Assessment of model performance of precipitation extremes over the mid-high latitude areas of Northern Hemisphere: from CMIP5 to CMIP6

This study explores the model performance of the Coupled Model Intercomparison Project Phase 6 (CMIP6) in simulating precipitation extremes over the mid–high latitudes of Asia, as compared with predecessor models in the previous phase, CMIP5. Results show that the multimodel ensemble median generally outperforms the individual models in simulating the climate means of precipitation extremes. The CMIP6 models possess a relatively higher capability in this respect than the CMIP5 models. However, discrepancies also exist between models and observation, insofar as most of the simulated indices are positively biased to varying degrees. With respect to the temporal performance of indices, the majority are overestimated at most time points, along with large uncertainty. Therefore, the capacity to simulate the interannual variability needs to be further improved. Furthermore, pairwise and multimodel ensemble comparisons were performed for 12 models to evaluate the performance of individual models, revealing that most of the new-version models are better than their predecessors, albeit with some variance in the metrics amongst models and indices.