Complete reply on CC1

Abstract

Worldwide, shallow landslides repeatedly pose a risk to infrastructure and residential areas. To analyse and predict the risk posed by shallow landslides, a wide range of scientific methods and tools to model shallow landslide probability exist for both local and regional scale However, most of these tools do not take the protective effect of vegetation into account. Therefore, we developed SlideforMap (SfM), which is a probabilistic model that allows for a regional assessment of shallow landslide probability while considering the effect of different scenarios of forest cover, forest management and rainfall intensity. SfM uses a probabilistic approach by distributing hypothetical landslides to uniformly randomized coordinates in a 2D space. The surface areas for these hypothetical landslides are derived from a distribution function calibrated from observed events. For each randomly generated landslide, SfM calculates a factor of safety using the limit equilibrium approach. Relevant soil parameters, i.e. angle of internal friction, soil cohesion and soil depth, are assigned to the generated landslides from normal distributions based on mean and standard deviation values representative for the study area. The computation of the degree of soil saturation is implemented using a stationary flow approach and the topographic wetness index. The root reinforcement is computed based on root proximity and root strength derived from single tree detection data. Ultimately, the fraction of unstable landslides to the number of generated landslides, per raster cell, is calculated and used as an index for landslide probability. Inputs for the model are a digital elevation model, a topographic wetness index and a file containing positions and dimensions of trees. We performed a calibration of SfM for three test areas in Switzerland with a reliable landslide inventory, by randomly generating 1000 combinations of model parameters and then maximising the Area Under the Curve (AUC) of the receiver operation curve (ROC). These test areas are located in mountainous areas ranging from 0.5–7.5 km2, with varying mean slope gradients (18–28°). The density of inventoried historical landslides varied from 5–59 slides/km2. AUC values between 0.67 and 0.92 indicated a good model performance. A qualitative sensitivity analysis indicated that the most relevant parameters for accurate modeling of shallow landslide probability are the soil depth, soil cohesion and the root reinforcement. Further, the use of single tree detection in the computation of root reinforcement significantly improved model accuracy compared to the assumption of a single constant value of root reinforcement within a forest stand. In conclusion, our study showed that the approach used in SfM can reproduce observed shallow landslide occurrence at a catchment scale.