Statistical downscaling and bias correction for daily precipitation in the South Korea using the PRIDE model

In this study, we produced grid climate data set of 1km×1km horizontal resolution in South Korea using 5 types of RCM (HadGEM3-RA, CCLM, RegCM4, WRF, GRIMs) results based on Socioeconomic Pathways (SSP) four scenarios (tier1: SSP1-2.6, SSP2-4.5, SSP3-7.0, SSP5-8.5) of the IPCC 6th report. The high-resolution future scenario data of South Korea were calculated using the PRIDE (PRism based Dynamic downscaling Error correction) model based on MK (Modified Korean)-PRISM (Parameter-elevation Regressions on Independent Slopes Model), a statistical downscaling method that can estimate grid data of horizontal high-resolution using observational station data in South Korea. And then, the QDM (Quantile Delta Mapping) method was used to correct bias due to climate change trend in high-resolution data of future period. The PRIDE model results were evaluated as realistically reflecting seasonal changes and topographical characteristics in South Korea. Furthermore, we assessed uncertainty for future climate data using the results of 5-ensemble models. As a result, in temperature, uncertainties due to internal variability and model were larger than due to scenario in the near future, and the influence of the scenario became greater as it progressed towards the end of 21st century. On the other hand, in the case of precipitation, the uncertainty according to the model over the entire future period was the largest, exceeding 60%.