Impact of cumulus parameterization on the extreme rainfall simulation according to the summertime weather patterns over South Korea

Abstract

This study identifies the characteristics of summertime extreme rainfall events (ERE) according to the weather patterns in South Korea and compares the model simulation skills depending on the use of cumulus parameterization. The Weather Research and Forecasting (WRF) model simulations with and without the Kain–Fritsch cumulus scheme (CU_ON and CU_OFF experiment, respectively) are conducted over a 3 km high-resolution domain. The ERE-occurring days are clustered into four representative weather patterns in terms of the 850 hPa geopotential height anomaly. The observed Clusters 1 and 2 rainfall is characterized by continuous low-intensity rainfall. In contrast, relatively high-intensity, short-duration rainfall occurs mainly in Clusters 3 and 4. For CU_ON WRF experiments, the spatial distribution of the daily rainfall anomaly composite in Clusters 1 and 2 is well depicted, but the overall rainfall intensity is underestimated. CU_ON better reproduces the Clusters 1 and 2 type rainfall characterized by long-duration rainfall than CU_OFF. The observed rainfall events exceeding 20 mm h−1 intensity with a short-duration are reproduced better in CU_OFF than in CU_ON, showing reasonable performance for sub-daily timescale rainfall in Clusters 3 and 4. Hence, the CU_ON well depicts the continuous low-intensity ERE type, while CU_OFF captures the ERE type where high-intensity rainfall with a short duration occurs relatively frequently.