Parkfield Mainshock Research Articles

Seismic attenuation and stress on the San Andreas Fault at Parkfield: are we critical yet?

The Parkfield transitional segment of the San Andreas Fault (SAF) is characterized by the production of frequent quasi-periodical M6 events that break the very same asperity. The last Parkfield mainshock occurred on 28 September 2004, 38 years after the 1966 earthquake, and after the segment showed a ∼22 years average recurrence time. The main reason for the much longer interevent period between the last two earthquakes is thought to be the reduction of the Coulomb stress from the M6.5 Coalinga earthquake of 2 May 1983, and the M6 Nuñez events of June 11th and 22 July 1983. Plausibly, the transitional segment of the SAF at Parkfield is now in the late part of its seismic cycle and current observations may all be relative to a state of stress close to criticality. However, the behavior of the attenuation parameter in the last few years seems substantially different from the one that characterized the years prior to the 2004 mainshock. A few questions arise: (i) Does a detectable preparation phase for the Parkfield mainshocks exist, and is it the same for all events? (ii) How dynamically/kinematically similar are the quasi-periodic occurrences of the Parkfield mainshocks? (iii) Are some dynamic/kinematic characteristics of the next mainshock predictable from the analysis of current data? (e.g., do we expect the epicenter of the next failure to be co-located to that of 2004?) (iv) Should we expect the duration of the current interseismic period to be close to the 22-year “undisturbed” average value? We respond to the questions listed above by analyzing the non-geometric attenuation of direct S-waves along the transitional segment of the SAF at Parkfield, in the close vicinity of the fault plane, between January 2001 and November 2023. Of particular interest is the preparatory behavior of the attenuation parameter as the 2004 mainshock approached, on both sides of the SAF. We also show that the non-volcanic tremor activity modulates the seismic attenuation in the area, and possibly the seismicity along the Parkfield fault segment, including the occurrence of the mainshocks.

A High-Resolution Earthquake Catalog for the 2004 Mw 6 Parkfield Earthquake Sequence Using a Matched Filter Technique

Abstract We present the high-resolution Parkfield matched filter relocated earthquake (PKD-MR) catalog for the 2004 Mw 6 Parkfield earthquake sequence in central California. We use high-quality seismic data recorded by the borehole High Resolution Seismic Network combined with matched filter detection and relocations from cross-correlation derived differential travel times. We determine the magnitudes of newly detected events by computing the amplitude ratio between the detections and templates using a principal component fit. The relocated catalog spans from 6 November 2003 to 28 March 2005 and contains 13,914 earthquakes, which is about three times the number of events listed in the Northern California Seismic Network catalog. Our results on the seismicity rate changes before the 2004 mainshock do not show clear precursory signals, although we find an increase in the seismic activity in the creeping section of the San Andreas fault (SAF) (about ∼30 km northwest of the mainshock epicenter) in the weeks prior to the mainshock. We also observe a decrease in the b-value parameter in the Gutenberg–Richter relationship in the creeping section in the weeks prior to the mainshock. Our results suggest stress is increasingly released seismically in the creeping section, accompanied by a decreasing aseismic creeping rate before the mainshock occurrence. However, b-value and seismicity rates remain stable in the Parkfield section where the 2004 mainshock ruptured. This updated catalog can be used to study the evolution of aftershocks and their relations to afterslip following the 2004 Parkfield mainshock, seismicity before the mainshock, and how external stresses interact with the Parkfield section of the SAF and the 2004 sequence.

Why the M w 6 parkfield earthquake expected in the 1985–1993 interval was postponed till 2004?

