Comment on “Coupling Semantics and Science in Earthquake

Eos,Vol. 85, No. 36, 7 September 2004
Comment on “Coupling Semantics and
Science in Earthquake Research”
PAGES 339–340
Kelin Wang and Timothy Dixon (Eos, 85(18),
4 May 2004,p.180) thoughtfully advocate paying
close attention to semantics in descriptions of
fault zone properties and kinematics,an increasingly important issue given the distinct usages
of terms such as “coupling” by separate disciplines involved in the multidisciplinary study
of earthquake faulting.We are in full accord
with their advocacy of unambiguous language,
such as the description of a nonsliding fault
segment as being “not slipping” rather than
“strongly coupled” in the absence of any information about the frictional or stress state of
that segment.While several of Wang and Dixon’s
recommended “simple expressions”have clear
merits, we feel that their advocacy of “locked”
to equate to “not slipping” is not an improvement, and that their accompanying illustration
of dislocation models of subduction zone
megathrusts is potentially misleading.
Wang and Dixon critique a simple,one-dimensional dislocation model for an interplate thrust
event, for which conventional thinking is that
the principle seismogenic zone is not sliding
between earthquake ruptures but that there is
steady sliding occurring along the shallow
and deep extensions of the thrust plane.These
stable sliding portions of the fault plane are
assumed to be regions of velocity-strengthening
frictional conditions; they accommodate relative plate motions without earthquake failure,
although portions may be conditionally stable,
driven to failure by the high strain rates (large
changes in slip velocity) that accompany rupture of the main seismogenic zone [e.g., Scholz,
1998].Wang and Dixon argue that this model
is “incorrect” and that the updip region is not
slipping steadily, and should be viewed as
“locked,” along with the unstable sliding region.
They invoke an analogy involving a book on a
level table with no shear stress being applied;
and it is correctly asserted that this stable
equilibrium state does not allow strength or
nature of frictional coupling to be deduced.
However, this analogy seems irrelevant to the
situation of interplate thrust faults, which are
not in a state of stable equilibrium and are
being continuously loaded by forces associated
with slab-pull, ridge-push, and lateral loading
by slip of adjacent segments both along the
strike and dip of the megathrust [e.g., Lay
et al., 1989].
A more complex but more realistic visualization of the megathrust frictional environment
is offered in Figure 1 [Lay and Bilek, 2004].
This cartoon adopts the frictional notions from
Pacheco et al.[1993] and Scholz [1998],describing two-dimensional variations in friction on
the fault contact as a distribution of stable,
unstable, and conditionally stable domains.The
shallowest part of the fault is viewed as primarily a region of velocity-strengthening material;
shear stress applied to this zone will cause it
to creep stably (possibly episodically) rather
than rupture in an earthquake [e.g., Byrne et al.,
1988].At greater depth the fault surface has
patches with velocity-weakening material that
fail unstably in stick-slip earthquakes. Regions
around the unstable zones and possibly isolated portions of the fault contact are conditionally stable, capable of failing in earthquake
rupture if driven by failure of nearby regions of
unstable slip.
In the generalized cartoon of Figure 1, the
application of shear stress to the overall fault
zone is not localized only to regions of unstable sliding potential, but is distributed across
the entire fault surface as a result of loads
applied to the system by slab-pull, ridge-push,
and failure of adjacent portions of the megathrust.
The notion of the updip region of stable sliding
conditions being “locked” (weak but not slipping), as advanced in the “correct” model of
Wang and Dixon,requires that there be no stress
on the shallowest portion of the fault; essentially one requires a stress shadow caused by
other “locked”(strong but not slipping) regions.
This seems unlikely in general. In addition, the
word “locked” carries with it the implication
that it would resist motion were stress to be
applied (one of those subtle semantic concerns
that Wang and Dixon alert us to), but if stress
is applied to a region of stable sliding frictional
conditions, it will simply stably slide. Even if the
region does slip following a deeper stick-slip
event, it will not fail coseismically because its
behavior is governed by different frictional
conditions. Using the same terminology to
describe regions of very different frictional
properties is likely to engender confusion.
While there may well be regions updip of seismogenic zones that are indeed completely
buffered by surrounding locked regions of
unstable friction and are not slipping, we feel
it is at least premature, if not generally wrong,
to assert that this is always the case.
