Persistent model biases in the CMIP6 representation of stratospheric polar vortex variability

Richard J Hall*, Daniel M. Mitchell, William J.M. Seviour, Corwin J. Wright

*Corresponding author for this work

Research output: Contribution to journalArticle (Academic Journal)peer-review


Sudden Stratospheric Warmings (SSWs) can have major impact on surface wintertime weather, especially at mid-high latitudes. We do not yet have a complete understanding of why some of these events influence our weather more than others, but one factor may be the dynamical nature of the SSW; whether it involves a split or a displacement of the polar vortex, and one way to explore this is through comprehensive climate models. Here we analyse the stratospheric dynamics of SSWs within models from the sixth Coupled Model Intercomparison Project (CMIP6). All CMIP6 models are able to simulate SSWs to some degree, and while the frequency of events is, on average more realistic than in older models, we find a persistent bias in the relative underrepresentation of split vortex events. When comparing with CMIP5 models, large biases persist despite significant model improvements in resolution and in representing atmospheric processes. We show that the simulated displacement frequency is strongly related to climatological lower stratospheric eddy heat flux. The split frequency, on the other hand, is not related to lower stratospheric eddy heat flux, but is strongly related to both the vortex geometry (aspect ratio) and lower stratospheric zonal winds. This suggests that those models with a large positive bias in zonal winds may inhibit the propagation of zonal wavenumber 2 planetary waves from the troposphere, which are associated with split events. Our results suggest how future model development may address these longstanding biases.
Original languageEnglish
JournalJournal of Geophysical Research: Atmospheres
Publication statusSubmitted - 10 Feb 2021


  • sudden stratospheric warming
  • CMIP6
  • splits
  • displacements
  • stratosheric polar vortex
  • model bias

Cite this