Periodic Pulsing of the San Andreas Fault: An Update

Robert M. Nadeau

Figure 5.1: (Left) Along fault depth section of background seismicity (gray points) and SAF CS locations (black circles) for an 80 km segment of the central SAF. Rupture zone of the 1989, M7.1 Loma Prieta earthquake is labeled LP. (Right, center) Profile of the 1984-2005.25 deep slip history for the segment from the CS data. Rates (in color) are in percent difference from the 1984-2005.25 average. Color intensity give 95% confidence bounds. Open circles are along fault locations and times of M $>$ 3.5 earthquakes. Their sizes are keyed to relative magnitudes, but are significantly larger than their actual rupture dimensions. Vertical gray line is start time of the extended analysis. (Right, bottom) Deep slip rates history for a representative 15 km sub-segment showing the 7 pulses (labeled 1 through 7) observed to date. (Right, top) Occurrence times and magnitudes of the M $>$ 3.5 earthquakes in the 80 km zone. Time of Loma Prieta and San Juan Bautista events are labeled at top (LP and SJB, respectively). Horizontal green bars are pulse onset periods before 2000. Horizontal red bars are predicted pulse onset periods based on the green bars. Gray extensions of the red bars are prediction uncertainties.
\epsfig{file=nadeau06_2_1.eps, width=9cm}\end{center}\end{figure*}

Slip Rates from Micro-quakes

A characteristically repeating micro-earthquake sequence (CS) is a sequence of small earthquakes (M $\sim<$ 3.5) whose seismograms, locations and magnitudes are nearly identical. Each earthquake in the sequence represents a repeated rupture of the same patch of fault, and the times between the ruptures (i.e., their recurrence intervals) are, in general, inversely proportional to the average tectonic loading rate on the fault (Nadeau and McEvilly, 1999). These unique properties allow CSs to be used to infer slip rates deep within fault zones, and they have been used to particular advantage for this purpose in regions where geodetic measurements are limited in their spatial and temporal coverage.

Repeating Quake Analysis along the Central SAF

Along much of the 175 km stretch of the San Andreas Fault (SAF) separating the rupture zones of California's two great earthquakes (i.e., the $\sim $ M8 1906 San Francisco and 1857 Fort Tejon events), geodetic measurements have been done relatively infrequently in campaign mode. Along this stretch, however, over 500 CSs have been identified with events occurring between 1984 and 1999 (inclusive), and analysis of these sequences reveal: 1) that the recurrence intervals within any given CS vary significantly through time, 2) that among different CS on a localized fault segment, the recurrence variations are coherent through time, and 3) that in many cases, the coherent variations recurred quasi-periodically (Nadeau and McEvilly, 2004).

Here we report results based on an analysis of the CSs that extend the period studied by (Nadeau and McEvilly, 2004) by an additional 5 years. A primary objective of this extended analysis is to see if systematics in time-varying patterns reported in the 2004 study (i.e., the periodic pulsing of deep slip and its correlation to variations in rates of larger earthquakes) are ongoing phenomena.

Correlation with Larger Earthquakes

Recurrence variation information for the CSs was used to construct a profile of deep fault slip rate histories along the 175 km study zone for the 1984-2005.25 study period. The profile reveals that along the northwestern-most 80 km segment of the study zone (Figure 5.1), deep fault slip rates commonly vary by over 100% and their variation patterns (i.e., pulse patterns) recur with a periodicity of $\sim $ 3 years. Shown at the right in Figure 5.1 is a comparison of this large-scale periodic deep slip pattern with the occurrence times of M3.5 to M7.1 earthquakes (i.e., magnitudes larger than those of the CS events) and with the occurrence times of three known slow slip events in the area (Gwyther et al., 2000). The comparison reveals a significant correlation between the onset periods of the repeating deep slip signals and the occurrence rates of the larger events. Nadeau and McEvilly (2004) observed this pattern for CS data during the period 1984-1999. With the additional 5+ years of CS data, the pattern and correlation observed by Nadeau and McEvilly appears to be continuing (Figure 5.1, right).

Excluding Loma Prieta aftershocks, 66 earthquakes with M $>$ 3.5 occurred in region during the 1984-1999 study period, and a general correlation is also observed between the occurrence times of these events and the 1-year onset periods of the pulses (i.e. the time interval where pulse slip velocities transition from low to high values). Forty-two of the 66 events were found to occur during the onset periods, this represents an occurrence rate that is 3.7 times larger than the rate observed during the non-onset periods. When Loma Prieta aftershocks are included into the analysis the onset period rate increased to 4.6 times that of the non-onset period rate.

To the resolution of the characteristic microearthquake slip rate data, the M7.1 Loma Prieta mainshock also occurred coincident with the onset of the P2 timed pulse. The times of the next two largest non-aftershock events in the area and study period (i.e., M5.4 San Juan Bautista mainshock in 1998 and a M4.7 event in 1986) are also coincident with the onset of pulses P5 and P1, respectively, and the P3, P4, and P5 pulse onsets correspond closely to the times of the three slow slip events in the area whose aseismic moment magnitudes were estimated to be $\sim $ 5$M_{w}$ (Gwyther et al., 2000).


Earthquake triggering induced by velocity weakening effects (Dietrich, 1986) associated with increasing fault slip velocities during pulse onsets may provide an explanation for the corresponding rate increases of the larger earthquakes. It is also possible that increased quake rates occur quasi-periodically due to some other mechanism, such as the accelerated rate of loading to failure during onsets of the quasi-periodic deep slip pulses.

Regardless of the causes, the success of predictions of time periods with increased likelihood for larger quakes during the additional 5+ year period, based on the pattern of pulsing that occurred prior to the end of 1999 (Figure 5.1), suggests that significant potential for refinement of time-dependent earthquake forecasts models exists for this area (WGCEP, 1999) on time scales comparable to the average pulse cycle duration of $\sim $ 3 years.


Thanks are given to Roland Bürgmann and Mark H. Murray for stimulating conversations and insightful comments regarding this work. This research was supported by the National Science Foundation through award EAR-0337308.


Dieterich, J.H., A model for the nucleation of earthquake slip, in Earthquake Source Mechanics, S. Das, J. Boatwright, and C.H. Scholz (Editors), American Geophysical Monograph 37, American Geophysical Union, Washington, D.C., 37-47, 1986.

Gwyther, R.L., C.H. Thurber, M.T. Gladwin, and M. Mee, Seismic and Aseismic Observations of the 12th August 1998 San Juan Bautista, California, M5.3 Earthquake, in Proceedings of the 3rd Conference on tectonic problems of the San Andreas Fault system, G. Bokelmann and R. Kovach, Eds., Stanford California School of Earth Sciences, Stanford University, 2000.

Nadeau, R.M., and T.V. McEvilly, Fault slip rates at depth from recurrence intervals of repeating microearthquakes. Science, 285, 718-721, 1999.

Nadeau, R. M. and T. V. McEvilly, Periodic Pulsing of Characteristic Microearthquakes on the San Andreas Fault, Science, 303, 220-222, 2004.

Working Group on California earthquake Probabilities (WGCEP). Earthquake probabilities in the San Francisco Bay Region: 2000 to 2030-a summary of findings, U.S. Geol. Surv. Open-File Rept. 99-517, 36 pp., 1999.

Berkeley Seismological Laboratory
215 McCone Hall, UC Berkeley, Berkeley, CA 94720-4760
Questions or comments? Send e-mail:
© 2006, The Regents of the University of California