That assertion can be tested by fully characterizing deformation over the entire fault plane
throughout the seismic cycle.This is a formidable challenge, but some progress is being
made. For example, recent dislocation modeling of GPS data from the Nicoya Peninsula,
Costa Rica [Norabuena et al., 2004] is inconsistent with the assertion.The Nicoya Peninsula
extends over the shallow portion of the
Cocos/Caribbean-Panama Block plate boundary and is one of the only places in the world
where land-based geodesy provides resolution
of interseismic deformation of the shallow
thrust interface. Geodetic modeling indicates
a strong transition from slip at more than 60%
to less than 25% of the plate rate 30–40 km from
the trench (fault depth of 5–8 km).Although
resolution falls off dramatically near the trench,
the strong gradient in shallow slip, at the very
least,requires modification of Wang and Dixon’s
“correct” model of interseismic deformation.
As coordinated efforts to carefully map the
relationship between geodetically defined notslipping regions and seismicity distributions
expand,complex relationships are being revealed,
as is the case along the Nicoya Peninsula
where the apparent seismic front defined by
microseismicity is downdip of the region with
the least slip [Norabuena et al., 2004]. It seems
that much more work is needed before we can
define end-member models of “incorrect” and
“correct” slip distributions on both thrust and
strike-slip faults,and this should not be done in
the context of one-dimensional models or stable
equilibrium analogs.
Fig. 1. Cartoon notion of heterogeneous frictional conditions on the interplate thrust fault in a subduction zone.Thermal, hydrological, and material properties give rise to shallow regions beneath
any accretionary wedge that are not observed to have earthquake slip and are inferred to be in a
velocity-strengthening stable sliding regime.At greater depth an increasing percentage of the fault
plane has velocity-weakening frictional conditions that give rise to unstable stick-slip failure.
These may be associated with areas of bathymetric roughness such as seamounts and horst and
graben structures on the subducting slab. Regions of conditional stability, which tend to stably
slide unless driven by strong velocity increases, may surround unstable slip regions or exist in isolated patches. Modified from Bilek and Lay [2002].
Eos,Vol. 85, No. 36, 7 September 2004
Bilek, S. L., and T. Lay (2002),Tsunami earthquakes
possibly widespread manifestations of frictional
conditional stability, Geophy. Res. Lett., 29(14),
1673, doi:10.1029/2002GL015215.
Byrne, D., D. Davis, and L. Sykes (1988), Loci and
maximum size of thrust earthquakes and the
mechanics of the shallow region of subduction
zones, Tectonics, 7, 833–857.
Lay,T., and S. Bilek (2004),Anomalous earthquake
PAGE 340
We thank Thorne Lay and Susan Schwartz
for their comment on our Forum article (Eos,
85(18), 4 May 2004, p. 180).They agree with
our main point that slip rates of a fault should
not be confused with stress conditions or frictional properties, but they criticize our use of
the word “locked” and the interseismic deformation model we used to illustrate a conceptual error.We agree with Lay and Schwartz
that the term “locked”has connotations beyond
purely kinematical and that “no slip” may be
more appropriate.The present reply is to further discuss the meaning of the simple deformation model.
In that 2-D (no along-strike variation) example,
the segment updip of the locked zone of a
subduction fault is assumed to be weak (and
may have a stable frictional behavior). Our
criticism was solely to the assumption that this
updip segment could slip steadily at the plate
convergence rate for a long time, not on other
aspects of the model.We have no general disagreement with the more complex fault model
presented in Lay and Schwartz’s Figure 1, but we
feel that how a real subduction fault behaves is a
separate issue (of course an important one!).
Our choice of using the simple 2-D model was
simply to clarify essential concepts.
Lay and Schwartz mistakenly think the alternative model that we labeled “correct” was
meant to be generally applicable.As clearly
stated in our figure caption, the condition for
this model being correct is “if the updip zone
is not slipping.” Whether or when the segment
could slip is a different issue, and is discussed
separately in our article.The following two
questions regarding this model need to be further addressed.To simplify discussion, here we
ignore deformation of the subducting plate.
1.What forces drive the frontal wedge of the
upper plate above the weak segment to move
seaward relative to the lower plate? Since
ruptures at shallow depths on subduction zone
megathrusts, in Interplate Subduction Zone Seismogenesis, edited by.T. Dixon et al., Columbia Univ.
Press, New York, in press.
Lay,T., L.Astiz, H. Kanamori, and D. H. Christensen
(1989),Temporal variation of large intraplate
earthquakes in coupled subduction zones, Phys.
Earth Planet. Inter., 54, 258–312.
