Seismically induced landslide hazard and exposure modelling in Southern California based on the 1994 Northridge, California earthquake event

Quantitative modelling of landslide hazard, as opposed to landslide susceptibility, as a function of the earthquake trigger is vital in understanding and assessing future potential exposure to landsliding. Logistic regression analysis is a method commonly used to assess susceptibility to landsliding; however, estimating probability of landslide hazard as a result of an earthquake trigger is rarely undertaken. This paper utilises a very detailed landslide inventory map and a comprehensive dataset on peak ground acceleration for the 1994 M_w6.7 Northridge earthquake event to fit a landslide hazard logistic regression model. The model demonstrates a high success rate for estimating probability of landslides as a result of earthquake shaking. Seven earthquake magnitude scenarios were simulated using the Open Source Seismic Hazard Analysis (OpenSHA) application to simulate peak ground acceleration, a covariate of landsliding, for each event. The exposure of assets such as population, housing and roads to high levels of shaking and high probabilities of landsliding was estimated for each scenario. There has been urban development in the Northridge region since 1994, leading to an increase in prospective exposure of assets to the earthquake and landslide hazards in the event of a potential future earthquake. As the earthquake scenario magnitude increases, the impact from earthquake shaking initially increases then quickly levels out, but potential losses from landslides increase at a rapid rate. The modelling approach, as well as the specific model, developed in this paper can be used to estimate landslide probabilities as a result of an earthquake event for any scenario where the peak ground acceleration variable is available.