Regional-scale liquefaction analyses

Regional-scale liquefaction hazard analyses are necessary for resilience planning and prioritization of seismic upgrades for critical distributed infrastructure such as levees, pipelines, roadways, and electrical transmission facilities. Two approaches are often considered for liquefaction hazard analysis of distributed infrastructure: (1) conventional, site-specific probe or borehole-based analyses, which do not quantify the uncertainty between investigation locations; or (2) surface geology-based analyses, which often neglect localized geotechnical properties and include a great amount of uncertainty. We describe an analytical method to unify the disparate site-specific and deposit-scale approaches using Gaussian processes. We use borehole data to produce spatial fields of random variables for liquefaction triggering analyses, such as groundwater elevation, soil texture classification, penetration resistance, and cyclic resistance ratio that converge to the site-specific uncertainty at sampling locations but also quantify the uncertainty in-between sampling locations. We demonstrate the effectiveness of Gaussian process models for regional-scale liquefaction hazard analyses in two example studies in Washington state and California, US.