Training exercise

Task 1

The Tavolese catchment in the Tuscany region of central Italy is well known for the occurrence of landslides, particularly after heavy rainfall when the soil is completely saturated with water. The regional government asks your consultancy firm to produce a landslide susceptibility map for that area. Your tasks are as follows:

• Generation of a map where susceptible areas are distinguished from less susceptible areas on the basis of raster cells;
• Estimation of the validity of this map with the available landslide inventory;
• Sensitivity analysis with regard to the cohesion and the angle of internal friction. Thereby, the ranges of cohesion and internal friction angle given below should be subdivided into three subranges of equal interval each, so that you arrive at 9 combinations of parameter ranges (i.e. nine parameter spaces). The landslide susceptibility should be calculated and validated separately for each of these nine parameter spaces.

The following data are available:

• A digital terrain model with a resolution of 20 x 20 m (tv_elev) and a hillshade (tv_hillshade) als ESRI rasters;
• The landslide inventory (tv_inventory - 1=landslide, 0=no landslide) as ESRI raster;
• The data on the geotechnical parameters are, unfortunately, very sparse or even nonexistent. However, after having consulted some geotechnical textbooks you know that the cohesion most likely ranges between 0 and 9 kN/m² and the internal friction angle most likely ranges between 21 and 45�. You can assume a dry density of the material of 1500 kg/m³ and a saturated water content of 40 Vol.-%. You can further assume a soil depth of 3 m, corresponding to the depth of the sliding surface.

Data access: Tavolese

The deliverable is a report containing at least the following elements:

• The landslide susceptibility map (directly inserted into the report);
• The result of the validation (a suitable validation parameter as well as its interpretation);
• The result of the sensitivity analysis (one suitable validation parameter for each model run, all of them summarized in a table, and the corresponding interpretation);
• A brief documentation of the work flow. For the case that the work flow has been automatized through a script, also this script should be added;
• A brief discussion on which issues you have to take into account when interpreting the results of the model and of the validation.

Task 2

In task 1, the soil depth - and therefore also the depth of the sliding surface - was assumed constant (3 m). Now it is your task to investigate whether the spatial pattern of landslide susceptibility corresponds better to the distribution of the observed landslides when assuming deeper soils in areas with a tendency to accumulate material (e.g. in valleys) than in those areas without such a tendency (e.g. on ridges). You have the same data available as for task 1. In addition you have the spatial distribution of soil depth available as ESRI raster (tv_soildepth). You should use the geotechnical parameter space which has resulted in the best model performance in task 1.

The deliverable is a report with:

• The two landslide susceptibility maps (constant vs. variable soil depth);
• Both validation results and their interpretation;
• A brief documentation of the work flow. For the case that the work flow has been automatized through a script, also this script should be added;
• A brief discussion on which issues you have to take into account when interpreting the results of the model and of the validation.

Return to Course Home