Comparative study of decoupled- and coupled-based predictive models for seismically-induced slope displacements during shallow crustal earthquakes

Various predictive models for the seismically-induced displacement (D) of flexible slopes during shallow crustal earthquakes are available in the literature, but the model-to-model difference of predictions remains unclear. This paper conducts a comprehensive comparative study of ten representative D models based on scenario- and probabilistic-based displacement analyses. Explicit expressions are provided to implement the probabilistic displacement hazard approach, in which copula theory is utilized for deriving the joint occurrence probability of ground-motion parameters. A total of 6048 displacement hazard analyses are performed considering different earthquake and slope scenarios. It is found that the model-to-model difference generally becomes larger with increasing the natural period (Ts) of a sliding mass, whereas the within-type difference among different coupled models is notably smaller than that of the decoupled models. The coupled-type models generally result in smaller total variability of D as compared to the decoupled-type models. Meanwhile, the coupled models tend to, on average, yield larger (or smaller) hazard-consistent D than the decoupled models for Ts larger (or smaller) than an intersection period Ts,inter, though the difference between the two model types diminishes as Ts approaches zero. The effects of ground-motion model selection and ground-motion parameter correlation are also discussed. This study could hopefully provide insights into the displacement model selection for the seismic slope performance assessment and regional landslide hazard mapping.