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 #172, SCEC Community Models (CXM)

A Level-Set Approach to Parsimoniously Updating the SCEC CVMs

Jack B. Muir, Robert W. Clayton, & Victor C. Tsai
Poster Image: 

Poster Presentation

2020 SCEC Annual Meeting, Poster #172, SCEC Contribution #10482 VIEW PDF
Effectively combining local high-resolution tomographic results with existing regional models presents an ongoing challenge as the SCEC community velocity models (CVMs) continue to mature. Currently extant boundaries between high and low resolution areas in the CVM-H model have lead to velocity models that contain geologically improbable recto-linear structures that produce false coherent scattering in wavefield simulations. In contrast, with the CVM-S model, smoothing introduced by full waveform tomography has attenuated the high-frequency scattering that are observed in the data. We have recently developed a framework based on the Tikhonov-regularized level-set method that provides a simpl...e means to infer where local data provides meaningful constraints on seismic observables beyond those found in the existing CVMs. This technique defines a volume within which updates are made to a reference CVM, with the boundary of the volume being part of the inversion rather than explicitly defined. By appropriately penalizing the complexity of the boundary, a minimal update which best explains the data is achieved.

To test this framework, we utilize the high-resolution data from the Community Seismic Network (CSN), a large 400-station permanent urban deployment, to invert Love wave dispersion, derived from eikonal tomography of two-station cross-correlation travel-time delays, and relative amplification data from the Mw 7.1 July 5 2019 Ridgecrest Earthquake. We invert for an update to CVMS-4.26 using the Tikhonov Ensemble Kalman Inversion scheme, a highly efficient derivative-free optimizer. We find that the Ridgecrest Earthquake data is best explained by a deepening of the LA Basin (compared to the CVMS-4.26 reference model) along its Northwest-Southeast axis relative to its deepest point, just south of downtown LA. Additionally, the deeper basin edge extends further to the East of downtown LA towards East Los Angeles. This result offers new progress towards the parsimonious incorporation of detailed local basin models within regional reference models utilizing an objective inverse-problem framework, and highlights the importance of accurate basin geometry models when accounting for the potentially significant amplification of surface waves from regional earthquakes in the high-rise building band.