Improving the resolution of co-seismic velocity change monitoring at active fault zones using the ambient seismic field

Jared T. Bryan, Kurama Okubo, Congcong Yuan, & Marine A. Denolle

Published August 14, 2019, SCEC Contribution #9585, 2019 SCEC Annual Meeting Poster #043

The release of accumulated strain energy during dynamic earthquake ruptures causes stress relaxation around the fault. Off-fault fractures are co-seismically generated around the rupture front due to high stress concentrations, and this damage can be intensified by geometrical complexities such as fault roughness, kinks, and edges. Co-seismic stress relaxation and off-fault damage contribute to the sudden change of elastic properties in the Earth’s crust. Here, we attempt to continuously monitor changes of the medium to validate whether or not these models are correct.

Material properties can be imaged using continuous measurements of ambient noise cross-correlation functions. The relative phase change between correlations sampled at unit intervals (e.g. days, months) is proportional to the perturbation in seismic velocity, dv/v. Accurate measurements of these phase changes require a sufficiently high signal-to-noise ratio, which is often achieved by averaging the correlations over long time periods and over multiple station pairs, reducing spatial and temporal resolution. In this study, we explore ways of improving spatial and temporal resolution by increasing the convergence rate, rather than the stacking duration, of the cross-correlation functions.

We first develop a transient signal removal algorithm that combines STA/LTA and kurtosis detection techniques. We then perform a selective stacking, where we threshold the incoherent short-time windowed cross-correlation functions against the reference cross-correlation function and omit them while stacking over the time interval. These techniques improve the convergence rate in both ballistic and coda sections of the stacked cross-correlation functions.

We then conduct a dv/v analysis using the moving-window cross-spectrum (MWCS), trace stretching, and dynamic time warping (DTW) methods, coupled with the wavelet transform. We compare the results obtained through these techniques to evaluate the robustness of the dv/v measurements. We perform a continuous monitoring of the region associated with the 2004 Parkfield earthquake. We explore the spatial correlation between our mapped dv/v and the coseismic stress change inferred from kinematic slip models, as well as aftershock locations.

Key Words
Ambient Noise, Parkfield, Velocity Change Monitoring

Citation
Bryan, J. T., Okubo, K., Yuan, C., & Denolle, M. A. (2019, 08). Improving the resolution of co-seismic velocity change monitoring at active fault zones using the ambient seismic field. Poster Presentation at 2019 SCEC Annual Meeting.

Related Projects & Working Groups
Seismology