Landslide hazard scenarios based on both past landslides and precipitation

Abstract

The goal of this contribution is to develop a set of methods and techniques for modelling landslide hazard, in order to obtain better predictions and, therefore, to reduce the risk associated to this type of process. The research has been carried out in North-Western Guipúzcoa (Spain), an area intensely affected by shallow landslides. Hazard predictions must be based on several assumptions, termed scenarios. To propose more realistic hazard scenarios, two different approaches have been applied: on one hand, by extrapolating the trends observed from recent past landslide activity and, on the other hand and indirectly, from the empirical analyses between landslides and their main trigger (precipitation). It was necessary to get long time series of landslide occurrences, through which sound future frequency could be estimated and relationships between landslides and their predisposing factors established. The landslide inventory includes 20 time intervals ranging from the 50s of last century to the present and has been obtained by analyzing photographic images from different dates. Rainfall records of several weather stations have been analyzed, establishing correlations between landslides and precipitation intensity-duration. According to rainfall quantity, duration and intensity, as well as landslide activity, six multiple occurrence regional landslide events (MORLE), regarding rainfall extreme events, have been identified and characterized. On the basis of both types of scenarios quantitative hazard models could be obtained.