SHALLOW LANDSLIDES HAZARDS IN A CHANGING CLIMATE

Rainfall-triggered shallow landslides pose hazards to communities, infrastructure, and ecosystems. In future climates the magnitude and frequency of extreme precipitation are expected to change, but the resulting effects on landslide abundance, size, and spatial distribution are not well understood. Changes in extreme precipitation can be relatively greater than those in mean precipitation and are thought to depend on local orography. We assess relative changes in extreme precipitation in the USA over the periods 1971-2000 to 2041-2070 using regional climate models (RCMs). We analyze topography and local winds associated with extreme precipitation to delineate areas where orography may alter the pattern of these extremes. We quantify precipitation changes on the lee and windward slopes and verify that RCMs reflect theoretical predictions. To assess the impacts of extreme precipitation changes on landslide characteristics, we apply a search algorithm that predicts landslide abundance, location, and size to a study site in the Oregon Coast Range (OCR) with a landslide inventory. We test a range of precipitation scenarios, forest management practices, and antecedent moisture conditions. We rescale observed precipitation for representative lee and windward orographic locations and find that fractional changes in mean winter precipitation are ~3 times larger on leeward slopes. Fractional changes in extreme precipitation intensity are much greater than those of mean precipitation, and increase with return period. In the Pacific Northwest, leeward increases are ~10% for 2-year events and ~20% for 30-year events. At our study site, a 20% precipitation or antecedent moisture increase corresponds to a 30-40% increase in landslide volume, but up to 250% increases are predicted leeward of the OCR and in undisturbed forest. This suggests that the largest relative change in future landslide hazards may occur in areas that currently experience only infrequent landsliding. Communities in these locations are likely to be less prepared and therefore more vulnerable than those in locations of high susceptibility. The modeled landslide response to climate change may reduce the dependence of communities on memory of historic events for their awareness of landslide risk.