news & views
SEISMOLOGY

Diary of a wimpy fault

Subduction zone faults can slip slowly, generating tremor. The varying correlation between tidal stresses and
tremor occurring deep in the Cascadia subduction zone suggests that the fault is inherently weak, and gets weaker
as it slips.

Roland Bürgmann
three centimetres of total offset during each
slip event 4. Houston finds that not only is
tremor (and thus fault slip) modulated by
tides, but the strength of the correlation
between the tides and tremors increases as
slow slip progresses. This implies that the
fault becomes increasingly sensitive to tides
and progressively weakens as slip reaches
a total of a few centimetres. So, not only
are tremor-producing faults easily pushed
around by external forces, they continue to
weaken as they go.
Tremors have a story to tell about how
deep faults work. The complex distribution

of tremors in space and time and their
response to external stresses illuminate
the mechanical properties of the deep
fault 5,6. Some frictional models of slow-slip
events feature slip-weakening distances
that are large compared with those found
in laboratory experiments (10–5 to 10–6 m);
however, there are a number of alternative
ways in which models can be parameterized
to produce such events6,7. The observation
of a slowly increasing correlation of tides
and tremors during the Cascadia slow-slip
events indicates that the fault may indeed
need to slip a relatively large distance, on the

0
Seattle
10
Depth (km)

T

he discovery of tectonic tremor in
the deep roots of plate-boundary
subduction zones in Japan and
North America was a great surprise. These
unusual and low-amplitude seismic signals
emanate from a depth in the Earth where
rocks should be too hot and too viscous to
produce seismic events. The tremors were
found to be caused by episodic slow-slip
events on the deep sections of plateboundary faults, well below the parts of
the fault that produce great earthquakes.
Tremors can be triggered and modulated
by the small stresses associated with body
and ocean tides, implying that the faults that
generate them must be very weak. Writing
in Nature Geoscience, Houston1 analyses
tremor events in the Cascadia subduction
zone between 2007 and 2012. She shows
that not only are the deep parts of the fault
very weak, but that the fault becomes weaker
during individual slip events.
Tremors are the weak cousins of regular
earthquakes. While it took decades of
painstaking work to document the modest
effect of tides on regular earthquakes,
tremors were found to be easily triggered
by these transient stresses2. This extreme
sensitivity to very small stress changes
indicates that the tremor-producing fault is
frictionally weak, in large part due to fluids
in the fault zone being at near-lithostatic
pressure. That is, the overburden of rock
is counteracted by fluid pressures of
comparable magnitude in the fault. The high
fluid pressure leads to a very low effective
frictional strength of the fault, and slip and
tremor can readily respond even to tiny tidal
shear stresses of less than 100 Pa (ref. 3).
Houston1 explores in detail the
occurrence of tens of thousands of tremors
created by six slow-slip events in the
Cascadia subduction zone between 2007
and 2012 (Fig. 1). Each slow-slip event
took several weeks to complete, with the
slip front propagating at rates of several
kilometres per day along a nearly 200-kmlong stretch of the fault. At any given point
on the fault, tremors and slip continue for
several days but accumulate only two to

10 km

20
30

20 km

40
30 km
50
60
127

40 km
126

125

Long
it

50 km
124
123
ude (
° W)

122

121

45.5

46

46.5

47

47.5

ude

it
Lat

48.5

48

49

)

(° N

Time during 2010 slip event
8 August

9 September

Figure 1   Recurring slip events deep on the Cascadia subduction zone are illuminated by tremors. The
tremor locations (circles) are coloured by date of occurrence during a month-long tremor-and-slip
episode in 2010. Houston1 finds that the correlation between tremor and tidal stress at a point on the fault
strengthens during a slip event, implying that the fault weakens as slip evolves. At depths of less than
about 20 km (purple shaded region), the megathrust appears to be locked and therefore builds up stress
towards the next great subduction earthquake. Figure courtesy of Aaron Wech, US Geological Survey.

