A high-resolution globally applicable model for near-real-time estimation of earthquake-induced landslides

Abstract:

Earthquake-induced landslides are a prominent secondary hazard of seismic activity, causing a significant portion of earthquake-related fatalities and economic losses, sometimes with long-lasting negative societal impacts. We present a new, high-resolution, globally applicable, statistical model to estimate the likelihood of seismically induced landslides across the globe in near-real-time. This empirical landslide probability model uses ground shaking estimates from USGS ShakeMap software, together with globally available landslide-susceptibility proxies. A previous version of the model has been adapted for use in the USGS Ground Failure product, which runs in near-real time for significant earthquakes around the globe. An important update to this model is the inclusion of higher resolution (3 arc-second, ~90 m) topographic data to more accurately represent the slope component of landslide vulnerability. We have incorporated more than 10 newly available landslide inventories into the model for training and testing purposes. To address known deficiencies in the current model, we test additional landslide-susceptibility factors including proxies for near-surface material strength, such as soil thickness, soil type, sediment thickness, proxies for landcover, such as percent green vegetation, and climatic parameters, such as the mean annual precipitation, maximum temperature, and precipitation seasonality. The models can be used to rapidly provide regional estimates of the probability and distribution of seismically induced landslides, as well as their potential impact on the surrounding population, within minutes of the occurrence of any significant earthquake globally. Other applications include estimating ground-failure impacts from earthquake scenarios and, when applied to known values of probabilistic ground motion, producing landslide susceptibility maps. The near real-time information on earthquake secondary effects produced by the models can be used to improve earthquake response, loss estimation, and hazard mitigation.