OPINION
Stress-forecasting Earthquakes
tice, computing and foreknowledge clearly fail by some very
substantial orders of magnitude.
This type of behavior is common in even quite simple
In recent Opinions in Seismological Research Letters Paul Silnonlinear
systems where negligibly small differences in iniver has suggested that earthquake prediction is not possible,
tial
values
can lead to substantially different results. It is just
whereas Lowell S. Whiteside suggests that it is. Earthquake
not
possible
to get sufficient resolution in initial conditions
prediction means many things. There are at least three types
to
predict
the
final result. This is known as deterministic
of (conventional) earthquake prediction. Deterministic prechaos,
and
it
is
one of the extraordinary properties of such
diction is where the behavior before the earthquake (the
systems
that
deterministic
equations can lead to effectively
stress interactions with the surrounding rocks, say) can be
chaotic
results.
Even
more
extraordinary, these chaotic syscalculated (by whatever techniques are available) so that the
tems
may
impose
some
form
of order on the behavior which
time, place, and magnitude of the future large earthquakes
may
be
common
across totally different
can be estimated within well defined
systems
and
can
in
some instances be calwindows. Statistical prediction is where
culated.
It
is
suggested
that stressseismicity in the past can yield estimates
Since
the
Earth
is
complex
induced
manipulations
of
fluid-satuofseismicity in the future. The third and
and
heterogeneous
at
all
rated
microcracks
is
one
such
calculable
most common type is where some key
system
(see
below).
Such
systems
are
scale lengths deterministic
precursory phenomenon or a group of
commonplace
and
embrace
a
huge
range
phenomena indicate that a large earthearthquake prediction is
of phenomena including weather sysquake is imminent. I suggest that all
impossible and consistent
tems, turbulent flow, convection curthree types cannot predict time, place,
precursory phenomena are
rents, geomagnetic reversals, cardiac
and magnitude of a future large earthextremely
unlikely.
fibrillations, dripping taps, chemical
quake. It is complexity and heterogenereactions, percolation systems, clustering
ity that prevents it each time.
on
freeways, and water drops on window
(1) The difficulty as far as determinpanes,
among
many
others.
It is not surprising that fluidistic prediction is concerned is that the Earth is immensely
rock
interactions
in
the
crust
are also such a nonlinear,
complex and heterogeneous at all scales. Heterogeneous feapotentially
chaotic
system.
tures include stress fields, geology, and water saturation, as
(2) Statistical analysis ofseismicity in the past in order to
well as the dimensions, frictional properties, orientation, and
attempt
to predict future behavior again fails because of
focal mechanisms of fault planes, among many others. These
complexity
and heterogeneity. It is clear from seismicity (and
range in size from submillimeter grains to faults and photofrom
episodes
of orogeny, say) that earthquake occurrences
lineaments of thousands of kilometers (more than
vary
enormously
and cluster in time and in space (more
nine orders of magnitude). If the stress-induced interaction
deterministic
chaos),
so that the seismicity of any area varies
of every grain-boundary crack in a stressed fluid-saturated
with
time
on
all
scales
from seconds to millions of years
rockmass before a large earthquake could be evaluated, then
(again
~
10
orders
of
magnitude).
Assumptions of statistical
I dare say it might be possible in principle to predict deterstability
can
be
expensive.
The
Parkfield
Earthquake Predicministically the time, place, and magnitude of the future
tion
experiment,
where
a
sixth
earthquake
was expected after
large earthquake. Since the strained rockmass before a large
a
sequence
of
five
34--6
earthquakes
repeated
regularly every
earthquake is at least a million cubic kilometers (and could
twenty-two
years,
is
already
at
least
five
years
beyond
its selleven be hundreds of millions of cubic kilometers for the largby
date.
