SCEC Award Number 22130 View PDF
Proposal Category Individual Proposal (Integration and Theory)
Proposal Title CANVAS: An Adjoint Waveform Tomography Model of California and Nevada
Investigator(s)
Name Organization
Richard Allen University of California, Berkeley
Other Participants Claire Doody
SCEC Priorities 4b, 3b, 1c SCEC Groups CXM, CS, SAFS
Report Due Date 03/15/2023 Date Report Submitted 05/10/2023
Project Abstract
 As the interests of SCEC expand beyond Southern California, high-quality velocity models with coverage across the entirety of the state are necessary to understand fault structure, earthquake source parameters, and earthquake ground motions. Therefore, we present the results of the California-Nevada Adjoint Simulations model (CANVAS), an adjoint waveform tomography model of California and Nevada iterated to a minimum period of 12 seconds. In order to calculate CANVAS, we began by testing the effect of different starting models on the final inversion results, and showed that when using a conservative, multi-scale inversion approach, the starting model has minimal effect on final inversion results using both model-space and data-space comparison metrics (Doody et al., 2023). With an intermediate version of the model at a minimum period of 20 seconds, we also inverted for moment tensor solutions using 3D Green’s Functions; the inverted moment tensors had an average of 54% higher variance reduction compared to the Global Centroid Moment Tensor (GCMT) catalogue (Dziewonski et al., 1981; Ekström et al., 2012) and more closely resembled local catalogue solutions from the Northern California Earthquake Data Center and the Southern California Earthquake Data Center (Doody et al., 2022). Finally, we used the inverted moment tensor parameters to calculate CANVAS down to 12 seconds minimum period. CANVAS significantly improves the dispersed surface wave compared to the starting model, WUS256 (Rodgers et al., 2022) and resolves the basement depth of major basins throughout California. We hope CANVAS can serve as a starting model for smaller-scale crustal tomography models throughout California.
Intellectual Merit This project produced an adjoint waveform tomography model of California and Nevada that could be an important contribution to the SCEC Community Velocity Model. The model not only resolves known tectonic structure and basement depths of large basins, but also shows accurate waveform fits, particularly to dispersed surface waves. This model could therefore be a good candidate for a starting model for other statewide velocity models or for regional crustal tomography models.
Broader Impacts This project partially funded a graduate student during the 2022-2023 academic year. Accurate velocity models are also necessary for rupture propagation modelling, which are used for seismic hazard analysis.
Exemplary Figure Figure 2. Cross-sections of the crust and uppermost mantle in CANVAS. The map on the left shows the location of each cross-section. Section A-A’ (39˚N) shows the vertical extent of the Geysers-Clear Lake low velocity anomaly (GCL). Section B-B’(36˚N) includes the Tulare Basin (TB), which is the southernmost extent of the Central Valley, the Sierra Nevada (SN) and its root, the Isabella Anomaly (IA),and a small portion of the Basin and Range province (B&R). Section C-C’ (33˚N) shows the Ventura Basin (VB), which extends to about 18km depth in our models. Finally, Section D-D’ (32˚N) passes through the Salton Sea (SS). The sharp contrasts in each cross-section denote the Moho, which is held fixed in the model and uses Moho depths from Crust1.0 (Laske et al., 2013).