The earthquake prediction released by USGS in 1985 to the public stated that there was an M w6.0 earthquake occurring in the Parkfield area by January 1993 with a 95% chance. This prediction has proved unsuccessful. Some researchers analyzed the reasons for the failure of the prediction on the basis of statistics. In this study, we explored this problem by means of b-value and the Earth's rotation-correlated seismicity. The results show that an increasing trend in b-value occurred prior to May 1992 and a decreasing trend with a relative decrease of more than 35% from June 1992 to early 1999. In several years before the 2004 M w 6.0 Parkfield earthquake, the b-value remained stable at the lowest level. A significant correlation between seismic activity and Earth's rotation occurred only when the 2004 mainshock was approaching. It has been well- known that when a stressed focal medium is approaching failure due to action of a critical stress, the focal medium will reach a critical state. At this point, seismicity could be significantly correlated to Earth's rotation. So, significantly Earth's rotation-related seismic activity indicated that the focal medium in the Parkfield area was at a critical state. Since the decline in b-value corresponds to increasing stress and the significantly Earth's rotation-related seismic activity indicates the critical state of the focal medium, we suggest that two precursors are the decline (several years to more than a decade) in b-value and the significantly Earth's rotation-related seismic activity. These two precursors did not emerge during or before the 1985 Parkfield earthquake prediction period. Instead, they appeared preceding the 2004 M w 6.0 Parkfield mainshock. This may be the reason that the M w 6 Parkfield earthquake did not occurred till 2004.

Modulation of Seismic Attenuation at Parkfield, Before and After the 2004 M6 Earthquake

AbstractThe crack density within a fault's damage zone is thought to vary as seismic rupture is approached, as well as in the postseismic period. Moreover, external stress loads, seasonal or tidal, may also change the crack density in rocks, and all such processes can leave detectable signatures on seismic attenuation. Here we show that attenuation time histories from the San Andreas Fault at Parkfield are affected by seasonal loading cycles, as well as by 1.5–3‐year periodic variations of creep rates, consistent with Turner et al. (2015, https://doi.org/10.1002/2015JB011998), who documented a broad spectral peak, between 1.5 and 4 years, of the spectra calculated over the activity of repeating earthquakes, and over InSAR time series. After the Parkfield main shock, we see a clear modulation between seismic attenuation correlated to tidal forces. Opposite attenuation trends are seen on the two sides of the fault up to the M6.5 2003 San Simeon earthquake, when attenuation changed discontinuously, in the same directions of the relative trends. Attenuation increased steadily of over one year on the SW side of the San Andreas Fault, until the San Simeon earthquake, whereas it decreased steadily on the NE side of the San Andreas Fault, roughly for the six months prior to the event. Random fluctuations are observed up to the 2004 M6 Parkfield main shock, when rebounds in opposite directions are observed, in which attenuation decreased on the SW side, and increased on the NE side.

Changes in repeating earthquake slip behavior following the 2004 Parkfield main shock from waveform empirical Green's functions finite‐source inversion

AbstractFinite‐source inversions are performed using small earthquake waveforms as empirical Green's functions (eGf) to investigate the rupture process of repeating earthquakes along the San Andreas Fault in Parkfield, California. The eGf waveform inversion method is applied to a repeating Mw 2.1 Parkfield earthquake sequence using three‐component velocity waveforms recorded by an array of borehole seismometers. The obtained models show a circular slip distribution with a ~20 m radius, a 3.0–4.2 cm average slip of the main asperity, and peak displacement of 10.6–13.5 cm. The static stress drop distribution shows that the main asperity has a peak stress drop of 69.5–94.7 MPa. The inversion results support an earlier finding by Dreger et al. (2007) that high‐strength asperities exist in the rupture areas of the Mw 2.1 events at Parkfield. In addition, notable temporal peak slip and stress drop reduction was observed after the 2004 Parkfield event while the average value remains constant (~12 MPa) over time. These events may represent mechanically strong sections of the fault, surrounded by regions that are undergoing continuous deformation (creep), Given repeated loading of the strong asperities, it would be expected that these similar repeating earthquakes should also have very similar slip distributions since surrounding regions are deforming aseismically. There are small differences in the waveforms of these repeating earthquakes, and this could be because of rupture nucleation points not being in exactly the same location within the region of the fault that is capable of stick‐slip behavior. Our result indicates that waveform slip inversion is needed to reveal spatial and temporal variations of the stress drop within the rupture area to improve understanding of fault healing and rupture mechanics.

