Aftershocks and Fault-Zone Properties

Carl Kisslinger    Cooperative Institute for Research in Environmental Sciences and Department of Geological Sciences, University of Colorado at Boulder, Boulder, Colorado 80309-0216

1 INTRODUCTION

Sequences of aftershocks are among the most frequently observed effects of moderate or stronger earthquakes. The spatial and temporal distributions of the events in an aftershock sequence contain information about the aftershock generating process in particular, and by extrapolation, the earthquake-generating process in general. The research to date, which is reviewed here, indicates that these distributions—which are linked closely to the geometry of the fault surface that ruptured to produce the mainshock—depend on the physical properties of the fault zone as well as the ambient conditions, especially the distributions of strength and stress and the temperature. In addition to the scientific interest in aftershock behavior, there are practical applications because strong aftershocks are a significant additional hazard associated with damaging earthquakes.

The Landers California earthquake sequence of 1992 provides a good example of the value of well-located aftershocks for delineating the fault that ruptured and for providing data on the earthquake-generating process in general. The entire June 1994 issue of the Bulletin of the Seismological Society of America (Volume 84, Number 3) is devoted to the analysis and interpretation of the copious field and instrumental data gathered for this major episode. The sequence was marked by three main events: a magnitude Mw-6.1 foreshock on April 23 (the Joshua Tree earthquake), which had a well-developed, slowly decaying aftershock sequence; the Mw-7.3 mainshock on June 28 (the Landers earthquake); and a Mw-6.4 aftershock on June 28 (the Big Bear earthquake) (Jones, 1994). All of the earthquakes in the region located by the Southern California Seismic Network between June 28 and December 31, 1992 are mapped in Fig. 1 (Jones, 1994). The Landers aftershocks and the aftershocks of the strong Big Bear aftershock delineate the faults that slipped in this episode. A gap in activity separates the principal aftershock lineation, approximately 100 km long, from the aftershocks associated with the Calico–Blackwater fault zone to the north, which also experience a small amount of slip in the mainshock.

The Northridge (California) earthquake, Mw-6.7 (Hauksson et al., 1994), occurred on January 17, 1994. This earthquake provides an outstanding example of the contributions to understanding of seismogenesis that can be derived from the data acquired and carefully analyzed for an aftershock sequence that occurred on a buried fault system with no surface expression, but that was recorded by a dense network of modern seismographs. The aftershocks served to identify which of the two planes from the instrumental focal mechanism solution is the fault plane: a plane striking N60°W, dipping to the south at about 40°. Because this earthquake occurred shortly before the writing of this chapter, the abundant results of the scientific analysis are available only in unpublished abstracts and manuscripts, and much work remains to be done. The region around this earthquake has been monitored by a variety of geophysical methods over a long period of time, and because of the implications of the Northridge earthquake for the assessment of earthquake hazards in a densely populated, highly industrialized urban area, this earthquake and its aftershocks should be among the most thoroughly studied events in the history of seismology.

2 THE PHENOMENOLOGY OF AFTERSHOCKS

2.1 Definitions and Basic Properties

It is easier to offer a working definition of an aftershock than it is to identify a particular event as being one. Frohlich (1989) points out that the working definitions differ among investigators. In this work, we define an aftershock as a secondary earthquake following a stronger primary one (the mainshock) whose location and time of occurrence are a direct result of the occurrence of the mainshock. This definition admits the possibility that an aftershock might be an earthquake that would have happened in the same place at a later time, without the occurrence of a mainshock. Procedures for identifying aftershocks are discussed in Section 3.1.

It is useful for analysis and interpretation of mapped distributions of aftershocks to define three classes, all of which meet the basic definition. Class 1 aftershocks occur on the same section of the fault surface that slipped in the mainshock, or in a narrow zone bordering it around the edges and possibly in thin sheets on both sides. It is usually assumed that the early aftershocks, those during the first 24 or 48 h, are all Class 1 events; in other words, they define the mainshock rupture surface by a distribution covering it or outlining it. Fault slip modeling based on crustal deformation observations and/or seismic waveforms is useful for obtaining the approximate dimensions of the slip surface, including faults that do not extend to the surface. The results of such modeling are checked against the aftershock distribution as a means of adding reliability to the estimates of fault dimensions and orientation.

Analysis of recorded waveforms is widely used to work out the pattern of slip or moment release on the fault surface. These analyses do show the sites of concentrated slip, but they do not define the limits of the entire slipped area. Mendoza and Hartzell (1988) have examined the relation between the distribution of aftershocks and the patterns of coseismic slip for two well-recorded sequences and have combined these with the results for other sequences to conclude that aftershocks seldom occur where mainshock slip is large. Rather, they “tend to cluster near the edges of areas of maximum coseismic displacement.” Further support for this conclusion is provided by the Northridge mainshock of January 17, 1994 and its aftershocks. The sequence occurred in a place with excellent instrumental coverage. As reported by Dreger et al. (1994), few aftershocks occurred during the first few days at the locations of the patches of maximum slip as determined by waveform analysis. Rather, the 24-72-h aftershocks more or less outline the main slipped area. Hauksson et al. (1994) report that “all of the aftershocks since January 18 have occurred within the zone as defined during the first 24 hours of activity.”

Engdahl et al. (1989) showed that clusters of the aftershocks of the May 7, 1986 Mw-8.0 earthquake in the Andreanof Islands (Alaska) tended to be located in the same places as clusters of background events during the 22 years before that major earthquake.

Class 2 aftershocks occur on the same fault that ruptured to generate the mainshock but are located outside the section of initial slip. Class 2 aftershocks represent the growth of the original aftershock zone, a frequently observed phenomenon discussed below. Class 3 aftershocks occur on faults other than the fault that produced the mainshock but are presumably triggered by the mainshock. Recent experience indicates that these triggered events may be at distances that are large compared to the dimensions of the mainshock rupture.

One characteristic of aftershock sequences known since the beginnings of observational seismology is that the rate at which the events occur decreases steadily with time after the mainshock. In this chapter, we view an aftershock sequence as a relaxation process, with a process relaxation time that is governed by a complex combination of fault-zone properties and ambient physical conditions. The appropriate mathematical relaxation function by which to describe aftershock rate decay is an important issue, because any physical theory that is developed to explain aftershock generation is constrained to yield this function.

2.2 Aftershocks of Normal Depth and of Deeper Mainshocks

Aftershocks are common effects of earthquakes at normal hypocentral depths, i.e., those that occur in brittle crustal material. From standard global earthquake catalogs, such as the Preliminary Determinations of Epicenters of the U.S. Geological Survey or the Bulletin of the International Seismological Center, it appears that even strong earthquakes at depths below about 50–100 km produce few aftershocks (Frohlich, 1987). These catalogs, based largely on data from the global system of seismograph stations, usually list only earthquakes with magnitudes above approximately 4.5, perhaps as small as 4.0 in favorable circumstances. Regional catalogs based on data from dense networks of stations often have events with magnitude 2.0 or smaller. Such networks, when located to monitor subduction zones (Kisslinger, 1993a), may detect clear aftershock sequences for intermediate-depth earthquakes that are similar in...