Massimiliano Alvioli, Michele Santangelo, Federica Fiorucci, Mauro Cardinali, Ivan Marchesini, Paola Reichenbach and Mauro Rossi, 2021, A data-driven method for assessing the probability for terrain grid cells of initiating rockfalls on a large area, Geomorphometry 2021, Perugia, 13-17 settembre 2021,
URL: http://www.cnr.it/prodotto/i/460033

Rockfalls are one harmful kind of landslide, due to their rapidity, destructive potential and high probability of occurrence on steep topographies, often found along transportation corridors. Various factors can trigger rockfalls, including intense rainfall and seismic activity, and diverse phenomena affect their occurrence, like rock weathering and fracturing. Existing approaches for the assessment of rockfall susceptibility range from purely phenomenological to purely deterministic, physically based methods. A common requirement for many approaches is the need to locate the potential point locations of source areas, often located uphill on cliffs. Application of a physically based model, in particular, allows the calculation of material runout stemming from rockfalls originating from such point locations. In this work, we propose a method for the location of rockfall source points, on a digital elevation model, suitable for large areas. We deem the method as data-driven, because it relies on expert delineation of potential source areas from Google Earth images in few sample locations, representative of the study area at large. We measure the slope distribution of grid cells encompassed by expert-mapped source areas, and generalize the distribution of sources to the whole of the study area. We apply the method to a corridor of about 17,000 km in length and varying width, containing the entire Italian railway network. The map of source areas represents the main input for a physically based simulation of rockfall trajectories with the model STONE, and likely of other similar physically based or phenomenological models for rockfall runout assessment.

Data from https://intranet.cnr.it/people/