Atmospheric fronts detection over Europe – methodological approach

Weather conditions, including the threat of extreme weather, in the temperate zone are mainly related to the atmospheric circulation, including the presence of atmospheric fronts. Despite many studies conducted on the distinction of atmospheric fronts in recent years, there is still no entirely satisfactory objective method of their discrimination with respect to traditional synoptic maps. Here we propose a method of distinguishing atmospheric fronts based on cluster analysis methodology treating border zones between air masses as fronts and testing the output against operational analyses The basic data used for the calculations were meteorological fields taken from the ERA5 database for the period 1951-2020 in hourly time steps across a domain given by coordinates 30°W to 50°E and 30°N to 80°N. The following elements were taken into account: temperature, specific and relative humidity, geopotential height on pressure levels from 925 to 700 hPa as well as temperature and wind speed near the ground. Using methods such as k-means and dbscan, division into 3, 4 and 5 clusters was conducted over Europe. This analysis was conducted for the whole period, as well as for specific months and seasons, and included considerations of spatial diversity and temporal changes in different regions of the domain. Validation of the results, including comparative studies with traditional synoptic maps from DWD and the Met Office, confirms the 3-classes distinction of air masses as the most reliable. The proposed method seems to be effective and self-dependent (contrary to, for example, supervised machine learning methods). This approach enables a method to be developed allowing calculations of severity of fronts and trends in the behaviour of fronts with time.