Predictive GIS-Based Model of Rockfall Activity in Mountain Cliffs

Abstract

Rockfall susceptibility has been analysed in mountain cliffs of the Cantabrian Range, North Spain. The main aim of this analysis has been to build a predictive model of rockfall activity from a low number of environmental and geological variables. The rockfall activity has been quantified in a GIS. The cartographic information used shows the spatial distribution of all the recent talus screes as well as their associated source areas in the rock-slopes. The area relation At/Ar (recent talus scree polygon/source basins) in the rock slopes has been used as the rockfall activity indicator. This relation has been validated in 50 pilot rock-slopes and compared with the relation number of recent rock fragments/source basin, obtained from field work. The environmental factors causing rockfall depend on the rock slope situation, and these are: altitude and sun radiation on the rock cliff. The geological factors considered are: lithology, relative position of the main discontinuities with respect to the topographic surface and two morphologic parameters: the roughness and slope gradient. A logistic regression analysis has been applied to a population of 442 limestone and quartzite rock cliffs. The dependent variable is the rockfall activity indicator, which allows the definition of two classes of rock cliff units: low and high activity. The independent variables are altitude, sun radiation (equinox radiation, summer solstice radiation, winter solstice radiation), slope roughness, slope gradient,anisotropy and lithology. Results suggest that it is possible tobuild a valid cartographic predictive model for rockfall activity in mountain rock cliffs from a limited number of easily obtainable variables. The method is especially applicable in massive rock slopes or in regions with uniform rock mass characteristics.