Machine Learning-Based Front Detection in Central Europe

Extreme weather phenomena such as wind gusts, heavy precipitation, hail, thunderstorms, tornadoes, and many others usually occur when there is a change in air mass the passing of front over certain region. The climatology fronts difficult, since they are drawn onto maps manually by forecasters; therefore, data concerning them limited process itself very subjective nature. In this article, we propose an objective method for determining position based on random forest machine learning technique, digitized from DWD database, ERA5 meteorological reanalysis. Several aspects leading to improvement scores presented, adding new fields or dates training database using gradients fields.