The discovery of nonvolcanic tremors in the region of Parkfield-Cholame, California, along the major San Andreas Fault in 2005 (Nadeau and Dolenc, 2005) has given a new perspective on these deep long-duration, high-frequency (1 to 15 Hz) events, which previously were only observed along subduction zones. The California tremors show many of the characteristics of the subduction zone events in terms of depth, frequency range, waveform, and the absence of apparent P- and S- waves. However they are shorter in duration, up to tens of minutes versus hours to days. The presence in the area of five seismic networks, including the High Resolution Seismic Network (HRSN), gives the opportunity to detect and locate the tremor activity over a large region, even for distances approaching 200 km between stations. We performed a multi-year tremor analysis between August 2001 and 2008 and we studied the influence of the 2003 M6.5 San Simeon and 2004 M6.0 Parkfield earthquakes, the largest events to occur in the study region and time period, on the Parkfield-Cholame tremor activity.
Using a similar cross-correlation detection method described by Obara (2002), we searched nearly 7 years of tremor activity starting in August 2001 and found more than 1,700 nonvolcanic tremors with a cumulative duration of more than 9,700 minutes (Figure 2.47). The locations of the tremors based on envelope cross-correlation and station pair time delay indicate that the events are mainly distributed across the San Andreas Fault over a 10-15 km wide area offset to the west of the fault, beneath Cholame, California (Figure 2.48).
The tremor catalog was compared to the catalog of M0+ earthquakes of the same 40-by-40 km region centered on the town of Cholame, California. Analysis of the spatio-temporal evolution of the tremors during the seven years of the study (Figure 2.47) revealed a strong correlation between tremor rates and occurrences of the two largest earthquakes of the region: the 2003 M6.5 San Simeon and 2004 M6.0 Parkfield earthquakes. Following the two mainshocks, the tremor activity increased in a step-like pattern, which was persistent over several months. We noticed a step change of a factor of 3 in the cumulative duration of tremor activity over 3 months before and after the San Simeon earthquake. Similar observations followed the Parkfield earthquake, when the cumulative duration of the events increased by a factor of 5 between 3 months before and 3 months after the mainshock. The tremor rate changes were not the consequence of the emergence of longer tremors but of a larger tremor frequency. On the other hand, no increase in the number of small earthquakes was observed above the tremors after San Simeon (70 km away from tremor zone). However, we noticed a very strong increase (34 times the background level) in the seismicity rate after Parkfield. The close proximity of the Parkfield earthquake rupture to the tremor region explains the large number of aftershocks following it (Figure 2.48).
Also, following the two events, an aftershock-like decay of the tremor activity was observed. Aftershock sequences are evidence of re-adjustements of the stress field in the seismogenic zone after a mainshock. The similar pattern in the tremor activity at Parkfield suggests that tremor activity is also related to stress change.
We performed Coulomb stress changes maps (Figure 2.48) of the two strong earthquakes using the finite-source models computed by Rolandone et al. (2005) and Kim and Dreger (2008), respectively, to define the rupture zone. The Coulomb failure stress changes were calculated for vertical planes parallel to the San Andreas Fault (140 strike, 90 dip and 180 rake) at the median depth of the tremors (i.e. 20 km) and at seismogenic depth (i.e. 8 km) in two 10-by-10 km boxes centered on the tremor and earthquake locations. Following the 2003 San Simeon earthquake, the static Coulomb stress increased up to 12.5 kPa (0.125 bars) at 20 km depth and to 14.4 kPa (0.144 bars) at the earthquake depth. Typical stress changes inducing triggered earthquakes are on the order of 1 to 10 bars (Stein, 2004). The small stress changes observed following San Simeon at 8 km depth explained the absence of earthquake activation. However, changes in the tremor rate for less than 12.5 kPa stress increase suggest that tremors can be stimulated by significant low static stress changes. Following Parkfield, up to a 5-bar increase was transmitted into the earthquake zone, in agreement with the change recorded in the seismicity. At 20 km depth, the static stress changes were also higher after Parkfield than after San Simeon, with a maximum of 22.9 kPa (0.229 bars). Evidences of correlation between higher tremor rate and higher stress changes suggest that tremors are sensitive to stress variations at depth and that the degree of tremor activation is related to the level of stress change experienced.
Since the Parkfield earthquake (from 2005 to present), the tremor activity has remained elevated relative to before the San Simeon earthquake (Figure 2.47). We have also noticed the emergence of quasi-periodic bursts of tremor activity starting after the end of the Parkfield aftertremor sequence in 2005. The episodes of tremors were not observed before the San Simeon earthquake or before the Parkfield earthquake. They consist of periods of higher frequency of the number of tremors and not of an increase in the duration (or amplitude) of individual tremors. The analysis of the episodes shows that the recurrence time between episodes is lengthening from 50 days in 2005 to 110 days in 2008, following a linear trend. The episodes of tremors may suggest very small stress variations in the deeper part of the crust generated after the Parkfield earthquake and possibly related to postseismic transients in depth.
The origin of the nonvolcanic tremors has yet to be determined; however several hypotheses have been proposed: fluid migration from subduction processes, shear coupling at depth, hydraulic fracturing, tidal stress variations, and small dynamic stress changes (less than 43 kPa) during the passage of teleseism surface waves. Our results suggest that tremors react to even smaller static stress change (less than 23 kPa) transmitted in their generating region. The degree of tremor activation also reflects the level of stress change at depth, and elevated activity persits well beyond the aftershock decay period following Parkfield.
Kim, A., and D. S. Dreger (2008), Rupture Process of the 2004 Parkfield Earthquake from Near-Fault Seismic Waveform and Geodetic Records, J. Geophys. Res., 113, B07308, doi:10.1029/2007JB005115.
Nadeau, R. M., and D. Dolenc (2005), Nonvolcanic Tremors Beneath the San Andreas Fault, Science, 307, 389.
Obara, K. (2002), Nonvolcanic Deep Tremor Associated with Subduction in Southwest Japan, Science, 296, 1,679-1,681.
Rolandone, F., D. Dreger, M. Murray, and R. Burgmann (2006),Coseismic Slip Distribution of the 2003 Mw6.6 San Simeon Earthquake, California, Determined from GPS Measurements and Seismic Waveform Data, Geophys. Res. Lett., 33, L16315.
Stein, R. S. (2004), Tidal Triggering Caught in the Act, Science, 305, 1,248-1,249.
Berkeley Seismological Laboratory
215 McCone Hall, UC Berkeley, Berkeley, CA 94720-4760
Questions or comments? Send e-mail: firstname.lastname@example.org
© 2007, The Regents of the University of California