San Andreas Fault dip, Peninsular Ranges mafic lower crust and partial melt in the Salton Trough, Southern California, from ambient‐noise tomography

Abstract

We use ambient-noise tomography to improve CVM-H11.9, a community velocity model of southern California. Our new 3-D shear-velocity model with 0:05 x 0:05 lateral and 1 km vertical blocks reveals new structure beneath the San Andreas Fault (SAF), Peninsular Ranges batholith (PRB), southern Sierra Nevada batholith (SNB), and the Salton Trough (ST). We use 4 years of data recorded on 849 broadband stations, vastly more than previous studies and including our own broadband Salton Seismic Imaging Project, a 40 station transect across the ST, as well as other campaign stations in both Mexico and the United States. Mean lower crust and upper mantle wave speeds (3.6 km/s at 20 km, 4.2 km/s at 40 km) are low by global standards. Across the SAF, southeast of San Gorgonio Pass, we observe vertical to steeply dipping lateral velocity contrasts that extend beneath the Moho. Beneath the western PRB and westernmost southern SNB, we observe relatively high shear velocities (!3.8 km/s) in the lower crust that we interpret as the mafic roots of the overlying arc. Relatively high-velocity upper mantle (up to $4.5 km/s) may be part of the intact arc, or possibly a remnant of the Farallon plate. Beneath the ST, we observe zones of low shearvelocity in the lower crust and upper mantle which permit up to $4.5% melt in the lower crust and up to $6% melt in the upper mantle, depending on the assumed composition and pore geometry. Our results preclude the existence of older continental crust beneath the ST and support the creation of new crust beneath the ST.

Eikonal Tomography of the Southern California Plate Boundary Region

“…Consistent with the increasing histogram peak velocity with period shown in Figure , faster velocities are observed for the isotropic phase and group maps at longer periods. The obtained Rayleigh wave phase and group speed maps are generally similar to results from previous studies (Barak et al, ; Roux & Ben‐Zion, ; Zigone et al, ) in the overlapping area, with our phase and group velocities being slightly slower beneath basins (e.g., LA basin and Ventura basin) and showing sharper velocity contrast across the SAF southeast to the SGP at shorter periods (<7 s).…”

“…To maximize the number of measurements at shorter periods (e.g., ≤3 s) while simultaneously minimizing false detections, we perform the automatic picking procedure not only on the symmetric signals but also on both the positive and negative time lags of the correlations. Comparisons between results obtained from different components of the correlation functions help filter out erroneous measurements identified through inconsistency. The determination of phase travel time dispersion picking employs model‐predicted travel time dispersion curve from the CVM‐S4.26 to avoid cycle skipping ( N in equation ). The group travel time dispersion picked using the method of several earlier studies (e.g., Barak et al, ; Zigone et al, ) is found to be sensitive to noise and has larger uncertainty. We therefore derive group travel time dispersion using the obtained phase travel time dispersion following equation , which improves the accuracy of the measurements (Figures d and a). In addition to the quality control criteria developed in Lin et al (), we introduce station configuration error to identify regions that have unreliable gradient estimates due to poor data coverage (section Figure S5).…”

“…This suggests that although Rayleigh wave group velocities at 3 s have sufficient sensitivity and can constrain the Vs structures in the top 3 km (e.g., Barak et al, 2015;Zigone et al, 2015), the Vs values in the top 3 km from such 1-D inversion are nonunique and likely be biased by the initial model. The resulting tomographic model of Vs using Rayleigh wave data is consistent overall with previous inferences on the large-scale velocity structure in SC (e.g., Allam et al, 2012;Barak et al, 2015;Berg et al, 2018;Fang et al, 2016;Lee et al, 2014;Share et al, 2019;Tape et al, 2010;Zigone et al, 2015). However, we find a large discrepancy between the surface wave dispersions obtained in this study and those predicted by the SCEC community velocity models, particularly for regions inside the basins and around fault zones ( Figures S11 and S15).…”

“…While consistent phase travel time dispersion curves are obtained from all three cases, significant discrepancies are observed between the t g ( ω ) dispersion curves, especially at longer periods (>15 s), suggesting that the peak of envelope function is sensitive to noise and often associated with large uncertainties (Figures d and a). Therefore, in this study, instead of determining group travel time based on t g ( ω ) (e.g., Barak et al, ; Zigone et al, ), we simply derive the group travel time from the smoothed phase dispersion using the following relation: where υ ph and υ gp represent the smoothed phase and resulting group dispersions in the form of average velocities (i.e., υ ph = Δ/ t ph , υ gp = Δ/ t gp ) and λ is the wavelength given by 2 π ⋅ υ ph / ω . Here, we use the observed phase dispersion to first invert for a 1‐D Vs model and then predict the smoothed phase velocity dispersion to stabilize the first derivative in equation (Figure ).…”

Subsurface Geometry of the San Andreas Fault in Southern California: Results from the Salton Seismic Imaging Project (SSIP) and Strong Ground Motion Expectations

“…The underthrusting in this region has recently been supported and emphasized by analysis of ambientnoise tomography (Barak et al, 2015). In a cross section between San Gorgonio and Cajon Passes, mafic lower crust of the Peninsular Ranges (Pacific plate) appears to underthrust North America by at least 30 km (Fig.…”