The following contributions by this project help to accomplish SCEC's mission:
It identifies seismogenic faults from seismicity and specifies their kinematic properties.
It formulates and tests models of stress changes and fault-earthquake interaction.
This project is pertinent to the following working groups (underline=closer affinity):
Group A, because we investigate the interaction between stress and earthquakes and provide physical understanding leading to estimates of the effect of earthquakes on earthquake probability.
Group C, because we map seismogenic faults, their location and kinematics, from seismicity and we relate these faults to structure inferred from geology.
Group D, because we construct a data set of focal mechanisms; we investigate the relation between earthquakes and structure in specific settings and in general; and we calculate stress parameters and stress changes from the seismicity.
Group G, because we investigate space-time clustering of small earthquakes in relation to the occurrence of large and intermediate-magnitude earthquakes and we offer data to test models.
Earthquakes cause and respond to stress change. But, while large and small earthquakes change stress by vastly different amounts, they may nucleate in response to similar mechanical parameters. Thus, the response of regional seismicity to the stress change induced by large earthquakes may provide information about several key issues concerning seismogenesis at both small and large magnitudes, with the advantage of a very large data sample. The role of earthquake-induced stress changes in controlling future seismicity, the value of friction, the nature of nucleation processes (i.e., the constitutive properties of faults), the poroelastic response of fault zones, and the dependency of these properties on depth, lithology, magnitude, etc. are within reach of this research avenue.
We have produced a data set of hypocenters with focal mechanisms for southern and central California. We have also interpreted them for "slip planes", the nodal planes associated with the ruptures. We use slip planes to test recent hypotheses on changes in stress caused by large ruptures and their effects on the nucleation of future earthquakes. Initial results from our ongoing study of the 1992 Landers sequence include:
1. Regional seismicity (>7.5km from mainshock rupture) is profoundly affected by the coulomb stress change (DCS) from the 1992 mainshock, down to a level of about 0.1bars, or 100km from the rupture (Fig.1);
2. We have modeled the 1992 mainshock rupture from off-rupture seismicity and we obtained a slip distribution very similar to Wald and Heaton (1994) (Fig.3). This suggests that the coulomb failure criterion is applicable and our slip-plane data are reliable.
3. The optimum friction coefficient for the deeper half of the seismicity is 0.8-0.9. The optimum friction value for shallow seismicity is generally somewhat lower;
4. The 1992 DCS effect on seismicity is strongest immediately after the mainshock, but it is still present 3.5y after the level of off-fault seismicity is back to pre-1992 levels (Fig. 2);
5. Our choice of "slip planes" is much better than random (Fig.2), but obvious interpretation errors can be identified. We have not corrected them; but we are seeking a systematic and non-circular approach to use stress-response in selecting slip planes.
Data and Method.
Our stress-interaction research makes use of the "slip planes" which we have interpreted from quality-controlled locations and focal mechanisms in this SCEC/NEHRP project. About 25,000 slip planes are available from 1980 to 1996 (these data and the visualization software QKVIEW are available in the SCEC Data Center). We use slip planes as seven-dimensional elements of brittle shear failure (lat., lon., depth, strike, dip, rake, and time), and we calculate the change in coulomb stress for each of these elements. These slip planes are considered negligible in terms of agent of stress change; they are manifestations of change caused by other factors.
We consider seismicity in southern California that may be affected by stress change induced by Landers (DCS), but we exclude earthquakes in the immediate vicinity (<7.5km) of the rupture, where non-elastic effects may also be present (Figure 1). We categorize each slip-plane as being encouraged (E) or discouraged (D) by DCS. Stress change from this rupture is small compared to the pre-existing stress and it affects the timing of the event, but not the kinematics of the slip plane, except in the immediate vicinity of the fault (e.g., King et al, 1994). Thus the pertinent shear stress component is in the slip direction. Figure 1 compares the spatial distribution of E's and D's before and after the Landers mainshocks (Joshua, Landers, Big Bear). Figure 2 show the temporal distribution of E's, D's and E/D.
