Precipitation and soil accumulation history modifies future landslide hazard

Abstract:

Landslides are a major global geohazard that are predicted to increase as anthropogenic climate change drives an increase in landslide-triggering storms. Humid mountains may be particularly important, as rainfall-induced shallow landsliding causes a significant proportion of global landslide fatalities. While precipitation is a significant driving force, future landslide susceptibility also depends on millennial-scale landslide history that limits the distribution of potential landslide material. However, the influence of landslide history on current and future landslide hazard is poorly understood.

We address this problem by first quantifying the distribution of shallow landslide potential across 1347 km² of the southern Appalachian Mountains using an unprecedented empirical dataset of hillslope soil depths and strength parameters. By accounting for landslide history, estimates of future landslide potential are lowered significantly. Slope stability modelling demonstrates that under current conditions, only 38% of potential landslide sites across the landscape could fail, regardless of the size of the storm. Of susceptible slopes, most can only fail during the largest possible precipitation events. This is because once a landslide occurs it takes thousands of years to accumulate enough soil to make a site unstable during precipitation. In contrast, the return period of large storms is tens to hundreds of years. This result challenges whether increases in precipitation predicted by climate models will lead to measureable increases in landslide frequency.

Next, we examine how the distribution of potential landslide material changes through time as storm-induced landslides periodically remove material, using a coupled hillslope stability and soil accumulation model applied to the Appalachian landscape. Our results reveal the spatial pattern of temporal variability in landslide potential, which represents a neglected source of uncertainty when assessing regional landslide susceptibility. Constraining landslide history through analysis of historical and paleo-climatic data, therefore provides a potential means of reducing this uncertainty, by making susceptibility assessments dynamic through time.