Non-technical summary: The Greater Tokyo area lies at the complex junction of three major tectonic plates: the Philippine Sea, Pacific, and Honshu/Eurasia plates.  Although the region contains many faults capable of producing damaging earthquakes, the largest and most destructive earthquakes occur along the PHS-Honshu dipping slab interface which descends from the Sagami trough.   The two most recent of these events, the M~8 1703 Genroku earthquake and M=7.9 1923 Kanto earthquake involved several meters of slip through a combination of strike-slip and dip-slip components (just like the recent M=9.0 Sumatra earthquake).  A better understanding of these recent earthquakes is needed in order to understand the processes of tectonic strain accumulation over 10s to 100s of years, a culminating seismic event, which occurs instantaneously, and subsequent postseismic readjustment over a period of years.

We investigate the source process of the 1923 earthquake using a large available geodetic dataset, consisting of first and second order triangulation data  and leveling data.  Nyst et al. (2005) investigate simple models of the 1923 source, employing uniform slip models on one or two rupture planes.  The complementary study by Pollitz et al. uses the fault geometry obtained by Nyst et al. as a starting point for deriving more detailed distributed slip models of the 1923 earthquake.  The novel aspect of both studies is the use of a very large second order triangulation dataset that had not been used in previous source studies of the 1923 earthquake.  This new data, which was provided by the Geographical Survey Institute, Japan, expanded the information provided by triangulation observations by a factor of 10 and, in conjunction with the leveling data, allows a very detailed examination of the slip distribution on the causative faults of the 1923 earthquake. We find that coseismic slip is localized in two main areas (Figure 1a) -- on a NE-dipping plane near Odawara with a maximum of ~8 meters slip and on a NE-dipping plane beneath the Miura peninsula and somewhat further east beneath the southern Boso peninsula.

Unusually intense aftershock activity after the 1923 mainshock has long suggested additional sources involved in the 1923 sequence which were suspected to be in the locality of the Boso peninsula and which may not necessarily be associated with the PHS-Honshu interface.  The second order triangulation provides a vital clue as to the nature of an additional fault.   Representation of the second order angle changes in terms of the direction and magnitude of the local principal strain axes (Figure 2) reveals a WNW-trending zone of high right-lateral shear strain through the southern Boso peninsula. The distributed slip model of Figure 1a cannot well predict this feature of the strain pattern, but we find that it can be predicted with a model of combined PHS-Honshu distributed slip plus ~3 meters right-lateral strike slip on a buried WNW-trending fault ("Boso Splay" fault)  in the southern Boso peninsula (Figure 1b).   This fault is thought to be a steeply dipping discontinuity extending possibly to the upper surface of the PHS slab.   The proposed rupture of this fault in the 1923 earthquake sequence is consistent with the aftershock data, the existence of mapped right-lateral faults on the Boso peninsula, and the gravity anomaly pattern.   Furthermore it resolves a kinematic discrepancy, identified by Nyst et al., between  PHS and Honshu relative motions during the earthquake versus their relative motions over geologic time as constrained by plate motion studies.  The PHS-Honshu slip required near the Boso peninsula is substantially reduced by the involvement of the Boso Splay in the 1923 sequence.

Significant slip appears small or absent south and east of the Boso peninsula.  This raises the question as to how slip has been accommodated on the PHS-Honshu interface during the most recent seismic cycle.  Interseismic strain accumulation appears evenly distributed on the shown PHS-Honshu planes (Nishimura and Sagiya, 2004), including that portion where 1923 slip was low.  This raises the possibility that a "slip deficit" remains on the interplate boundary south and east of the Boso peninsula.  However, aseismic slip may occur on this portion of the plate boundary as suggested by the occurrence of a "silent earthquake" just east of the Boso peninsula in 1996 (Sagiya, 2004).

Figure 1




Figure 2