Investigation:

We investigated the relations between stress changes associated with subduction and seismicity in the upper plate.  Oblique subduction segments were studied to understand how cycle-related stress changes induced by the main event promote or decrease the likelihood of strike-slip in back-arc regions.  The relevant quantity here is the change in Coulomb shear stress on active back-arc faults, equal to the sum of the change in shear stress plus the change in extensional stress multiplied by a coefficient of friction f.  We applied 3D elastic models (Dmowska et al., EOS, 1996; Taylor et al., EOS, 1996, 1997) to study two cases from the Aleutians in which the back-arc regions are seismically active.

Results:

The Andreanof Islands earthquake of May 7, 1986 (Mw = 8.0) was followed in the first 1.5 months by a series of shallow upper-plate earthquakes, the five largest of which range in magnitude between Mw = 5.3 and Mw = 6.5 and are consistent with right-lateral motion on arc-parallel transform faults (Ekström and Engdahl, JGR, 1989, Fig. 1).  The February 4, 1965 Rat Islands earthquake (Mw = 8.7) was followed on July 4, 1966 by a shallow mb = 6.2 upper-plate strike-slip event (Stauder, JGR, 1968), which can be interpreted as a right-lateral earthquake in a relative position and time very similar to the Andreanof Islands earthquakes, here associated with slip on the eastern, strongest asperity of the 1965 main event, (Fig. 3).  Fig. 3 also shows two other strike-slip events (mechanisms from Newberry et al., J Geodyn., 1986), close to Near Island, north of the most western asperity of the 1965 event, forming a doublet of mb 5.9 and 6.0 of February 2, 1975.  This doublet was interpreted as right lateral strike-slip motion along a northwesterly fault plane (Cormier, Geol. Soc. Am. Bull., 1975), as an extension of strike-slip faulting described by Cormier (EOS, 1975) in the Komandorsky Islands.  Newberry et al. (J Geodyn., 1986) prefer to choose the northeast trending nodal plane for these earthquakes, noting that Agattu Canyon to the south of Attu Island and major faults on Attu Island itself are aligned with the trend of this nodal plane. Agattu Canyon has been interpreted as the surface expression of the boundary between two tectonic blocks of the Aleutian Arc.  Choice of the northeast striking nodal plane as fault plane for these events would suggest that this boundary may extend north of Attu Island.

The Aleutians is a convergent margin formed by the right-lateral subduction of the Pacific beneath the North American Plate.  Great earthquakes occur with the highest degree of slip on isolated asperities and so we employ a 3D model of a generic subduction zone as in Dmowska et al. (JGR, 1996) to calculate extensional and shear stress changes in the upper plate arising from oblique slip localized on an asperity sustaining strong stress drop in the center of an interface rupture zone with free slip around it.  Model parameters, such as angle of dip and oblique subduction, are tailored to the Andreanof Islands and Rat Islands separately, but in both cases, the resulting calculated stress changes across the surface of the upper plate form complex distributions.  The greatest magnitude changes occur above the edges of the asperity, but the pattern simplifies somewhat at a distance from the trench beyond the down-dip end of the thrust zone (the area of interest here).  There, for arc-parallel faults, the distribution of right-lateral Coulomb shear stress change, , separates into distinct regions of increase (to the East) and decrease (to the West) relative to the asperity, as shown in Fig. 2 for parameters appropriate for the Andreanof Islands and a mid range friction coefficient.  The arc-parallel line of five back-arc events following the 1986 main event seems to indicate that they represent right-lateral slip on arc-parallel faults and the five x marks indicate the corresponding positions of the events in Fig. 1, measured relative to the asperity from the inversion of Das and Kostrov (JGR, 1990).  They all lie well within the zone of increased static right-lateral Coulomb shear stress change, with values in the range 1-2 bars and as such are consistent with the interpretation of being caused by stress changes from the mainshock.

