Evaluation of multidimensional simulations of summer air temperature in China from CMIP5 to CMIP6 by the BCC models: From trends to modes

To determine whether the capability of the CMIP6 version of Beijing Climate Center (BCC) models (BCC-ESM1 and BCC-CSM2-MR) in simulating China summer surface air temperature (SAT) has improved, we presented a multidimensional evaluation of the summer SAT in China including the trends, modes, and influencing factors. Critical comparisons are also made with the results of CMIP5 (BCC-CSM1.1 and BCC-CSM1.1m). In general, the CMIP6, especially BCC-CSM2-MR, has smaller deviations in the trends, the means, the mutations, the maximum centers, the variances, and the spatial patterns of the dominant modes from observatio relative to those of CMIP5. However, the BCC CMIP6 models still underestimate the SAT variation in the Qinghai‒Tibetan Plateau and the northeastern regions of China, and the performance is unsatisfactory with respect to the physical drivers of the dominant modes. Importantly, all the BCC models can capture the spatio-temporal characteristics of the first mode well and can, in general, characterize the spatial pattern of the second mode, but none of the models perform well in the principal component of the second mode (PC2) due to the low performance with respect to the interannual variation of PC2. Furthermore, the factors influencing the leading two modes are evaluated. The two CMIP6 can simulate better the Northern Hemisphere subtropical high northern boundary affecting the first mode. Another factor, the Asia polar vortex area, can only be simulated better by two low-resolution models (BCC-CSM1.1 and BCC-ESM1). For the second mode, all four models simulate the influence of Asian zonal circulation well, but poorly simulate that of the southern Indian Ocean dipole due to a large deviation in the Indian Ocean surface temperature.