Modes of European Circulation Variability from a Central European Perspective

Understanding the large-scale variability of atmospheric circulation over Europe has always been a key challenge for interpreting weather extremes and climate trends. The classical Euro-Atlantic weather regimes, usually defined by clusters of geopotential height anomalies, put the emphasis primarily on Western Europe and the connection to the North Atlantic Oscillation. However, they often fail to explain much of the variability of basic meteorological variables in Central Europe and their spatial patterns across the entire European continent. Therefore, we have implemented a new, machine learning-based approach to define the main modes of European circulation variability. Mean daily fields of 500 hPa geopotential height anomalies were calculated from the ERA5 reanalysis for domain centered unusually towards Central Europe (40°W–60°E, 30°–80°N) for the 1981–2023 period. Then 300 individual Self-Organizing Map (SOM) neural networks were trained on this daily dataset, using a topology of 25×25 nodes with 248 training cycles (epochs) and subsequently combined into a single SOM model, by averaging weights in individual nodes. The entire model was clustered using Wards hierarchical clustering method and tested by silhouette analysis, which lead to identification of four main modes of European circulation variability: Atlantic blocking (ABL), Continental blocking (CBL), Southern zonal track (SZT), Northern zonal track (NZT). Then, one mode was assigned to each day, using calculation of the Best Matching Unit (BMU), representing the SOM node with the highest similarity (smallest Euclidean distance) to the given day, including days outside the SOMs’ training period, therefore the time series was extended to the 1940-2023 period. Employing fields from ERA5 and E-OBS, it was discovered that each mode represents a very distinctive spatial patterns of weather conditions over the entire Europe. In particular, SZT/NZT represent very dry/wet patterns over Central and Western Europe (documented by change in soil moisture), even though both modes occur more often during positive NAO phase. We believe these new modes of European circulation variability have the potential to be a useful tool in dynamic climatology and in season-to-season weather forecast.