Characterizing the scale of regional landslide triggering from storm hydrometeorology

Abstract. Rainfall strongly affects landslide triggering; however, understanding how storm characteristics relate to the severity of landslides at the regional scale has thus far remained unclear, despite the societal benefits that would result from defining this relationship. As mapped landslide inventories typically cover a small region relative to a storm system, here we develop a proxy for landslide-inducing rainfall, A*, based on extremes of modelled soil water relative to its local climatology. We calibrate A* using four landslide inventories, comprising over 11,000 individual landslides over four unique storm events, and find that a common threshold can be applied to estimate regional shallow landslide triggering potential across diverse climatic regimes in California (USA). We then use the spatial distribution of A*, along with topography, to calculate the landslide potential area (LPA) for nine landslide-inducing storm events over the past twenty years, and test whether atmospheric metrics describing the strength of landfalling storms, such as integrated water vapor transport, correlate with the magnitude of hazardous landslide-inducing rainfall. We find that although the events with the largest LPA do occur during exceptional atmospheric river (AR) storms, the strength of landfalling atmospheric rivers does not scale neatly with landslide potential area, and even exceptionally strong ARs may yield minimal landslide impacts. Other factors, such as antecedent soil moisture driven by storm frequency, and mesoscale precipitation features within storms, are instead more likely to dictate the patterns of landslide-generating rainfall throughout the state.