A hybrid statistical-dynamical approach for seasonal predictions of the boreal winter stratosphere

The North Atlantic Oscillation (NAO) is a large-scale alternation of atmospheric mass between subtropical high surface pressure, centred on the Azores, and subpolar low surface pressure, centred on Iceland. Ensemble-based dynamical seasonal prediction systems (SPSs) are known to skilfully predict the winter NAO index for a season ahead and this ability improves increasing the ensemble size. However, recent studies by Dobrynin et al. (2018, 2022) prove the efficiency of a multimodel subsampling approach in increasing skill of predicted NAO index and temperature in the NH. This improvement on NAO predictability could also reflect in a better prediction skill of many variables and features which are strongly influenced by NAO phase and variability. The present work focuses on how this could affect the prediction of the boreal winter stratosphere with a reference to a recent study by Portal et al. (2022) in which the predictability of the winter stratospheric polar vortex (SPV) is investigated in the NH with specifical attention to the connection between the SPV and lower-stratosphere wave activity (LSWA). The first objective of this work is to perform a multi-model subsampling approach using five SPSs which contribute to Copernicus Climate Change Service (C3S) and cover the period from 1994 to 2017. The subsampling technique is performed analysing initial autumn conditions to identify ensemble members with well-established relationships between initial autumn conditions and the winter NAO. The main goal is to improve the prediction skill of NAO variability, phase and strength. The second objective of the present work is to underline the effect of the improved NAO skill on the prediction skill of winter SPV focusing on the analysis of zonal-mean zonal winds at 10 hPa with the same database used for the subsampling. Particular attention is dedicated to the probability forecast of Sudden Stratospheric Warming (SSW) events and to LSWA.