On the spring stratospheric final warming in CMIP5 and CMIP6 models

This study evaluates the performance in simulating the stratospheric final warming events (SFWs) that lead to the final collapse of the stratospheric polar vortex in spring in both Southern and Northern Hemispheres (SH and NH, respectively) based on the historical simulations provided by the Coupled Model Intercomparison Project Phases 5 and 6 (CMIP5 and CMIP6, respectively). Overall, CMIP5 and CMIP6 models can reproduce the main characteristics of the occurrence of SFWs. However, the SFW onset date (SFWOD) is 7 and 9 days later than in observations in the SH and NH, respectively. Moreover, the intensity of SFWs in models is 50% to 70% of that in observations. Compared with CMIP5 models, CMIP6 models have an ameliorated capability to simulate NH SFWs. However, this improvement does not manifest as significantly earlier SFW onset, but as more intense stratospheric planetary wave activities before the SFW and as a larger interannual variability of the SFWOD. By contrast, in the SH, the capability of CMIP6 models is roughly unchanged, even deteriorated in the simulation of SFWOD and stratospheric planetary wave activities before the SFW onset. The performance of CMIP6 high-top models is better than that of low-top models. Specifically, in the NH, high-top models are considerably improved in terms of intensity of circumpolar zonal wind around the SFWOD and stratospheric planetary wave activities before the SFW onset. In the SH, high-top models show fairly earlier SFWOD by 11 days, which is closer to observations.