A hybrid statistical–dynamical approach for seasonal prediction of the boreal winter stratosphere

Abstract The winter variability of the Northern Hemisphere Stratospheric Polar Vortex (SPV) is characterized by extreme events known as Sudden Stratospheric Warmings (SSWs). SSWs feature a rapid temperature increase and a reversal of the zonal winds in the SPV, exerting a downward impact on the troposphere. Yet, despite their importance, current seasonal prediction systems (SPSs) show limited skill in forecasting SPV variability and SSW occurrence. This study quantifies to what extent existing statistical–dynamical approaches originally designed for the prediction of tropospheric modes of variability can improve seasonal SPV predictions. We apply North Atlantic Oscillation (NAO)-based teleconnection subsampling to Copernicus Climate Change Service (C3S) SPSs, selecting ensemble members according to statistical predictions of the winter NAO. This approach is motivated by the established link between NAO and stratospheric variability. Results show that teleconnection-based subsampling increases the prediction skill of winter SPV variability and the number of SSW days across all individual C3S models. These improvements come together with a better representation of extra-tropical lower-stratosphere wave activity during December and January, providing quantitative evidence that predictive information can be transferred from the NAO to the SPV through ensemble subsampling. However, the study further demonstrates that this transfer is effective only when using NAO statistical predictors that are also dynamically linked to SPV variability, whereas predictors lacking such a link fail to improve, or even degrade, SPV forecast skill. These findings clarify the applicability of teleconnection-based subsampling and suggest that successful transfer of predictability across components of the climate system requires predictors that are not only informative for the teleconnection used for subsampling, but also physically linked to the forecast target.