Ground motion prediction using ambient seismic noise on a large-N array in the LA basin

Julian F. Schmitt, Tim Clements, Nan Wang, Kim B. Olsen, & Marine A. Denolle

Published August 14, 2020, SCEC Contribution #10579, 2020 SCEC Annual Meeting Poster #212 (PDF)

The presence of a seismic waveguide can significantly amplify ground motion from large earthquakes, as shown by numerical simulations. For example, the predictions of shaking in Los Angeles due to a large, southern San Andreas fault (SAF) earthquake depend on the presence, or absence of, a seismic waveguide in the northern basins of the greater Los Angeles. However, the accuracy of the ground motion predictions depends critically on the details of the basin structure along the wave guide. The use of cross correlation of ambient seismic noise has the potential to both improve our knowledge of the velocity structure for numerical simulations and provide more accurate predictions with virtual earthquakes as the ambient-field Green’s function encodes the basin structure.

In this study we perform an in-depth comparison between point source theoretical Green’s functions and the ambient-field Green’s function obtained from a dense, large-N nodal and broadband array in the greater Los Angeles area, the BASIN experiment. Using temporary and permanent stations located along the SAF as virtual sources and combining the nodal arrays with the SCSN stations, we compare all 9 components of the correlation tensor and focus on Love and Rayleigh wave properties.

The analysis informs numerical simulations of the seismic wavefield which uses a 3D velocity model via the AWP-ODC finite-difference solver. Confronted with the size of the large-N nodal array’s and permanent stations, we use Julia language, a high performance scripting language, and AWS cloud computing to process the N-squared cross-correlations. With 14 source stations located along the SAF and the dense spatial array covering more than 300 permanent and nodal receiver stations in the LA area, the calculation of all 9 components yields ~40,000 correlations. The dense coverage with station pairs in the area between downtown LA and the SAF is expected to dramatically improve our knowledge of the basin structure and therefore ground motion prediction.

We consider the cloud computing framework and high performance dynamic computing languages like Julia the next direction for ambient noise computation, and present this methodology alongside fundamental descriptions of the 3D wavefield using the empirical ambient-field and theoretical Green’s function.

Key Words
Ground motion prediction,ambient seismic noise,large-N,LA basin,AWS,Julia

Citation
Schmitt, J. F., Clements, T., Wang, N., Olsen, K. B., & Denolle, M. A. (2020, 08). Ground motion prediction using ambient seismic noise on a large-N array in the LA basin. Poster Presentation at 2020 SCEC Annual Meeting.

Related Projects & Working Groups
Ground Motions