Evaluation of model performance in simulating extreme precipitation indices over eastern China: A comparison of CORDEX and NEX-GDDP models

Abstract

Coarse-resolution global climate models (GCMs) have limited capability to simulate mean and extreme precipitation over eastern China, thus requiring the use of downscaling techniques to improve model performance. This study evaluates two types of downscaled climate models: regional climate models (RCMs) from the Coordinated Regional Downscaling Experiment Phase II for East Asia (CORDEX), which use a dynamical downscaling approach, and the NASA Earth Exchange Global Daily Downscaled Projections (NEX-GDDP, NEX), which apply a statistical downscaling method. CORDEX and NEX models are driven by CMIP5 and CMIP6 GCM outputs, respectively. For all precipitation indices (PRCPTOT, R95p, R99p, Rx5day, CWD, and SDII), the CORDEX ensemble mean marginally improves their representation south of 30°N compared to CMIP5. Among the four model groups, NEX models achieve the highest skill in reproducing PRCPTOT over eastern China. However, CORDEX models outperform NEX models in simulating extreme precipitation events, with NEX models systematically underestimating extremes, indicating that dynamical downscaling is generally more effective than statistical downscaling for capturing extreme precipitation. Both CORDEX and NEX models successfully capture the timing and peak of the annual precipitation cycle over eastern China, despite notable inter-model discrepancies. These results provide a comprehensive reference for selecting appropriate models to project mean and extreme precipitation over eastern China.