Performance Evaluation of CMIP5 and CMIP6 Models on Heatwaves in Korea and Associated Teleconnection Patterns

Abstract

This study assesses the performance of the Coupled Model Intercomparison Project (CMIP) models for simulating summer heatwaves in Korea during a historical simulation period (1979-2014) using four diagnostic indices that represent the teleconnection mechanism of summer heatwaves in Korea. Four skill metrics are used for the model evaluation, namely, relative error (RE), interannual variability skill-score (IVS), correlation coefficient (CC), and total ranking (TR) based on daily maximum temperature (TMAX) in Korea and the four diagnostic indices. The results show that the REs of CMIP5 models do not differ significantly from those of the CMIP6 models while the IVSs in the CMIP6 models are significantly improved compared with the CMIP5 models. Observations show that the heatwave circulation index (HWCI) contributes more to the interannual variability in TMAX in Korea than that of the Indian Monsoon Rainfall Index (IMRI), indicating that the teleconnection from the northwestern Pacific is more important than that from northwestern India. Interestingly, the CMIP6 models simulate this property better than the CMIP5 ensemble. The higher TR of CMIP6 models than CMIP5 supports that CMIP6 models are better overall in simulating heatwaves in Korea and the associated diagnostic indices.