Exploring the Sudden Stratospheric Warming Events in High-top and Low-top Climate Model Large Ensembles

Sudden stratospheric warming (SSW) events, followed by a characteristic circulation regime in the lower troposphere, are crucial for the subseasonal weather prediction. What remains controversial is whether or not increasing the height of top layer in a climate model and the vertical resolution improve the representation of SSW events and the subsequent influences on the near-surface climate. In this study, we examine the SSW events simulated in a high-top climate model (the Whole Atmosphere Community Climate Model version6, WACCM6) and those simulated in a low-top model (Community Atmosphere Model version 6, CAM6) with 30 ensemble members. The two sets of experiments are forced by identical observational sea-surface temperature and sea-ice concentration, as well as the radiative forcings, during the 1979-2013 period. We find that WACCM6 produces about two times more SSWs than CAM6 (i.e., 759 v.s. 357 events), and, in terms of occurring frequency, SSWs in WACCM6 happen about 7 times per decade, closer to the SSW frequency in reanalysis datasets. Analyses on the thermodynamical and dynamical components of SSWs, including the sea-level pressure precursor, the preceding Eliassen-Palm fluxes into the stratosphere, the stratospheric temperature increases, and the downward propagation features, reveal that WACCM6 in general gives weaker signals than CAM6. This is likely attributed to the weaker mean stratospheric circulation in WACCM6. We also find that the WACCM6-CAM6 differences can be amplified during the years of El Niño and La Niña events. Finally, we perform the vortex moments diagnostics to gain further insights into the vortex structure and separate the splitting and displacement SSWs. The diagnostics shows that CAM6 generates more symmetric polar vortices than WACCM6, while the vortices from CAM6 deviate less often from the North Pole. However, the ratio of displacement and splitting SSWs in both model is about 70%, larger than about 60% in reanalysis datasets. Our study suggests that the high-top configuration leads to better performance of stratospheric circulation variability.