Determining fault zone structure and examining earthquake early warning signals using large datasets of seismograms

Seismic signals associated with near-fault waveforms are examined to determine fault zone structure and scaling of earthquake properties with event magnitude. The subsurface structure of faults is explored using fault zone head and/or trapped waves, while various signals from the early parts of seismograms are investigated to find out the extent to which they scale with magnitude.;Fault zone trapped waves are observed in three arrays of instruments across segments of the San Jacinto fault. Similarly to previous fault zone trapped wave studies, the low velocity damage zones are found to be 100-200m wide and extend to a depth of ∼3-5km. Observation and modeling indicate that the damage zone was asymmetric around the fault trace. A similar sense of damage asymmetry was observed using detailed geological mapping by Dor et al. (2006) nearby on the San Jacinto fault at Anza. Travel time analysis and arrival time inversions of fault zone head waves were used to produce high resolution images of the fault structure of the San Andreas fault south of Hollister. The contrast of P wave velocities across the fault was found to be ∼50% in the shallow section, lowering to 10-20% below 3 km, with the southwest side having faster velocities. Inversions making use of different subsets of stations suggest that a low velocity damage zone also exists in this area and that it is more prominent on the faster velocity side of the fault. The patterns of damage from these studies of fault zone head waves and trapped waves are consistent (Ben-Zion and Shi, 2005) with the theoretical prediction that earthquake ruptures on these fault sections have statistically-preferred propagation directions.;The early parts of P waveforms are examined for signals that have previously been proposed to scale with the final event magnitude. Data from Turkey and a deep South African gold mine show that scaling is present in signals related to the maximum displacement amplitude and frequency content. The high sampling rate of the instruments in the gold mine enables the reduction of the time window in which measurements are made to below the estimated rupture duration of the largest events. Using increasingly small time windows has only a minimal effect on the scaling of the signals with event magnitude, implying that the size of earthquakes is affected statistically by some property of the early part of the rupture.