Calculation of  for arc-parallel faults in the Rat Islands gives a very similar plot to Fig. 2 (despite the increased angle of oblique slip in the 1965 main event relative to that in the Andreanof Islands) and so the position of the July 4, 1966 back arc event is also consistent with right-lateral slip on an EW trending fault plane.  However, Stauder interprets the event as occurring on an arc-perpendicular NS striking fault plane and a plot of the coseismic change in left-lateral Coulomb shear stress resolved onto arc-perpendicular faults,  (Fig. 4), shows that such a mechanism is also consistent with the stress changes at that position in the back-arc, relative to the main asperity (Fig. 3).  These results are fairly insensitive to change in friction coefficient.  Of the two February 2, 1975 back-arc strike-slip events (Fig. 3) the more Westerly of the two occurred just over an hour after the other and was probably as a direct result of the stress changes induced by that event and the one we show results for here.  Figure 5 shows the left-lateral Coulomb shear stress change resolved onto the NE striking plane (as proposed by Newberry et al.) for a f = 0.4. An x marks the associated position of the February 2 seismicity relative to the most Westerly (and smallest) of the three asperities (Fig. 3), and lies in a region of Coulomb shear stress decrease of about 1.5 bars (for typical model parameters). The distribution of right-lateral shear stress resolved onto the alternative (NW, as proposed by Cormier) striking fault plane for the same parameters and f is very similar to that of the left-lateral shear stresses on the NE plane (as expected from a comparison of Figs. 2 and 4), with x in a region of Coulomb shear stress decrease of about 1 bar.  Consequently, it would appear that, for a mid-range f, coseismic Coulomb shear stress changes would tend to suppress seismicity on either of the two fault planes.  In fact, the pattern of Coulomb shear stresses would have changed somewhat in the ten years between the main event and the Feb. 1975 upper plate events.  Experience of stress changes through an entire earthquake cycle (Taylor et al., JGR, 1996 and Zheng et al., JGR, 1996) indicates that the magnitude of the stresses would reverse through the cycle, but the distributed pattern would remain and hence the relative position of the Feb. 1975 seismicity would still lie in a region of decreased Coulomb shear stress (though less in magnitude) for both fault planes and so we cannot rule out the possibility that the back arc seismicity was induced by stress changes from activity on other nearby faults, and not from the main event.  Simpson and Reasenberg, USGS Prof. Paper, 1994, indicate that f might be another time-dependent quantity, due to drainage over time of (coseismically present) pore fluids, which would cause f to increase.  Figure 6 shows right-lateral Coulomb shear stress change for the NW striking fault plane, but with f = 0.8, which illustrates the growth of the region of positive Coulomb shear stress change to the West to encompass the region of the Feb. 1975 seismicity (x), thus promoting seismicity there.  This is not true of the pattern for NE striking plane as the coseismic stress decrease was greater.  If there indeed exists a change in f with time, it may provide another mechanism for triggering of subsequent seismicity in the back arc other than simply the coseismic response to the main event.  Our results also indicate that, beyond the down-dip end of the thrust interface, the difference in length of the asperities along strike (Fig. 3) only modulates the magnitude of the stress changes and the distribution is almost identical.

Non-technical Summary:

An understanding of heterogeneous coupling along subducting plate boundaries is sought through coordinated seismicity observations and computational modeling of stress variations associated with large or great subduction events.  Observed variations in seismicity and deformation signals are interpreted in terms of stress accumulation, release and transfer in the earthquake cycle.  Progress is being made in interpreting seismicity in the over-riding plate, along arc-parallel strike-slip features in the back-arc regions of oblique subduction zones in the Aleutians, in terms of heterogeneous coupling and Coulomb measures of stress change.

Reports:

Taylor, M. A. J., R. Dmowska and J. R. Rice, Coulomb shear stress changes along subduction segments and seismicity in the upper plate (abstract), EOS Trans. Amer. Geophys. Union, 1997, in press.
 
Taylor, M. A. J., R. Dmowska and J. R. Rice, Upper plate stressing and back arc seismicity in the subduction earthquake cycle, submitted to J. of Geophys. Res., 1997.

Page established by:  Mark Taylor, Division of Engineering and Applied Sciences, Harvard University, Nov. 26, 1997.