Results.
DCS has a dramatic effect on the seismicity. While the average rate of D's in the post-Landers period (4.5 years in Figure 1) changes little, the average rate of E's raises by a factor of 4. Although the overall seismicity rate is higher by a factor of 2 after than before Landers, areas where seismicity has been turned off can be identified in Figure 1 (after) on the compressive quadrants of the mainshock right-lateral rupture (northwest and southeast corners). The prevalence of E's over D's is even more dramatic immediately after the mainshock, thereafter diminishing gradually. The ratio E/D just after Landers is higher than before Landers by an order of magnitude in Figure 2 (thick line). This increase is caused by an increase in E's (dashed), but also a decrease in D's (dotted). After Landers, E/D decreases gradually, but has not yet recovered to pre-Landers level. This persistency may in part reflect discouraged earthquakes that have been delayed more than 4.5y.
Figure 2 shows the response of seismicity in the far field only (>27km). The E/D behavior in the near-field is more complex than in the far field. Some of this complexity can be ascribed to intense swarms (near Yucaipa and northeast of Big Bear Lake) that may reflect significant local strain violating the assumption that all the DCS in from Landers mainshocks. The tendency for E/D to remain high in the post-Landers period may be ascribed to sufficiently large DCS to hold many of the discouraged "earthquakes" away from failure. Erroneously picked nodal planes probably contribute to some of the complexity, as well. But, the post-Landers increase in E/D in Figure 2 is much less for nodal planes opposite to the ones chosen in the tectonic interpretation (thin). Thus, our slip-plane interpretation is much better than random. Finally, the E/D effect is clearly detected for earthquakes in the 0.1<DCS<0.3bars range, but not in the DCS<0.1bars range. Similar thresholds were found in other studies (e.g., Simpson and Reasenberg, 1997; Hardebeck et al., 1997).
We modeled the Landers mainshock rupture from the seismicity following the rupture. Our model is compared to the model proposed by Wald and Heaton (1994) in Figure 3. The fit is remarkably good, considering that it is derived with the simplistic assumption of maximizing the E's and minimizing the D's. Note the contrast with the result from seismicity before Landers. The result is rather independent of initial conditions (uniform slip in Figure 3) and several iteration procedures. We also tried to model the rupture with three successive data sets of about 1000 earthquakes each. The match to Wald and Heaton (1994) became progressively worse, but most of the misfit occurred at the northern end of the rupture, where substantial post-slip is supposed to have occurred. The reasonable fit obtained from data after the first post-Landers year suggests that the late effect in the seismicity is still primarily from the coseismic DCS, and not from secondary deformation (e.g., viscoelastic relaxation or post-slip on the rupture).
Recent and Expected Products and Publications
Armbruster and Seeber 1996, updated files of focal mechanisms and slip-planes for s. California as well as a version of QKVIEW to visualize these data in SCEC data base.
Seeber L. and J.G. Armbruster, The San Andreas fault system through the Transverse Ranges as illuminated by earthquakes, J. of Geoph. Res., 100, pp. 8285-8310, 1995.
Seeber, L. and Sorlien, C.C., Listric Thrusts in the western Transverse Ranges, Resubmitted to GSA Bull., 1997.
Geiser, P. A. and L. Seeber, Three-dimensional seismo-tectonic imaging in the California Transverse Ranges, submitted to JGR
Manuscripts in Preparation:
"Earthquakes and fault kinematics in the central Transverse Ranges" by Seeber and Armbruster;
"The California Transverse Ranges as a Compressional Wedge" by Geiser and Seeber;
"The kinematics of small 'background' earthquakes and mesoscopic-scale faults: similar and distinct from the kinematics of large faults", by F. Ghisetti and L. Seeber.
"Translation, rotation, and extension in the Salton Trough: how California is falling into the ocean" Seeber and Armbruster.
