Learning the Low Frequency Earthquake Activity on the Central San Andreas Fault. Issue 13 (3rd July 2021)

• Record Type: Journal Article
• Authors: Johnson, Christopher W.; Johnson, Paul A.
• Is Part Of: Geophysical research letters. Volume 48:Issue 13(2021)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2021-07-03
• Subjects: low‐frequency earthquake -- machine learning -- San Andrea Fault -- tremor; Geophysics -- Periodicals; Planets -- Periodicals; Lunar geology -- Periodicals; 550
• DOI: 10.1029/2021GL092951
• Languages: English
• ISSNs: 0094-8276

Abstract

Abstract: Low frequency earthquakes (LFEs) originating below the central San Andreas Fault are associated with slow‐slip beneath the seismogenic zone within the more ductile portion of the crust. Monitoring efforts over 15 years detected >1 million LFEs. We train a gradient boosted tree model using statistical features describing the seismic waveforms to estimate the hourly LFE event count. The burst‐like LFE behavior is reproduced, while lower amplitudes are predicted during the most active periods. The hourly event counts are up to 18% greater than the catalog. The ability to continuously monitor LFE activity provides insight to when geodetic measurements of slow slip are possible, without the need for developing a computational‐intensive template‐matching catalog. Similar waveform statistical features are found between detecting LFEs and tremors, which provides additional evidence tremors are composed of LFEs. The approach extracts information contained in continuous seismic waveforms that might benefit detecting precursory signals. Plain Language Summary: Low frequency earthquakes (LFEs) are a class of events occurring deep in the fault core beneath the seismogenic zone. This type of event has been observed along the central San Andrea Fault and occurs much more frequently than regular earthquakes. This study applies gradient boosted tree models using statistical features derived from continuous daily seismic waveforms to train a model that is capable of estimating the hourly LFE event count. Inferring the hourly rate of LFEs allows continuous monitoring of the fault zone using statistical features of daily seismic waveforms, without developing a computationally expensive LFE catalog. Bursts of these events are associated with deep slow‐slip at the base of the fault that is integral to quantifying the entire slip budget. The model uses features that quantify the energy released and varying frequency content in daily seismic waveforms to estimate the LFE activity. Similarities are found between monitoring for LFEs and detecting tremors, providing additional evidence that tremors are composed of LFEs. The technique exemplifies the abundant information contained in seismic waveforms that can be applied to training machine learning models to identify processes deep in the fault zone, with the potential to obtain a deeper understanding of slip events. Key Points: Gradient boosted tree regression model estimates the number of low frequency earthquakes per hour from seismic waveform statistical features Machine learning model reproduces bursts of LFE activity and predicts more events occurring than previously cataloged Average daily estimate of LFE count is similar to catalog values and a useful monitoring technique to track deep creep on the fault