is one of the factors causing flood. The surface runoff volume prediction is one of important activities to be conducted for flood mitigation. CFSv2 (Climate Forecast System version 2) is a seasonal prediction system that can be used to predict surface runoff volume. However, CFSv2 outputs are relatively coarse which cannot represent well the climate information at local scale. Thus, downscaling of these outputs is needed before continuing to use them to predict surface runoff. One of the methods is statistical bias correction applied to improve the prediction accuracy and skills, and to reduce the bias. The study areas of this research are located at Balangan and Tabalong sub-watershed in South Kalimantan, Indonesia. The study focuses on wet season (December-February). The predicted result of surface runoff volume from CFSv2 output is verified using Brier Score (BS) and Brier Skill Score (BSS). In this study, two schemes are performed to predict surface runoff volume. The results show slightly different values of BS between corrected and raw CFSv2 in both schemes. Scheme 1 uses water surface runoff variable. The corrected CFSv2 output of scheme 1 has BS values ranging from 0.5 to 0.7, and it has positive and negative BSS values where the negative BSS values indicate a low prediction accuracy in predicting high surface runoff. While scheme 2 uses prate (rainfall) variable. The corrected CFSv2 output of scheme 2 has a fairly good accuracy based on BS values ranging from 0.32 to 0.46, and it has positive BSS values. Thus, scheme 2 can be considered to predict the surface runoff volume for flood mitigation.