Evaluation of CMIP6 precipitation and temperature data based on multi-model ensemble

In order to improve the applicability of the CMIP6 climate model in the North China Plain, the quantile mapping method was used to correct the daily temperature and precipitation data of five typical GCMs (CanESM5, FGOALS-g3, IPSL-CM6A-LR, GFDL-CM4, and MRI-ESM2-0, denoted as M1–M5, respectively) and their multi-model ensembles (MME). Based on the historical observational data from 1979 to 2021, the performance of the models before and after correction in terms of temperature, precipitation distribution, spatial deviation and Taylor diagram was evaluated. The results show that after the deviation correction, the simulation results of the models in the main temperature range (0-15°C and 15-30°C) and low precipitation range (<1mm and 1-3mm) are significantly improved, especially for the M1, M2 and M5 models. Taylor diagram analysis shows that the correlation coefficient between temperature and precipitation of the corrected model is generally more than 0.95, the standardized standard deviation is close to 1, the root mean square error is significantly reduced, and the MME performs the best. In summary, deviation correction can effectively improve the simulation accuracy of temperature and precipitation of GCMs in the North China Plain, and the combination of multi-model ensemble and bias correction is an important means for regional climate prediction in the future.