Tectonic origins of seismic velocity changes inferred from ambient seismic noise in 18 years at Parkfield

Abstract:

We conduct auto- and cross- correlation of ambient seismic noise from the High Resolution Seismic Network (HRSN) at Parkfield, CA from 2002-2020. The goal of this study is to elucidate the tectonic mechanisms controlling observed relative velocity changes (dv/v). Previous studies have observed a coseismic decrease in dv/v associated with the 2004 Parkfield earthquake, and its recovery which is thought to reflect the damage and subsequent long-term healing of the SAF. This study focuses on additional temporal variations observed in the dv/v time history and explores possible connections between both fast and slow earthquakes, tectonic tremor, tidal and thermoelastic strains. In order to avoid and mitigate bias in the observed velocity change caused earthquakes and non-volcanic tremors we develop an algorithm to remove transient signals from continuous seismic waveforms using a combination of kurtosis and STA/LTA. After removal of the spurious signals, we use a wavelet transform and explore various stacking methods (linear, selective, robust, phase-weighted) to enhance the convergence of stacked correlation functions and continuous dv/v measurement. In order to handle the large amount of data analyzed, we improve the computational efficiency we use the Julia-based tools, SeisIO and SeisNoise. The seasonal variation caused by thermoelastic strains are then estimated and removed. Finally, we compare independent catalogs of seismic tremors and the evolution of surface strain to identify the signature of tectonic strain change remaining in our dv/v time history. We find that the long-term post-earthquake healing of the 2004 Parkfield earthquake continues today, and the current dv/v is equivalent to, or larger than, the level before the mainshock. Additionally, we investigate the coseismic velocity changes associated with regional earthquakes, including the 2014 South Napa and the 2019 Ridgecrest earthquakes.