ABSTRACT In ground-motion models (GMMs), site terms are typically derived from site-specific velocity profiles, such as VS30 (the time-averaged shear-wave velocity to a depth of 30 m) and Z1.0 (the depth to the 1.0 km/s shear-wave velocity horizon). However, these variables are not universally spatially available. This study presents a proof-of-concept model for linear site effects in GMMs for California, developed using broadly accessible geospatial datasets with spatial coverage across the entire study area. The model incorporates data from 95,898 recordings at 2,339 stations, as provided by the DesignSafe Ground-Motion Database (Ji et al., 2022). Mapped geospatial variables, such as sediment thickness, elevation, and surficial geologic units, are compiled for each station. The target site amplification of peak ground acceleration, peak ground velocity, and pseudospectral accelerations (PSA) from 0.01 to 10 s is decomposed from the residuals of the Boore et al. (2014) (BSSA14) GMM. A mixed-effects linear regression model is developed using the compiled geospatial variables and validated through fivefold cross-validation. A novel mapped categorical variable, the regional geological unit—defined as a combination of geomorphic provinces and geological units—is introduced due to its strong correlation with site amplification. The geospatial-based site term shows a substantial reduction in site-to-site variability for site amplifications across all frequencies compared to the conventional velocity profile-based site terms; on average, a 10.6% reduction in site-to-site variability is achieved compared to BSSA14 (with a 9.5% reduction at a PSA of 1 s). The model is an improved ergodic model, constructed using geospatial parameters that are mapped across California. This study illustrates the effectiveness of geospatial data as site proxies in developing site terms for GMMs and highlights the potential of geospatial variables for predicting site effects, especially in areas with sparse ground-motion records or limited site-specific velocity measurements.
Roberts et al. (2025) studied this question.