INVESTIGATIONS
This project consists of two tasks that address modeling of stress changes associated with past earthquakes in southern California. The first task is a quantitative investigation of how well mainshock-induced static stress change explains the triggering of aftershock sequences. The second task addresses stress evolution history and fault interaction in southern California.
RESULTS
Static stress change triggering of earthquakes (Jeanne Hardebeck)
Static stress change triggering of earthquakes has been proposed as a model for evaluating short-term earthquake hazards. We have quantitatively evaluated how well this model explains the apparent triggering of aftershocks by the 1992 Landers and 1994 Northridge earthquakes. Specifically, we have tested whether the fraction of aftershocks consistent with triggering by mainshock-induced static stress changes is larger than the fraction of random events which would appear consistent by chance.
The static stress change triggering model works well for the Landers sequence, especially for events within one fault length of the mainshock. We find that 85% of event between 5 and 75 km distance from the mainshock fault plane (Figure 1a), or with average static stress changes between 0.01 and 1 MPa (Figure 1b), are consistent with static stress change triggering, compared to approximately 50% of random events. The minimum distance is probably controlled by limitations of the modeling, while the maximum distance may be because static stress changes of less than 0.01 MPa trigger too few events to be detected with our datasets. The static stress changes appear to influence the regional seismicity for at least 4.5 years.
The static stress change triggering model, however, can't explain the first month of the Northridge aftershock sequence, or any subset we tested, significantly better than it explains a set of random events (Figures 1c and 1d.)
The difference in the performance of the static stress change triggering model between the Landers and Northridge sequences implies that it may be a useful model only in zones of weak crust and low shear stresses, such as the Landers area (Hauksson, 1994). Presumably, this is because the small static stress changes are a more significant fraction of the failure stress of weak faults.
The poor performance of the static stress triggering model for the Northridge sequence and presence of many Landers aftershocks not consistent with static stress change triggering imply that other triggering mechanisms are also involved.
Our results suggest that the static stress triggering model has some validity and can be useful in explaining apparently triggered seismicity on weak faults. The variability of the usefulness of the model for areas with different stress states and fault strengths indicates that these factors should be incorporated into the model when used for seismic hazard assessment.
A paper describing this study has been submitted to the Journal of Geophysical Research and is currently in the revision process. Preliminary results were also presented at the March, 1997, SCEC workshop on "Earthquake Stress Triggers, Stress Shadows, and their Impact on Seismic Hazard."
Cumulative Coulomb failure function (Jishu Deng)
We continued the study of Deng and Sykes (1997a) by calculating the cumulative Coulomb failure function, CFF, for southern California as a function of time with respect to an arbitrary zero baseline in 1812 (Deng and Sykes, 1997b). We take into account tectonic stress loading associated with 98 fault segments as well as coseismic stress changes associated with 36 earthquakes of M > 6. We then examine the distribution of events of M > 1.8 in terms of their occurrence in space and time with respect to our calculations of the evolving stress field. Our calculations show that between 1933 and the present, more than 85% of the M> 5 earthquakes occurred in regions of positive CFF. Most other M > 5 events occurred very close to the calculated boundaries between stress shadow and stress-enhanced zones. The locations of several of those boundaries are very sensitive to the slip distributions of the older large to great earthquakes in the 19th century. Only one event of M > 5, the 1993 Wheeler Ridge earthquake, occurred in the middle of what we calculate to be a large shadow zone, in that case the one created by the 1952 Kern County earthquake.
The locations of small- (3.0 < M < 5) and micro-size (M < 3) earthquakes are also well-constrained by the stress model. Our result shows that from 1981 until just before Landers earthquake sequence of 1992 more than 85% of the examined 2497 right-lateral strike-slip earthquakes of M > 1.8 occurred in stress-enhanced zones (Figure 2). The ratio of encouraged to all small events reaches a high value of about 88% if the apparent coefficient of friction m is between 0.0 and 0.6. A significant number of the other small- and micro-size shocks occurred close to the stress boundary, which again is sensitive to the length of rupture and displacements in the 1812 Wrightwood earthquake. The highest percentage of events occurred at locations where stress is about 1 MPa above the 1812 baseline. While the size of the peak stress, 1 MPa, may be related to the timescale of almost 200 years we used for our calculations, it is comparable to many of the stress drops calculated from the spectra of seismic waveforms. Our calculations also show that most events occurred in areas of CFF between 0.0 and 2.0 MPa, independent of magnitude. This result indicates that the stress on a fault can vary up to a maximum of about 2 MPa during an earthquake cycle.
