A strategy for GIS-based 3-D slope stability modelling over large areas

• Main Authors: M. Mergili, I. Marchesini, M. Alvioli, M. Metz, B. Schneider-Muntau, M. Rossi, F. Guzzetti
• Format: Article
• Language: English
• Published: Copernicus Publications 2014-12-01
• Series: Geoscientific Model Development
• Online Access: http://www.geosci-model-dev.net/7/2969/2014/gmd-7-2969-2014.pdf

GIS-based deterministic models may be used for landslide susceptibility mapping over large areas. However, such efforts require specific strategies to (i) keep computing time at an acceptable level, and (ii) parameterize the geotechnical data. We test and optimize the performance of the GIS-based, 3-D slope stability model r.slope.stability in terms of computing time and model results. The model was developed as a C- and Python-based raster module of the open source software GRASS GIS and considers the 3-D geometry of the sliding surface. It calculates the factor of safety (FoS) and the probability of slope failure (<i>P</i>_f) for a number of randomly selected potential slip surfaces, ellipsoidal or truncated in shape. Model input consists of a digital elevation model (DEM), ranges of geotechnical parameter values derived from laboratory tests, and a range of possible soil depths estimated in the field. Probability density functions are exploited to assign <i>P</i>_f to each ellipsoid. The model calculates for each pixel multiple values of FoS and <i>P</i>_f corresponding to different sliding surfaces. The minimum value of FoS and the maximum value of <i>P</i>_f for each pixel give an estimate of the landslide susceptibility in the study area. Optionally, r.slope.stability is able to split the study area into a defined number of tiles, allowing parallel processing of the model on the given area. Focusing on shallow landslides, we show how multi-core processing makes it possible to reduce computing times by a factor larger than 20 in the study area. We further demonstrate how the number of random slip surfaces and the sampling of parameters influence the average value of <i>P</i>_f and the capacity of r.slope.stability to predict the observed patterns of shallow landslides in the 89.5 km² Collazzone area in Umbria, central Italy.