Vol. 127, No. 4 * Pages 421–504 * October - December 2023

Certain types of medical meteorological phenomenontransitions can have a significant deteriorating effect on road safety conditions. Hence, a system that is capable of warning road users of the possibility of such conversions can prove to be utterly useful. Vehicles on different levels of automation (i.e., ones equipped with driver assistance systems – DAS) can use this information to adjust their parameters and become more cautious or warn the drivers to be more careful while driving. In this paper, we prove that identifying the critical type of weather front transition (i.e., no front to unstable cold front) is possible based on locally observable meteorological information. We present our method for classifying weather front transitions to non-critical versus critical types. Our developed machine learning model was trained on a dataset covering 10 years of meteorological data in Hungary, and it shows promising results with a recall value of 86%, and an F1-score of 60%.
As the developed method will form the basis of a patent, we are omitting key components and parameters of our solution from this paper.