Räumliche und zeitliche Vorhersage der Eintrittswahrscheinlichkeit eines ertragsgefährdenden Mehltauereignisses an Winterweizen mit der Random-Forest-Methode

Wheat is one of the most grown crops in the world. The yield of the wheat can be reduced by fungal infections, like the powdery mildew. The prediction of infestations, that can endanger the yield, is of corresponding importance. This report presents an approach to predict such infestation events in a spatial and temporal dimension, considering hourly weather and longtime climate data. Furthermore it is based on a random forest simulation. This prediction resulted in 71.9 % correct prognoses applied to the German state Schleswig-Holstein considering uninfluenced validation years.