Site-specific selection of conditional spectrum-based motions through modified stochastic ground motion modeling

Abstract

Despite the utilization of the prominent ground motion selection and modification (GMSM) method—the exact conditional spectrum (CS-exact)—for site-specific ground motion selection, limited ground motion availability may result in records that may not fully represent regional site characteristics or capture the underlying distribution of causal parameters. In this article, we explore an alternative record selection approach, termed conditional spectrum utilizing stochastic ground motion modeling (CS-SGMM), to select site-specific CS-based ground motions by leveraging an illustrative site-based stochastic ground motion model, specifically aimed at addressing the challenges caused by data limitations. This approach involves modifying the parameters of the stochastic ground motion model through constrained optimization to match the target CS within the desired period range of interest while aligning with regional trends. This involves incorporating causative parameters to select site-specific ground motions matching the target CS. Subsequently, we implement a redistribution procedure to ensure that the selected ground motions effectively represent the distribution of causal scenarios, thereby achieving higher hazard consistency in CS-based selection. Illustrative examples from a site in the Western United States demonstrate the effectiveness of our approach across different structural periods and ground motion intensity levels. Finally, we evaluate the viability of our approach by comparing the selected set of ground motions with those chosen using contemporary GMSM methods, such as CS-exact and generalized conditional intensity measure (GCIM), which select records based on spectral shape and both spectral shape and duration, respectively.