From scenario-based seismic hazard to scenario-based landslide hazard: rewinding to the past via statistical simulations

Abstract The vast majority of statistically-based landslide susceptibility studies assumes the slope instability process to be time-invariant under definition that “the past and present are keys future”. This assumption may generally valid. However, trigger, it a rainfall or an earthquake event, clearly varies over time. And yet, temporal component trigger is rarely included in only confined hazard assessment. In this work, we investigate population landslides triggered response 2017 Jiuzhaigou ( $$M_w = 6.5$$ M w = 6.5 ) including associated ground motion analyses, these being carried out at Slope Unit (SU) level. We do by implementing Bayesian version Generalized Additive Model assuming across SUs study area behaves according Bernoulli probability distribution. procedure would produce map reflecting spatial pattern specific therefore limited use for land planning. implement first analytical step reliably estimate effect, its distribution, on unstable SUs. then assume effect time-invariant, enabling statistical simulations any scenario occurred from 1933 2017. As result, obtain full spectrum potential coseismic patterns last century compress information into model/map representative all possible since 1933. backward can also further exploited opposite direction where, accounting scenario-based motion, one forward future slopes.