Based on individual tree damage data dating back to the gale “Lothar” (winter 1999) in Baden-Württemberg, Germany, a statistical model was developed to estimate the risk of storm damage for individual trees. The data were compiled from the National German Forest Inventory. The model attempts to separate the effects of tree-specific variables, topography, site conditions and flow field related effects on damage probability. The crucial problem of missing information on the actual flow field parameters was solved by applying a generalized additive model that enables the simultaneous fit of a spatial trend function. The geographical location of risk hotspots as predicted by the model correspond well to the actual distribution pattern of storm damage as assessed by the forest service. Tree height proved to be one of the most important factors affecting the level of damage, while height to diameter at breast height ratio influences damage probability to a much lesser extent. The Norway spruce ( Picea abies (L.) Karst.) group has the highest potential to be damaged followed by the silver fir ( Abies alba Miller) – Douglas-fir ( Pseudotsuga menziesii (Mirb.) Franco) group and the Scots pine ( Pinus sylvestris L.) – larches ( Larix spp.) group. Predicted probabilities for deciduous trees are generally lower than those of conifers. West- to south-exposed locations bear a considerably higher damage risk and waterlogged soils show an increased predicted probability compared with slightly or not waterlogged soils.

Forest health monitoring schemes were set up across Europe in the 1980's in re sponse to concern about air pollution related forest die back (Waldsterben) and have continued since then. Recent threats to forest health are climatic extremes likely to be due to global climate change, increased ground ozone levels and nitrogen deposi tion. We model yearly data on tree crown defoliation, an indicator of tree health, from a monitoring survey carried out in Baden-Württemberg, Germany since 1983. On a changing irregular grid, defoliation and other site specific variables are recorded. In Baden-Württemberg the temporal trend of defoliation differs between areas because of site characteristics and pollution levels, making it necessary to allow for space-time in teraction in the model. For this purpose we propose to use generalized additive mixed

We investigated the effects of site properties, forest structure, and time on snow breakage, insect outbreaks, windthrow, and total damage for predominantly planted forests. A time series of forest damage in southwestern Germany spanning 77 years, from 1925 to 2001, was available along with a database on site properties and forest structure. The statistical modeling procedure successively addressed (i) probability of damage occurrence, (ii) timber loss in damaging events, and (iii) interaction among damage agents over time. Logistic and linear regressions were combined with multivariate autoregressive techniques. Natural disturbances were responsible for a total timber loss of 3.0 m 3 Á ha -1 Á year -1 . The distribution of the timber loss values over the years and over sites and stands with different properties was modeled with a standard error of 6.7 m 3 Á ha -1 Á year -1 . Disturbances are more likely to occur in previously damaged stands. Storm events typically provoke subsequent insect outbreaks between 2 and 6 years later. Large windthrow and snow breakage events tend to occur periodically, once every 10th, 11th, or 15th year. Analysis of disturbances as a time series significantly enhances understanding of forest risk processes.Résumé : Nous avons étudié les effets des propriétés de la station, de la structure de la forêt et du temps sur les bris causés par la neige, les épidémies d'insectes et les chablis ainsi que sur les dommages totaux dans des forêts issues principalement de plantations. Une série temporelle des dommages causés aux forêts couvrant une période de 77 ans, de 1925 à 2001, dans le sud-ouest de l'Allemagne était disponible ainsi qu'une base de données sur les propriétés de la station et la structure de la forêt. La procédure de modélisation statistique a successivement abordé : (i) la probabilité que surviennent des dommages, (ii) la perte de matière ligneuse lors d'événements entraînant des dommages et (iii) l'interaction dans le temps entre les agents responsables des dommages. Des régressions logistiques et linéaires ont été combinées à des techniques autorégressives multidimensionnelles. Les perturbations naturelles étaient responsables d'une perte totale de 3,0 m 3 Áha -1 Áan -1 de matière ligneuse. La distribution de ces pertes dans le temps et parmi les stations et les peuplements avec différentes propriétés a été modélisée avec un écart type de 6,7 m 3 Áha -1 Áan -1 . Des perturbations ont plus de chances de survenir dans les peuplements déjà endommagés. Typiquement, les tempêtes sont suivies d'épidémies d'insectes deux à six ans plus tard. Les chablis importants et les épisodes de bris causés par la neige ont tendance à survenir périodiquement à tous les 10, 11 ou 15 ans. L'analyse des perturbations sous forme de série temporelle améliore de façon significative la compréhension des processus de risques en forêt.[Traduit par la Rédaction]