An unsupervised learning approach to identifying blocking events:the case of European summer

Abstract. Atmospheric blocking events are mid-latitude weather patterns, which obstruct the usual path of the polar jet streams. They are often associated with heat waves in summer and cold snaps in winter. Despite being central features of mid-latitude synoptic-scale weather, there is no well-defined historical dataset of blocking events. Various blocking indices (BIs) have thus been suggested for automatically identifying blocking events and are frequently used to study their occurrence historically as well as in climate model simulations. However, BIs can show significant regional and seasonal differences and therefore several indices are typically applied in parallel to test scientific robustness. Here, we introduce a new blocking index using self-organizing maps (SOMs), an unsupervised machine learning approach, and compare its detection skill to some of the most widely applied BIs. To enable this intercomparison, we first create a new ground truth time series classification of European blocking based on expert judgement.We then demonstrate that our method (SOM-BI) has several key advantages over previous BIs because it exploits all of the spatial information provided in the input data and avoids the need for arbitrary thresholds. Using ERA5 reanalysis data (1979–2019), we find that the SOM-BI identifies blocking events with a higher precision and recall than other BIs. We present case studies of the 2003 and 2019 European heat waves and highlight that well-defined groups of SOM nodes can be an effective tool to reliably and accurately diagnose such weather events. We further compare the skill at detecting historic blocking events by applying our new SOM-BI to several meteorological variables that are associated with the study of blocking, including geopotential height, sea level pressure and four variables related to potential vorticity. The 500 hPa geopotential height anomaly field is the variable that most effectively supports the identification of blocking events with our new approach. Finally, we evaluate the SOM-BI performance on around 100 years of climate model data from a pre-industrial simulation with the new UK Earth System Model (UK-ESM1). For the model data, all blocking detection methods have lower skill than for the ERA5 reanalysis, but SOM-BI performs noticeably better than the conventional indices. SOM-BI performs well using at least 20 years of training data, which suggests that observational records are sufficiently long to train our new method effectively. Overall, our results demonstrate the significant potential for unsupervised learning to complement the study of blocking events in both reanalysis and climate modelling contexts.