Statistical downscaling of hourly precipitation in South Korea using the MS-PRISM method

In this study, we developed a grid climate dataset with a horizontal resolution of 500m × 500m for South Korea, utilizing observational station data from the Korea Meteorological Administration (KMA). The high-resolution 500m data were calculated using a newly developed Multi-Step (MS) PRISM (Parameter-elevation Regressions on Independent Slopes Model) method, which enhances the Modified Korean (MK) PRISM—a statistical downscaling technique for estimating high-resolution gridded data from observational data. First, to produce high-resolution hourly precipitation data, we performed quality control on the hourly precipitation observation data to select valid entries. Next, we created geographic information data, including Digital Elevation Model (DEM), aspect, and coastal proximity, all at a resolution of 500m. This geographic data was then applied to the MS-PRISM method to calculate hourly precipitation data at the same resolution. To confirm the reliability of the 500m resolution hourly precipitation produced, we conducted a verification of the final estimated data. We compared and analyzed the daily precipitation estimation errors as well as the hourly precipitation estimation errors at the same spatial resolution. Additionally, we evaluated the estimation results based on changes in spatiotemporal resolution by comparing the estimation errors associated with different spatial resolutions while maintaining the same temporal resolution.