Frequency‐dependent effects of rupture for the 2004 Parkfield main shock, results from UPSAR

AbstractThe frequency‐dependent effects of rupture propagation of the Parkfield, California, earthquake (28 September 2004, M6) to the northwest along the San Andreas Fault can be seen in acceleration records at UPSAR (USGS Parkfield Seismic Array) in at least two ways. First, we can see the effects of directivity in the acceleration traces at UPSAR, which is about 11.5 km from the epicenter. Directivity or the seismic equivalent of a Doppler shift has been documented in many cases by comparing short‐duration, high‐amplitude pulses (P or S) in the forward direction with longer‐duration body waves in the backward direction. In this case we detect a change from a relatively large amplitude, coherent, high‐frequency signal at the start of rupture to a low‐amplitude, low‐coherent, low‐frequency signal at about the time the rupture front transfers from the forward azimuth to the back azimuth at about 34–36 s (time is UTC and is the seconds after day 272 and 17 h and 15 min. S arrival is just after 30 s) for rays leaving the fault and propagating to UPSAR. The frequency change is obvious in the band about 5 to 30 Hz, which is significantly above the corner frequency of the earthquake (about 0.11 Hz). From kinematic source models, the duration of faulting is about 9.2 s, and the change in frequency is during faulting as the rupture extends to the northwest. Understanding the systematic change in frequency and amplitude of seismic waves in relation to the propagation of the rupture front is important for predicting strong ground motion. Second, we can filter the acceleration records from the array to determine if the low‐frequency energy emerges from the same part of the fault as the high‐frequency signal (e.g., has the same back azimuth and apparent velocity at UPSAR), an important clue to the dynamics of rupture. Analysis of sources of strong motion (characterized by relatively high frequencies) compared to kinematic slip models (relatively low frequency) for the 11 March 2011 Tohoku earthquake as well as Maule (27 February 2010) and Chi‐Chi (20 September 1999) earthquakes show that high‐ and low‐frequency sources do not have the same locations on the fault. In this paper we filter the accelerograms from UPSAR for the 2004 main shock in various passbands and then recompute the cross correlations to determine the vector slowness of the incoming waves. At Parkfield, it appears that for seismic waves with frequencies above 1 Hz, there is no discernible frequency‐dependent difference in source position (up to 8 Hz) based on estimates of back azimuth and apparent velocity. However, at lower frequencies, sources appear to be from shallower depths and trail the high frequencies as the rupture proceeds down the fault. This result is greater than one standard deviation of an estimate of error, based on a new method of estimating error that is a measure of how broad the peak in correlation is and an estimate of the variance of the correlation values. These observations can be understood in terms of a rupture front that is more energetic and coherent near the front of rupture (radiating higher frequencies) and less coherent and less energetic (radiating in a lower frequency band) behind the initial rupture front. This result is a qualitative assessment of changes in azimuth and apparent velocity with frequency and time and does not include corrections to find the source location on the fault.

Depth of Rock Damage from Strong Seismic Ground Motions near the 2004 Parkfield Mainshock

Changes in the S - P delay of repeating earthquakes near Parkfield, California, after strong shaking in the 2004 mainshock, occur at surface stations but not at borehole stations. This result indicates that rock damage occurs mainly in the upper few tens of meters in fractured rock with a low seismic velocity. In addition, changes in coda-primary delay are comparable to changes in S - P delay. This observation along with the lack of S - P delay changes at borehole stations yields a simple model for the coda. Direct waves pass through the shallow damaged layer once. Coda waves scatter or refract in the deep undamaged subsurface and hence also pass through the subsurface just once. The coda, however, contains some brief reverberations within the shallow subsurface where the additional path length scales to the thickness of the shallow layer. The coda-primary delay measured by correlation includes both the zero delay change of direct coda and the increased delay of reverberating coda. Published changes in Rayleigh-wave group velocity before and after the Parkfield mainshock are compatible with their main cause being from the shallow velocity changes inferred from repeating earthquakes but have little spatial resolution. Stacking of earthquake seismograms resolved surface reflections at one shallow (63 m deep) borehole station GHIB. A change in P -wave travel-time body of ∼8 msec was marginally resolved and is compatible with the repeating earthquake S - P delay changes. Autocorrelation passive seismology resolved ∼32 msec P -wave reverberation delay change above the 251 m deep borehole station CCRB.

