Avalanche Protection Capacity and Disturbance Dynamics of Mountain Forests

Abstract

One of the most important forest functions in the mountains is the protection capacity, safeguarding many people and their infrastructure against natural hazards. The attention on forest protection capacity complements other topics (e.g. related to post-disturbance dynamics in mountain forests) discussed in all the presented papers. This thesis aims to improve our knowledge on representation of changing forests and vegetated areas in avalanche simulations, and on the post-disturbance forest development after two typical Central European disturbance agents: bark beetle and windthrow. In the beginning of this thesis I compared various remote sensing methods (lidar and photogrammetry) for determining forest parameters (as tree height, crown cover or surface roughness), to the field investigation in total of 107 plots. These parameters were further used in avalanche simulations using the simulation tool RAMMS (Paper 1). The use of lidar or photogrammetry to estimate forest parameters as input for avalanche simulation proved appropriate, as the simulated runout distances did not differ significantly from the observed runout. The correct estimation of forest cover in release areas using photogrammetry was challenging but crucial as it may strongly affect the simulated runout distance of avalanches. The input data for this estimation should be as recent as possible in order to avoid an incorrect estimation of forest cover influencing the avalanche simulation. The precise estimation of forest cover in avalanche release areas is particularly important in the treeline ecotone (where sparse vegetation may be neglected in avalanche simulations) and in disturbed forests (which are expected to increase). Treeline ecotones and disturbed forests with high surface roughness have thus potentially important effects on avalanches. We thus tested several roughness algorithms in order to improve the quantification of terrain roughness in treeline ecotones and disturbed forests (Paper 2). The algorithm vector ruggedness measure was selected based on our analysis as the best performing algorithm in distinguishing between different roughness classes. We also proposed an improvement for roughness in gullies, which are typically overestimated in hazard modelling. To overcome this issue, we proposed to adopt the directional roughness, which may be applied using e.g. the algorithm standard deviation (SD) of residual topography, calculating surface roughness in the direction of flow. Surface roughness maps, based as well on digital surface models (DSMs), may serve as a good basis to improve natural hazard modelling. The two following papers aimed at better understanding the disturbance dynamics in forests impacted by bark beetle infestation (Paper 3) and windthrow (Paper 4). Post-disturbance forest development was studied in the field examining parameters on deadwood and tree regeneration (e.g. number of trees, tree species, tree height, deadwood cover, decay stage of deadwood). As deadwood decays, trees are growing taller and denser. Thirty years after both disturbances, forests showed good recovery. Based on our analysis, the future forest is expected to be more diverse with higher shares of broadleaves and with different age classes thanks to secondary tree regeneration growing on deadwood. Since trees establish only on rather decayed deadwood, this regeneration comes after at least 20 years from disturbance. Furthermore, the protection capacity was monitored over the years. We used the surface coverage of deadwood and trees creating surface roughness as a proxy for the protection capacity. After the bark beetle disturbance, we also assessed the protection capacity using avalanche simulations (modelled with RAMMS) based on remote sensing data. We observed that disturbed forests maintain their protection capacity for some time. In the case of windthrow through high surface roughness created by large amounts of lying deadwood and root plates, and in the case of bark beetle by long-standing dead trees. The protection capacity reaches its minimum after 10-15 years after disturbance with slight delay in the case of bark beetle infestation, as the decay proceeds slower. Remote sensing methods serve for different purposes and are suitable in delivering products like digital models. Such models are valuable in estimating forest parameters, calculating surface roughness or as a basis for avalanche modelling. Surface roughness calculated from DSM may represent some of the underestimated forest areas as young and shrub forests or changing forests and, therefore, improve the estimation of protective capacity of such areas. Analysis of data from repetitive fieldwork after bark beetle and measuring forest parameters at different stages after a windthrow revealed that the minimum in protective capacity is reached at similar time after these two common forest disturbances. It remains to be verified if remote sensing methods as photogrammetry, are able to deliver similar data on tree regeneration and deadwood, which are often being overgrown by other vegetation; and in case of lidar method the possibility of automatic object segmentation and extrapolation over larger areas.