Subseasonal predictability of European winters using a weather regimes approach

The ability of predicting winter anomalies of surface variables like surface temperature (t2m) is important for economic sectors like energy production and trade. Sub-seasonal predictions can provide a useful basis for the early detection of these anomalies. However, the skill varies considerably depending on the season and the location, and identifying periods of increased predictability is nontrivial. This study aims to work in this direction targeting atmospheric patterns leading to skillful winter predictions at S2S lead times using a weather regime (WR) approach with the overall goal to build confidence in the forecast. With a focus on Europe, we explore extended range t2m predictability (up to week 5) over 20 extended winters (1999-2018) in the ECMWF reforecasts with ERA5 reanalysis as the reference. Using the Euro Atlantic weather regimes, computed with 500 hPa geopotential height daily data, we identify the predominant WR weekly in both reanalysis and reforecast data. We propose a framework that allows for quantifying the degree of similarity between the reanalysis and the forecast WRs and assess whether higher similarity brings about improved skill. This is done by considering the difference in skill between all of the start dates with a predominant WR at week 1 and a subset made of occurrences with a degree of similarity that is higher than the climatology (i.e., the forecast system predicted the reanalysis WR better than the climatology did).  Results indicate that the framework proposed helps identifying more skillful forecasts making use of the WR similarity, particularly during NAO+ and NAO- conditions. This study constitutes an important step in the direction of exploiting flow dependent predictability to improve confidence in the forecast and can be considered as a preparatory step to the use of the more comprehensive real-time ensemble forecasts.