Shear wave anisotropy in the crust around the San Andreas fault near Parkfield: spatial and temporal analysis

SUMMARY We systematically analysed shear wave splitting (SWS) for seismic data observed at a temporary array and two permanent networks around the San Andreas Fault (SAF) Observatory at Depth. The purpose was to investigate the spatial distribution of crustal shear wave anisotropy around the SAF in this segment and its temporal behaviour in relation to the occurrence of the 2004 Parkfield M 6.0 earthquake. The dense coverage of the networks, the accurate locations of earthquakes and the high-resolution velocity model provide a unique opportunity to investigate anisotropy in detail around the SAF zone. The results show that the primary fast polarization directions (PDs) in the region including the SAF zone and the northeast side of the fault are NW–SE, nearly parallel or subparallel to the SAF strike. Some measurements on the southwest side of the fault are oriented to the NNE–SSW direction, approximately parallel to the direction of local maximum horizontal compressive stress. There are also a few areas in which the observed fast PDs do not fit into this general pattern. The strong spatial variations in both the measured fast PDs and time delays reveal the extreme complexity of shear wave anisotropy in the area. The top 2–3 km of the crust appears to contribute the most to the observed time delays; however substantial anisotropy could extend to as deep as 7–8 km in the region. The average time delay in the region is about 0.06 s. We also analysed temporal patterns of SWS parameters in a nearly 4-yr period around the 2004 Parkfield main shock based on similar events. The results show that there are no appreciable precursory, coseismic, or post-seismic temporal changes of SWS in a region near the rupture of an M 6.0 earthquake, about 15 km away from its epicentre.

Kinematic Inversion of the 2004 M 6.0 Parkfield Earthquake Including an Approximation to Site Effects

The 2004 M 6.0 Parkfield earthquake yielded one of the largest amounts of near-source strong ground motion seismic data ever. We invert strong-motion seismograms to obtain a model for the space-time distribution of coseismic slip on the fault. To reduce noise in the inversion, we take into account local amplifications that affect each station by using records of the 1983 M 6.5 Coalinga earthquake. Site amplification correlates well with large peak ground velocities registered during the 2004 Parkfield mainshock. The inversion for a kinematic rupture model yields a nonunique solution; we therefore analyze various rupture models that explain the data equally well. Our preferred rupture model identifies a primary zone of high slip surrounding the hypocenter, where the maximum slip is 57 cm. A secondary slip area, over which contours are not well resolved, is located northwest of the hypo- center. The rupture speed is highly heterogeneous. We infer an average rupture ve- locity of 2.8 km/sec close to the hypocenter, and of 3.3 km/sec in the secondary region of large slip to the northwest of the hypocenter. By correlation of our rupture model with both microseismicity and velocity structure, we identify six patches on the fault plane that behave in seismically distinct ways. Online material: Kinematic rupture model parameters.

Updating seismic hazard at Parkfield

The occurrence of the September 28, 2004 M w = 6.0 mainshock at Parkfield, California, has significantly increased the mean and aperiodicity of the series of time intervals between mainshocks in this segment of the San Andreas fault. We use five different statistical distributions as renewal models to fit this new series and to estimate the time-dependent probability of the next Parkfield mainshock. Three of these distributions (lognormal, gamma and Weibull) are frequently used in reliability and time-to-failure problems. The other two come from physically-based models of earthquake recurrence (the Brownian Passage Time Model and the Minimalist Model). The differences resulting from these five renewal models are emphasized.