NATURE GEOSCIENCE   ADVANCE ONLINE PUBLICATION   www.nature.com/naturegeoscience 

© 2015 Macmillan Publishers Limited. All rights reserved

1

 news & views
scale of a centimetre, before it weakens to its
frictional sliding strength.
Houston suggests that the observed
drop in fault strength over the course
of a slow-slip event could involve the
breakage of mineral precipitates in the fault
zone. The minerals rapidly regrow in the
approximately 14-month-long recurrence
interval between slow-slip events, during
which time the fault strength recovers
and stress accumulates8. Interestingly, a
transition from tremor modulated by the
rate of peak tidal shear stress to tremor
modulated by the amplitude of peak tidal
shear stress during a large slow slip event
has recently been documented in Cascadia9.
This observation adds another twist to the
relationship of tides and tremors and should
further illuminate the underlying physics of
the slow-slip process.
Tremor-producing faults are weak, but
it was not known if the weakness stems
only from very high fluid pressures or if
the fault-zone rocks also have a very low
friction coefficient. Taking advantage of
differences in the temporal variations
of tidal fault-perpendicular stress and
mean stress (pressure), Houston is able
to estimate the intrinsic, fluid-pressureindependent friction coefficient for the

2 

fault in situ. She finds values for the average
friction coefficient well below 0.2 — much
lower than most known rock types, with
friction values between 0.6 and 0.8. Such
low values require unusually weak fault
zone materials, such as talc, graphite and
saponite, to effectively lubricate the fault
zone. Saponite breaks down at temperatures
well below those found in the tremor zone,
but talc and graphite can be stable to very
high temperatures and pressures10, so
could exist in the deep fault zone beneath
the Cascades.
Above the weak tremor zone in Cascadia,
the seismogenic part of the fault is strongly
coupled and capable of producing great
megathrust ruptures. The last such event
in 1700 was a not-so-wimpy magnitude 9
earthquake. Given that the Cascades
are home to large cities such as Seattle,
Portland and Vancouver, we must continue
to investigate potential links between
slip on the deep subduction zone and
megathrust earthquakes on locked parts
of the fault. In the process, we are bound
to learn more about the physics of deep
plate-boundary faulting with improved and
targeted seismologic, geodetic, gravity, and
electro-magnetic observations, including
those from space-borne systems, borehole

sensors, and seafloor instrumentation.
As large earthquakes often initiate close
to the tremor zone, insights gained from
these observations may enable improved
characterization of time-dependent
earthquake hazard and of precursory
processes that appear to precede some large
earthquake ruptures. 
❐
Roland Bürgmann is in the Department of Earth
and Planetary Science, University of California,
Berkeley, California 94720–4767, USA.
e-mail: burgmann@seismo.berkeley.edu
References

1.  Houston, H. Nature Geosci. http://dx.doi.org/10.1038/ngeo2419
(2015).
2.  Rubinstein, J. L., La Rocca, M., Vidale, J. E., Creager, K. C. &
Wech, A. G. Science 319, 186–189 (2008).
3.  Thomas, A. M., Nadeau, R. M. & Bürgmann, R. Nature
462, 1048–1051 (2009).
4.  Wech, A. G., Creager, K. C. & Melbourne, T. I. J. Geophys. Res.
114, B10316 (2009).
5.  Beeler, N. M., Thomas, A., Bürgmann, R. & Shelly, D.
J. Geophys. Res. 118, 5976–5990 (2013).
6.  Hawthorne, J. C. & Rubin, A. J. Geophys. R.
118, 1216–1239 (2013).
7.  Liu, Y. & Rice, J. R. J. Geophys. Res. 112, 1978–2012 (2007).
8.  Audet, P. & Bürgmann, R. Nature 510, 389–392 (2014).
9.  Royer, A. A., Thomas, A. M. & Bostock, M. G. J. Geophys. R.
120, 384–405 (2015).
10. Moore, D. E. & Rymer, M. J. Nature 448, 795–797 (2007).

Published online: 27 April 2015

NATURE GEOSCIENCE   ADVANCE ONLINE PUBLICATION   www.nature.com/naturegeoscience

© 2015 Macmillan Publishers Limited. All rights reserved