Fault Structure

From geological investigations of active fault traces an earthquake is considered to rupture several fault segments, which are connected by jogs, steps and branches with each other [Aydin and Du (1995)]. During an earthquake, the rupture propagates mainly along the preexisting fault segment and sometimes jumps to the neighbouring segment by creating a new fault segment.

Along the rupture zone of the 1992 Landers, California, earthquake (Mw7.3), a complicated fault system existed before the earthquake [Aydin and Du (1995)]. During the earthquake, the rupture propagated along the fault traces by selecting its route by itself. In Fig. 1, the preexisting fault

Fig. 1 Distribution of active fault traces around the source region of the 1992 Landers earthquake. Thick lines indicate the fault traces where the rupture propagated during the Landers earthquake. The star symbol stands for the hypocenter location of the Landers earthquake [Modified from Aochi and Fukuyama, (2002)].

Fig. 1 Distribution of active fault traces around the source region of the 1992 Landers earthquake. Thick lines indicate the fault traces where the rupture propagated during the Landers earthquake. The star symbol stands for the hypocenter location of the Landers earthquake [Modified from Aochi and Fukuyama, (2002)].

traces and those followed by the rupture are shown. We found that how the rupture selected the fault trace at the branch basically depends on both the stress field applied to the fault branch and the rupture velocity [Aochi et al. (2000b)].

At seismogenic depths, we cannot see fault traces directly except for very ancient activity at depth, now exposed on the surface [e.g. Swanson (1988)]. We can see the currect fault structure at seismogenic depths in indirect ways. Along the fault trace, many microearthquakes occur. If these events are located very precisely, fault traces can be imaged. Very dense seismographic networks now enable us to locate very small earthquake with sufficient accuracy, which provides us with a detailed image of the fault system at seismogenic depth.

On October 6, 2000, a Mw6.6 earthquake with strike slip faulting occurred in western Tottori, southwest Japan. This was the first earthquake that occurred after the densely distributed seismographic network had been constructed. Following this earthquake, more than 10,000 aftershocks were recorded by the network and relocated [Fukuyama et al. (2003a)] by a very accurate technique called Double Difference Method [Waldhauser and Ellsworth (2000)]. The aftershock distribution shows a complicated image of the fault structure [Fukuyama et al. (2003a)] as shown in Fig. 2. The main-shock fault system consists of four fault segments (#1-#4 in Fig. 2(c)). Other fault segments (#5-#13) were created by the aftershock sequence, some of which were responsible for the post seismic deformation observed by GPS measurements [Sagiya et al. (2002)].

Focal mechanisms of aftershocks were calculated using the regional broadband seismic network [Fukuyama et al. (1998); Fukuyama et al. (2003a)], whose moment magnitudes are greater than 3.5. The fault strike directions of these aftershocks and the fault traces recognised from relocated hypocentral distribution are found to be consistent with each other, as shown in Fig. 2(b). Most aftershocks whose focal mechanisms were determined occurred along the pre-existing fault traces or parallel to them, and the lineaments inferred from the aftershock distribution are considered to be the fault structure at seismogenic depth.

We are therefore able to use the information on the geometry of the fault based on the active fault traces appearing on the surface, as well as seismic activities along the fault traces. But this information is sometimes insufficient, especially for an earthquake occurring in a seismically inactive region. This is sometimes called a blind fault. In order to overcome these situations, active seismic surveies such as a shallow reflection survey

Okhotsk Plate

Fig. 2 (a) Location of the 2000 western Tottori earthquake plotted with plate configurations. PH, EU, NA and PA represent Philippine Sea, Eurasia, North American and Pacific plates, respectively. (b) Hypocentre distribution relocated by double difference method. Dots are hypocentres and straight lines are strike directions of the fault determined by the moment tensor inversion of regional broadband waveforms. Optimum fault direction for two possibilities for the focal mechanism are chosen based on the aftershock distribution. (c) Fault models based on the hypocenter distributions. Faults #1—#4 were created during the mainshock and other faults (#5—#13) are related to aftershock activity. Faults #22—#24 are caused by the largest aftershock two days later [Modified from Fukuyama et al. (2003a)].

Fig. 2 (a) Location of the 2000 western Tottori earthquake plotted with plate configurations. PH, EU, NA and PA represent Philippine Sea, Eurasia, North American and Pacific plates, respectively. (b) Hypocentre distribution relocated by double difference method. Dots are hypocentres and straight lines are strike directions of the fault determined by the moment tensor inversion of regional broadband waveforms. Optimum fault direction for two possibilities for the focal mechanism are chosen based on the aftershock distribution. (c) Fault models based on the hypocenter distributions. Faults #1—#4 were created during the mainshock and other faults (#5—#13) are related to aftershock activity. Faults #22—#24 are caused by the largest aftershock two days later [Modified from Fukuyama et al. (2003a)].

would be useful [Sato et al. (2004)]. However, for strike slip faults, these experiments may not work properly because of the unclear vertical offsets of layers at depth.

Was this article helpful?

0 0

Post a comment