Coseismic Strain and the Transition to Surface Afterslip Recorded by Creepmeters near the 2004 Parkfield Epicenter

The Mw 6 Parkfield earthquake at 11:15 local time, 28 September 2004 has dealt a possibly fatal blow to the notion that significant surface slip precedes earthquakes. In 1966, anecdotal reports of surface fissures along the fault zone had been reported in the preceding week, and a water pipe fractured 11.6 hours before the mainshock. This offered hope that substantial surface slip may have occurred had displacement sensors been in place to measure it. In the days, hours, and minutes prior to the 2004 Parkfield mainshock, displacement sensors were in place, and nothing happened. According to data from eight creepmeters crossing the fault, the fault remained effectively locked at the detection level of each sensor (6-20 μm) (Langbein et al. , 2005, this issue). This absence of surface slip was accompanied by an absence of significant strain on the borehole dilatometer and shear-strainmeter array in the region, and by the absence of any unusual displacements on the GPS array embracing the 2004 epicenter. Although preseismic strain signals seen on some borehole strainmeters suggest that epicentral strain was not completely indifferent to the pending earthquake, the signals are close to instrumental noise levels and pose new challenges for their reliable future detection. In contrast to the absence of preseismic fault displacement, the creepmeters all recorded an abrupt coseismic dextral offset at the time of the mainshock. The step corresponds to dextral shear of the fault zone, but in the absence of surface rupture, the creepmeters during early afterslip acted as extensometers, measuring strain rather than fault creep. The serendipitous location of creepmeters at Work Ranch both across and eccentric to the fault zone has provided insight into the development of surface faulting. The u.s. Geological Survey (USGS) has maintained a creepmeter array in Parkfield since the 1966 earthquake, increasing …

Mapping the rupture process of moderate earthquakes by inverting accelerograms

We present a waveform inversion method that uses recordings of small events as Green's functions to map the rupture growth of moderate earthquakes. The method fits P and S waveforms from many stations simultaneously in an iterative procedure to estimate the subevent rupture time and amplitude relative to the Green's function event. We invert the accelerograms written by two moderate Parkfield earthquakes using smaller events as Green's functions. The first earthquake (M = 4.6) occurred on November 14, 1993, at a depth of 11 km under Middle Mountain, in the assumed preparation zone for the next Parkfield main shock. The second earthquake (M = 4.7) occurred on December 20, 1994, some 6 km to the southeast, at a depth of 9 km on a section of the San Andreas fault with no previous microseismicity and little inferred coseismic slip in the 1966 Parkfield earthquake. The inversion results are strikingly different for the two events. The average stress release in the 1993 event was 50 bars, distributed over a geometrically complex area of 0.9 km2. The average stress release in the 1994 event was only 6 bars, distributed over a roughly elliptical area of 20 km2. The ruptures of both events appear to grow spasmodically into relatively complex shapes: the inversion only constrains the ruptures to grow more slowly than the S wave velocity but does not use smoothness constraints.

Seismicity alert probabilities at Parkfield, California, revisited

AbstractFor a decade, the U.S. Geological Survey has used the Parkfield Earthquake Prediction Experiment scenario document to estimate the probability that earthquakes observed on the San Andreas fault near Parkfield will turn out to be foreshocks followed by the expected magnitude 6 mainshocks. During this time, we have learned much about the seismogenic process at Parkfield, about the long-term probability of the Parkfield mainshock, and about the estimation of these types of probabilities. The probabilities for potential foreshocks at Parkfield are reexamined and revised in light of these advances. As part of this process, we have confirmed both the rate of foreshocks before strike-slip earthquakes in the San Andreas physiographic province and the uniform distribution of foreshocks with magnitude proposed by earlier studies. Compared to the earlier assessment, these new estimates of the long-term probability of the Parkfield mainshock are lower, our estimate of the rate of background seismicity is higher, and we find that the assumption that foreshocks at Parkfield occur in a unique way is not statistically significant at the 95% confidence level. While the exact numbers vary depending on the assumptions that are made, the new alert probabilities are lower than previously estimated. Considering the various assumptions and the statistical uncertainties in the input parameters, we also compute a plausible range for the probabilities. The range is large, partly due to the extra knowledge that exists for the Parkfield segment, making us question the usefulness of these numbers.

