GIS-based modeling for evaluation of wind damage probabilityin forests in Southwest Germany

摘要

The weights of evidence (WofE) methodology and logistic regression models (LRM) were usedfor GIS-based estimation of the wind damage probability (PDAm) in the forests across the federalstate of Baden-Wuerttemberg (Southwest Germany). P_(DAM)-calculations were based on the stormdamages caused by the severe winter storms `Vivian'/`Wiebke' and 'Lothar'. For each stormevent six out of 20 tested evidential themes were favorably associated with wind damage in theforests. Out of these six evidential themes soil type, geology, and soil moisture were importantpredictor variables for PDAm after both storm events. Further important evidential themes relatedwith `Vivian'/`Wiebke' were slope, altitude, and soiSdrainage. Maximum gust speed, foresttype, and soil acidification were further predictor variables that were favourably associated withPDAM after 'Lothar'. On the basis of the combined evidential themes, for each storm event GIS-based wind damage probability maps were produced. These maps were then classified intozones of high, moderate and low wind damage probability. The WofE-methodology predictedwind damage with an overall success rate of 76 % after `Vivian'/ `Wiebke' and 78 % after 'Lo-thar'. The area under the ROC curve calculated from the LRM results for PDAM after `Vivian'/`Wiebke' equalled to 78 %. After 'Lothar' the area under the ROC curve amounted to 79 %.The most wind damage-prone regions within the study area are the northern Black Forest, thenorthern Upper Rhine Valley, and the Swabian Alb.