The above calculations indicate that current moderate-, small-, and micro-size earthquakes, even shocks as small as magnitude 1.8, are related to coseismic stress changes in large to great historic events, even those that occurred 100 to 200 years ago. The stress shadows from those major earthquakes have not been completely restored in all areas by tectonic stress reaccumulation. The distributions and mechanisms of more modern earthquakes might be combined with paleoseismological techniques to better constrain or to invert for the slip distribution of significant older earthquakes.
The fact that so many of the earthquakes that we studied occurred in regions of calculated positive values of CFF indicates that many future moderate, large, and great earthquakes will occur in stress-enhanced zones. In the future, detailed studies focused on those regions where stress is above or close to 1 MPa might help to better constrain the locations of future earthquakes. It is also surprising that a simple half-space model is so successful in predicting regions of earthquake occurrence in a tensorial sense. Clearly, viscoelastic effects do occur on timescales of years to centuries and will have to be modeled in future work.
References
Seeber, L., and J. G. Armbruster, The San Andreas fault system through the Transverse Ranges as illuminated by earthquakes, J. Geophys. Res., 100, 8285-8310, 1995.
Hauksson, E., State of stress from focal mechanisms before and after the 1992 Landers Earthquake Sequence, Bull. Seismol. Soc. Amer., 84, 917-934, 1994.
Reports
Deng, J., and L. R. Sykes, Evolution of the stress field in southern California and triggering of moderate-size earthquakes: A 200-year perspective, J. Geophys. Res., 102, 9859-9886, 1997a.
Deng, J., and L. R. Sykes, Stress evolution in southern California and triggering of moderate-, small-, and micro-size earthquakes, J. Geophys. Res., 102, 24,411-24,435, 1997b.
Hardebeck, J. and J. J. Norris, A Quantitative Investigation of the Static Stress Change Model of Earthquake Triggering for Aftershock Sequences, SCEC Annual Meeting Program with Abstracts, 51, 1996.
Hardebeck, J. L. and J. J. Norris, A Quantitative Investigation of the Static Stress Change Model of Earthquake Triggering for Aftershock Sequences, Eos Trans. AGU 77, Fall Meet. Suppl., F482, 1996.
Hardebeck, J. L. and E. Hauksson, Static Stress Drop in the 1994 Northridge Aftershock Sequence, Bull. Seism. Soc. Am., in press, 1997.
Hardebeck, J. L., E. Hauksson and J. J. Norris, Quantitative observations of static stress change triggering in two Southern California aftershock sequences, J. Geophys. Res., in review, 1997.
Hardebeck, J. L. and E. Hauksson, Static Stress Drop in the 1994 Northridge Aftershock Sequence, SCEC Annual Meeting Program with Abstracts, 65, 1997.
Figure Captions
Figure 1. The percent of events consistent with static stress change triggering (the Coulomb Index) versus distance from the mainshock and magnitude of static stress change. The asterisks indicate the Coulomb Index of the first month of the observed aftershock sequences. The vertical error bars are the 2s error estimates and the horizontal error bars indicate the bins. The horizontal dotted and dashed lines represent the mean Coulomb Index and 95th percentile, respectively, of a set of random synthetic sequences. If the Coulomb Index of the observed sequence is above the dashed line, we can conclude with 95% certainty that the static stress change triggering model explains the aftershocks better than it can a random set of events. (a) Coulomb Index versus distance to the nearest point on the fault plane, Landers. (b) Coulomb Index versus magnitude of static stress change, Landers. (c) Coulomb Index versus distance, Northridge. (d) Coulomb Index versus stress change, Northridge.
Figure 2. CFF just before the 1980 Victoria, Baja California, earthquake of Mw 6.4 for vertical right-lateral strike-slip faults trending 321o. Focal mechanisms for earthquakes of 1.8 < M < 3 between 1981 and just before the Landers earthquake of 1992 involving a right-lateral slip vector within 25o from one striking 321o and plunging 0o are selected from the catalog compiled by Seeber and Armbruster (1995) and superimposed on map of CFF. Note that most of the events are located in regions of positive CFF.
