Exciting news! We're transitioning to the Statewide California Earthquake Center. Our new website is under construction, but we'll continue using this website for SCEC business in the meantime. We're also archiving the Southern Center site to preserve its rich history. A new and improved platform is coming soon!

Poster #211, Ground Motions

Ground-Motion Time-Series Interpolation within the Community Seismic Network using Gaussian Process Regression: Application to the 2019 Ridgecrest Earthquake

Aidin Tamhidi, Nicolas M. Kuehn, Monica D. Kohler, Farid Ghahari, Ertugrul Taciroglu, & Yousef Bozorgnia
Poster Image: 

Poster Presentation

2020 SCEC Annual Meeting, Poster #211, SCEC Contribution #10623 VIEW PDF
At the present time, the density of ground motion recording stations within California is sparse. As a rough estimate, there are about 2000 recording stations within California. The Community Seismic Network is one of the strong-ground-motion networks within the state that was designed to densify monitoring in seismically active regions such as greater Los Angeles. This network includes 400 microelectromechanical accelerometers installed in Los Angeles Unified School District school campuses. After an earthquake, ground-motion intensity measures (GMIMs) can be estimated at locations between stations by employing Ground Motion Prediction Equations (GMPEs) or interpolated using regression mode...ls. However, the estimated GMIMs are not adequate if nonlinear time history analysis of structures at uninstrumented locations is required. Therefore, the estimation of the entire ground-motion time series is crucial in order to recognize the dynamic responses of the structures. In addition, one can have a better damage and loss estimation by estimating the entire ground motion time series at various uninstrumented locations after an earthquake, which will provide the responsible agencies with an accurate post-earthquake assessment. In this study, an interpolation method to predict the entire ground motion time series from neighboring stations’ recorded motions is presented. The interpolation process is done using a Gaussian Process regression, which models the real and imaginary parts for various frequencies as random Gaussian variables. The Gaussian Process regression model eventually estimates the entire ground motion time series, given the geographic coordinates and site conditions of the target locations. The training and verification of the proposed model are carried out using physics-based simulated ground motions of the 1906 San Francisco earthquake. Eventually, the 257 ground-level recording stations of the Community Seismic Network that recorded the 2019 Ridgecrest earthquake within Los Angeles are employed to test and evaluate the performance of the Gaussian Process regression model.