Norabuena, E., et al. (2004), Geodetic and seismic
constraints on some seismogenic zone processes
in Costa Rica, J. Geophys. Res., in press.
basal traction (along the fault) can only resist
slip, the driving force comes from the upper
plate material landward of the wedge, that is,
compressive stresses within the upper plate.
Change of gravitational potential due to vertical
deformation in earthquake cycles may also play
a small role. Ridge push and slab pull, alluded
to by Lay and Schwartz, are irrelevant here
because they act only on the subducting plate.
2. Can this weak segment slip at all if the
downdip zone is locked? As we stated in our
Forum, it indeed can, but “slipping at the plate
convergence rate, as often modeled, is unlikely
to be sustained over the entire inter-seismic
period.” If the locked zone slips, either as an
earthquake or aseismic event, but the updip
segment does not or slips more slowly, compressive stress is increased in the upper plate
above the two segments.The relief of this
incremental stress (release of strain energy)
can cause the segment to slip in a transient
fashion while the downdip segment is locked.
But it cannot slip if the stress is relieved. It is
not a surprise that, at a given point in time, we
may see an updip segment slip faster than a
downdip segment. Determining when, where,
and why such transients occur is a key scientific goal.
In most cases, land GPS data cannot uniquely
resolve whether the far offshore part of the
fault is slipping very slowly (e.g., at the plate
convergence rate) or is not slipping. It is a
common and possibly incorrect practice to
assign a plate convergence rate (i.e., zero
backslip) to the most seaward part of the plate
interface when inverting GPS data to infer the
state of fault locking.Using a different boundary
condition (no slip or slipping at a lower rate)
will change the results, especially for the offshore part. Direct near-trench observations are
needed to resolve this issue. It is noteworthy
that new seafloor geodetic data from the Peru
subduction zone [Gagnon et al., 2004]
supports the concept of a “no slip” condition
on the plate interface all the way to the trench
axis as portrayed by our “correct” model.
Pacheco, J. F., L. R. Sykes, and C. H. Scholz (1993),
Nature of seismic coupling along simple plate
boundaries of the subduction type, J. Geophys.
Res., 98, 14,133–14,159.
Scholz, C. (1998), Earthquakes and friction laws,
Nature, 391, 37–42.
Sciences Department and Center for the Study of
Imaging and Dynamics of the Earth, University of California, Santa Cruz
Lay and Schwartz quoted the result at the
Nicoya Peninsula that microseismicity occurs
downdip of the region with the least fault slip
[Norabuena et al., 2004].This result and the
many other more intriguing observations
[e.g., Freymueller et al., 2000; Ozawa et al.,
2002; Rogers and Dragert, 2003; Uchida et al.,
2003; Yagi et al.,2003] reflect the heterogeneous
and transient nature of fault motion in the
real world, but in our view they do not negate
simple models that are designed to explain
basic concepts.
Freymueller, J.T., S. C. Cohen, and H. J. Fletcher
(2000), Spatial variations in present-day
deformation, Kenai Peninsula,Alaska, and their
implications, J. Geophys. Res., 105, 8079–8101.
Gagnon, K. L., D. Chadwell, and E. Norabuena (2004),
Seafloor geodetic measurements of Nazca-South
America plate stick-slip behavior, Eos Trans.AGU,
85(17), Jt.Assem. Suppl.,Abstract G21B-04.
Norabuena, E., et al. (2004), Geodetic and seismic
constraints on some seismogenic zone processes
in Costa Rica, J. Geophys. Res., in press.
Ozawa, S., M. Murakami, M. Kaidzu,T.Tada,T. Sagiya,Y.
Hatanaka, H.Yarai, and T. Nishimura (2002), Detection and monitoring of ongoing aseismic slip in
the Tokai region, central Japan, Science, 298,
Rogers, G. C., and H. Dragert (2003), Episodic tremor
and slip on the Cascadia subduction zone: The
chatter of silent slip, Science, 300, 1942–1943.
Uchida, N.,T. Matsuzawa,A. Hasegawa, and T. Igarashi
(2003), Interplate quasi-static slip off Sanriku,
NE Japan, estimated from repeating earthquakes,
Yagi,Y., M. Kikuchi, and T. Nishimura (2003),
Co-seismic slip, post-seismic slip, and largest
aftershock associated with the 1994 Sanriku-haruka-oki,
—KELIN WANG, Geological Survey of Canada,
Sidney, British Columbia; and TIMOTHY DIXON, University of Miami, Coral Gables, Fla.