Attribution of subseasonal cold‐spell forecasts to remote drivers using nudging experiments

Abstract

A set of nudging experiments with the European Centre for Medium‐Range Weather Forecasts (ECMWF) model is used to attribute subseasonal cold‐spell forecasts over northern Europe to remote drivers. Four cases were selected where forecasts confidently predicted cold anomalies in Week 3: two associated with weakened stratospheric polar vortex (January 2013 and January 2010) and two occurring during active Madden–Julian Oscillation (MJO, February 2013 and February 2005). For each case, ensemble experiments were performed with either the zonal‐mean stratosphere or the tropical belt relaxed towards ERA5 reanalysis or towards climatology. Results show that the January 2013 cold spell was primarily stratosphere‐driven: nudging the stratosphere towards climatology removed the Week 3 cold anomaly and the negative North Atlantic Oscillation (NAO) signal. In contrast, the stronger January 2010 cold anomaly was reduced but persisted after removing the stratospheric signal, indicating an additional contribution from tropospheric processes. In the tropical cases, tropical nudging affected NAO but did not affect the cold anomalies, suggesting additional contributing mechanisms. While the February 2005 cold spell coincided with a negative NAO, a negative NAO was not predicted for February 2013, highlighting the complexity of factors shaping European cold spells. To demonstrate feasibility of operational attribution, additional experiments were conducted with either the stratosphere nudged towards the ensemble member with the weakest predicted vortex or the tropics nudged towards the ensemble member with the strongest MJO. The results of these experiments are similar to those obtained with ERA5 (fifth generation of ECMWF atmospheric reanalyses of the global climate) nudging. Such experiments can serve as a counterpart to the climatological nudging when observations are unavailable, providing an opportunity to attribute predicted signals in operational forecasting. Overall, the results demonstrate that cold spells can arise through different dynamical pathways, potentially entailing different uncertainty. Targeted nudging experiments provide a practical attribution framework to assess the origin and confidence of subseasonal signals, linking physical attribution with operational prediction.