Fault model for preseismic deformation at Parkfield, California

We construct an earthquake instability model to estimate precursory faulting and ground deformation before the next moderate (M 5.5–6) earthquake on the San Andreas fault near Parkfield, California. The quasi‐static model simulates fault slip, fault shear stress, and ground deformation for all stages of repeated earthquake cycles. Unstable slip, which is the analog of a moderate Parkfield mainshock, is caused by overall failure of a postulated strong area or patch of the fault zone. The brittle patch and surrounding weaker viscous fault are defined by the sign of a spatially varying coefficient of a slip rate dependent fault law. Stable failure of the patch leading up to mainshock failure causes slip rate to increase at depth. Amplitudes and timescales of the resulting preseismic deformation anomalies are found for the current earthquake cycle at Parkfield, starting just after the 1966 mainshock, in two ways. In the first way we vary parameter values in a sequence of simulations so as to find the simulation giving best agreement with locations, moments, and recurrence times of past moderate earthquakes and also with fault creep and trilateration line length changes since the 1966 mainshock. In the second way we assign values of most parameters from laboratory friction experiments and a temperature‐depth profile inferred from heat flow data. Overall agreement between model results and observations is comparable for the two methods. In all simulations, preseismic deformation anomalies arise from slip rate increase near the model earthquake focus and start several years or less before instability. Preseismic anomalies in surface fault creep and line length are always too small to detect in current measurements. However, for some simulations constrained mainly by field data the predicted anomalies in borehole dilatation are large enough to detect.

Seismic quiescence: a test of the hypothesis and a precursor to the next Parkfield, California, earthquake

SUMMARY The Parkfield segment of the San Andreas fault shows seismic quiescence for M > 2.0 earthquakes since early to mid-1986. The rate decreases by 45 to 70 per cent depending on the magnitude band. We interpret this change as real because it is concentrated in the larger magnitudes and cannot be explained by a reasonable magnitude shift. The rate decrease is present in the Parkfield segment of the fault between 35.6 and 36.1°N, and no other rate change as significant has been observed during the time when high quality data were available (since 1975). We interpret this observation as a seismic quiescence precursor to the next Parkfield mainshock. Although information on the expected duration of precursory quiescence is weak, we estimate that the Parkfield earthquake should occur in the interval 1990 February to 1992 February. Precursory seismic quiescence, which lasted approximately 1.5 yr, was found in, and possibly around, the source volume of an earthquake withM= 3.6 which occurred on 1986 August 29 in the Parkfield area. This observation demonstrates that the quiescence hypothesis is applicable to the Parkfield segment of the San Andreas fault. The false alarm rate was estimated in a detailed analysis as a function of alarm level for the fault segments surrounding the M= 5.0 Stone Canyon (1982 August) and M= 3.6 Parkfield (1986 August) mainshocks. For these two events, the duration and statistical significance of false alarms as a function of space, time and magnitude band was determined. We propose that in this tectonic environment, and with the data characteristics of these earthquake catalogues, attempts to predict mainshocks based on the quiescence hypothesis may be successful and generate one to two false alarms covering about 10 per cent of the time-space dimensions of the catalogue.

Changes of mean magnitude of Parkfield seismicity: A part of the precursory process?

The average magnitude of earthquakes produced along the Parkfield segment of the San Andreas fault was decreased significantly between late 1986 and late 1988 along with the seismic quiescence [Wyss et al., 1990a]. The change corresponds to an increase of the b‐value from 1.1 to 1.4. During the last year (1989) however, the average magnitude produced was approximately normal again. This same phenomenon had been identified before several other mainshocks. Therefore it may be interpreted as a precursor to the predicted Parkfield mainshock.

Quarry blast sources and earthquake prediction: The Parkfield, California, earthquake of June 28, 1966

Quarry blasts provide an excellent source of waves with which to study temporal variations of seismic travel times. Using sources from several active quarries we found no time changes at Gold Hill (GHC), California from 7 months before the 1966 Parkfield mainshock to at least 13 months following the event. GHC is within 0.5 km of the zone of ground breakage associated with the earthquake. This event was one of the largest to occur in central California since 1906. The negative result of our experiment does not rule out the existence of an anomaly prior to the event, rather it emphasizes the difficulty of detecting such anomalies without a fortuitous combination of sources and stations. The routine monitoring of temporal changes in seismic velocity along the San Andreas fault may require not only dense networks of high quality seismic stations, but also sources whose spatial location and origin time can be controlled.