Evaluation of multi-RCM ensembles for simulating spatiotemporal variability of Asian summer monsoon precipitation in the CORDEX-East Asia Phase 2 domain

Abstract

Five regional climate models (RCMs)-CCLM, RegCM, HadGEM3-RA, SNURCM and WRF-participating in the Coordinated Regional Climate Downscaling Experiment-East Asia (CORDEX-EA) Phase 2 project are evaluated for their ability to simulate spatiotemporal variability in Asian summer precipitation. For this purpose, two dynamical downscaling sets, experiments forced by ERA-Interim reanalysis data (reproduction experiment) and historical data from three Coupled Model Intercomparison Project 5 (CMIP5) general circulation models (GCMs) (historical experiment) are analysed. The horizontal resolution of the downscaled results is 25 km, and the analysis period is from 1981 to 2005 (25 years). The RCMs show reasonable performance in simulating the spatial and temporal characteristics of summer precipitation in CORDEX-EA Phase 2 domain. To assess spatiotemporal patterns in Asian precipitation, cyclostationary empirical orthogonal function (CSEOF) analysis is used. In the first mode representing seasonal variations, the model ensemble results of both the reproduction experiment (Rep_ENS) and the historical experiment (His_ENS) simulate the peak times, location of precipitation and progression of the monsoons in Asia reasonably. Their features are greatly influenced by the moisture flux, indicating that the relation between precipitation and 850 hPa moisture flux is depicted well by RCM ensembles. In the second mode, which represents intraseasonal variations, the Rep_ENS depicts the western North Pacific summer monsoon break (WNPSM break) phenomenon similar to the observation. Although the His_ENS simulates the WNPSM break later than the observation, it does present the intraseasonal variation in the East Asian summer monsoon region. Based on these results, the five RCM ensemble results show the ability to capture spatiotemporal characteristics of summer precipitation in CORDEX-EA Phase 2 domain, as well as added value from dynamical downscaling.