Regional estimation of liquefaction-induced ground deformations using a data-informed probabilistic approach

This paper presents an approach for estimating liquefaction-induced ground deformations on a regional scale in support of risk analyses of spatially distributed infrastructure systems. Logic trees are used to represent uncertainty in subsurface conditions (i.e., results from cone penetration testing and the water table depth) as well as uncertainty between models for liquefaction susceptibility, triggering, and deformations. Finally, a Gaussian process model is used to assign realistic spatial patterns of liquefaction-induced ground failure. The approach described here emphasizes the uncertainty in regional liquefaction risk analysis and takes measures to reduce the potential for overestimation of this risk. The method was developed for application in analyzing the risk faced by spatially-distributed gas infrastructure in California.