Forest Damage by Extra-Tropical Cyclone Klaus-Modeling and Prediction

Windstorms may have negative consequences on forest ecosystems, industries, and societies. Extreme events related to extra-tropical cyclonic systems remind us that better recognition understanding of the factors driving damage are needed for more efficient risk management planning. In present study, we statistically modelled caused by windstorm Klaus in south-west France. This event occurred 24 January 2009 severe maritime pine (Pinus pinaster) stands. We aimed at isolating best potential predictors can help build predictive models damage. applied random (RF) technique find classifiers binary response variable. Five-fold spatial block cross-validation, repeated five times, forward feature selection (FFS) were control model over-fitting. addition, variable importance (VI) accumulated local effect (ALE) plots used as performance metrics. The RF was prediction probability mapping. ROC AUC 0.895 0.899 training test set, respectively, while accuracy 0.820 0.837 set. FFS allowed isolate most important predictors, which distance from trajectory, soil sand fraction content, MODIS normalized difference vegetation index (NDVI), wind exposure (WEI). general, their influence positive a wide range observed values. area applicability (AOA) confirmed be construct map almost entire study area.