Contrasting Skills of S2S Models in Predicting Downward Propagation of Weak and Strong Polar Vortex Events and Impacts on Surface Temperature

Abstract Variations in the stratospheric polar vortex (SPV) intensity can influence large‐scale tropospheric circulation and surface weather through downward propagation of Northern Annular Mode (NAM) signals, offering a source of subseasonal‐to‐seasonal (S2S) predictability. This study evaluates the skill of S2S models in predicting the downward propagating (DP) characteristics of weak and strong stratospheric polar vortex (SPV) events, respectively, and their associated 2m‐temperature ( T 2m ) anomalies. Using ERA5 reanalysis and real‐time forecasts from S2S models during winter 2014–2024, SPV events are identified based on the 10 hPa NAM and classified into DP and non‐downward‐propagating (NDP) types according to their vertical propagation to 500 hPa. Results reveal that models capture well DP events but overestimate downward propagation for NDP events, beyond ∼20‐day leads. This bias is slightly stronger for weak events. Forecast skill of T 2m anomalies over Northern Hemisphere continents is enhanced for DP‐type weak and strong SPV events, with a larger advantage for weak events. However, skill is significantly weakened for NDP‐type weak events over Asia, where T 2m anomaly patterns that follow DP‐ and NDP‐type weak events are opposite. For NDP‐type strong events, skill degradation is over Europe and North America, with a generally smaller DP versus NDP contrast than weak events. Beside the forecasting failure of NDP‐type SPV signals, tropospheric teleconnection biases also play a role in determining the location of severe skill degradation. These findings call for more attention to the NDP weak SPV events, both in terms of better modeling and in the application of stratospheric signals to subseasonal prediction.