Application of an analog downscaling technique to the assessment of future landslide activity-a case study in the Italian Alps

Abstract

Slope stability and hence landslide activity is in many cases related to climate, which influences groundwater and pore pressure fluctuations of hillslopes. An approach is presented which transforms transient GCM output by statistical downscaling to local precipitation scenarios, which together with directly derived temperature scenarios are subsequently fed into a slope hydro- logical/stability model to derive future landslide activity. This model chain is applied to a landslide in the Dolomites, Italy. Validation of the approach against independent observed records suggests its applicability for estimating future landslide activity based on GCM results. One possible way of esti- mating the quality of the approach is to determine sources of uncertainty introduced by the GCM sim- ulations and by different fitting periods of the downscaling technique. Differences between the GCM experiments are found to be more important than differences between the 2 fitting periods. The most striking result is the significant reduction of landslide activity in spring in all cases. This is attributable to the rise of winter temperature which impedes future storage of winter precipitation as snow. As one consequence, less melt water is available for the hillslope in spring, causing the decrease in activity. It can be concluded that differences between GCMs and differences between fitting periods permit quantification of part of the uncertainty inherent in climate change impact assessments. Impacts which emerge in all model combinations, such as the decrease in landslide activity in spring shown here, have a high level of confidence.