The rainfall simulation in General Circulation Models (GCMs) is crucial in hydrological applications. In this study, the accuracy of rainfall from 15 GCMs in CMIP6 is evaluated over the Yangtze Delta. The results demonstrate that these GCMs can reproduce the seasonal cycle. However, summer rainfall is underestimated by 70% GCMs, and the increasing trend of rainfall amount is not captured by most of the GCMs. Overall, NorESM2-LM, EC-Earth3, CanESM5, IPSL-CM6A-LR are recommended models applied to the Yangtze Delta. The evaluation results provide fundamental information to future rainfall projection and change in flood risk due to climate change.


Presently, climate change has become a great challenge for human society (IPCC, 2022). Under the impact of climate change, the rainfall characteristics are varying temporally and spatially. Precipitation-related disasters like floods have enormous impacts on agriculture, water resource management, and transportation. The Yangtze Delta is one of the most developed and densely populated metropolitan clusters in China, which is also vulnerable to flood disasters. Plum rain and tropical cyclones, e.g. typhoons, inducing the flood events, result in huge economic loss in the region (Zhu et al., 2021). Thus, studying precipitation in the flood season and providing reliable rainfall patterns under future climate change are of great importance in this region.

General circulation models (GCMs) generate atmospheric variables in history and under future scenarios, are important tools in climate change studies. The World Climate Research Programme (WCRP) established the Coupled Model Intercomparison Project (CMIP) to compare GCMs worldwide and standardize climate simulation outputs. The CMIPs form the data foundation of the Inter-governmental Panel on Climate Change (IPCC) Assessment Reports.

The evaluation of CMIP models is a significant work. The precipitation simulation in China was evaluated, starting with the release of CMIP3 models (Miao et al., 2012; Chen and Sun, 2015), and illustrated the simulation of CMIP3 models generally underestimated the Plum Rain phenomenon for the Yangtze basin area (Tu et al., 2015). Later, Chen and Sun (2015) demonstrated the advantage of CMIP5 over CMIP3 in China. And Kusunoki and Arakawa (2015) considered due to the improved representation of the western Pacific subtropical high (WPSH) in the CMIP5, models in CMIP5 showed higher skill in simulating summer rainfall than CMIP3 over East Asia, however, CMIP5 models still underestimate the precipitation in the warmer season (May-September). Huang et al. (2013) indicated 14 CMIP5 models spread most in early summer over the Yangtze River basin, which is attributed to the uncertainties in simulating large-scale circulations, e.g. WPSH and East Asian Summer Monsoon. Model resolution (Wang et al., 2017) and physical parameterizations (Qian et al., 2015) also influence simulation results. Relatively, multi-model ensembles (MME) have better ability than individual ones (Chu et al., 2015). A further selection of higher-skill models can yield a more reliable projection than MME (Kwon et al., 2019).

You can access this article if you purchase or spend a download.