Previous
seismicity
is
clearly
important,
and
if one
est earthquakes), the required interaction involves possibly
wanted
a
really
aseismic
place
to
live
then
history
would
sugup to some 1024 millimeter-diameter grains each with graingest
that
Saharan
Africa,
much
of
Arctic
Canada,
and
Arctic
boundary cracks. Even if I am exaggerating by a factor of
Asia would appear to be seismically safe, although very
1012, say, the remaining 1012 plus interactions (a fault plane
uncomfortable (is there a message?). Hsewhere, even in the
of single grains, say) are still far too great for any conceivable
centers
of what are presumed to be stable continental shields,
computation and far too great for conceivable evaluation of
there
are
occasional large earthquakes which would not be
initial starting conditions around each grain. So although in
expected
from
historical seismicity.
principle deterministic prediction may be possible, in prac-
Seismological ResearchLetters Volume70, Number3 May/June1999 291
My favorite example of unpredictability was near my
pores at different orientations to the stress field. Consequently, the immediate effect of any deformation, however
home village. The U.K. has low to moderate seismiciry.
small, is to modify the geometry of the intergranular fluidProbably the most damaging British earthquake in the past
saturated microcracks by reorganizing the
400 years was the 1884 M L = 4.6 Colchester
distribution of crack aspect ratios. This modearthquake, which had a shallow focus and
But earthquakes can
ifies the anisotropic elastic symmetry and
produced high intensities. Many houses lost
means that changes in stress-induced deforchimneys and roof tiles and some churches
be stress-forecast.
mation can be directly monitored by analyzwere damaged, but direct casualties were liming changes in shear-wave splitting. In the
ited to a few people injured possibly by fallpast, such changes have been recognized
ing tiles. Apart from a small foreshock a few
before only a small number of earthquakes.
days previously and doubtful accounts of aftershocks, there
The reason for the small number of observations is that
is no other known earthquake within 50 km of Colchester in
monitoring changes in shear-wave splitting before large earthat least 500 years. Not easy to predict!
quakes requires very stringent source-to-recorder geometries
(3) The difficulty as far as the third type of prediction,
almost immediately above small-magnitude seismicity in the
seeking and evaluating precursors, is again the immense comneighborhood of the epicenter of a larger earthquake. Until
plexity and heterogeneity of the Earth at all scales. Although
recently, such changes had only been observed (with hindlarge numbers of different precursors have been observed
sight) on four occasions around the world when by chance
before earthquakes, their occurrences are extremely variable.
these geometrical conditions were met.
No two earthquakes are exactly the same, and it is extremely
The rapidly expanding SIL seismic network in Iceland
unlikely that there will be some common recognizable
combined with strong but localized seismicity has meant
unique phenomenon or phenomena that signals the time,
that changes in shear-wave splitting are now seen routinely
place, and magnitude of an imminent earthquake. The fact
(but again with hindsight) before larger earthquakes in a
that we have been searching for such earthquake precursors
small area of southwest Iceland. Such changes were also seen
since John Milne in the 1880's, and that no large earthquake
and recognized before the 1996 volcanic eruption beneath
has been successfully and unquestionably predicted (even
the Vatnaj6kull ice cap. Had we had more confidence in our
with hindsight), suggests that the hopes for consistent preresults, we could have predicted that something was going to
cursors can be discounted. Only an incurable optimist could
happen, although at that time we would not have expected
believe that there is still a "magic" precursor packed with
an eruption. It was thefirst five months of data from thefirst
information about future earthquakes awaiting discovery.
seismic station in Iceland which had been examined for
So what can be done? Paul Silver suggests up to 2,000
shear-wave splitting, and we did not believe our luck!
"plate boundary deformation sites" at 10 km intervals over
Although we have demonstrated that shear-wave splitCalifornia each with GPS receivers, borehole strainmeters,
ting can identify increases in stress before earthquakes, smallborehole three-component seismometers, and borehole
scale seismicity is far too intermittent to be relied on for rouaccelerometers. This would certainly be a powerful tool for
tine monitoring. What is needed is some form of controlled
investigating plate boundary deformation, but there is no
source seismology. Standard exploration techniques, such as
direct evidence that it would be any use at all in mitigating
reflection surveys and vertical seismic profiles, although proearthquake hazards. We would not know what to look for in
ducing reliable and consistent observations of shear-wave
such an enormous flood of data. Strain evaluated from GPS
splitting, lack resolution because of the scattering and attenobservations is deformation of the largely destressed unconuation of the near-surface rocks. They also do not readily
solidated surface layers of the Earth, which may have behavsample the crucial range of angles of incidence, between 15 ~
ior substantially different from the strain at the depths of
and 45 ~ to the vertical plane of the approximately parallel
earthquake generation. In addition and more importantly,
cracks, where stress-induced changes of shear-wave splitting
earthquakes are the result of the release of stress, not strain.
are principally observed. What is needed are crosswell obserStrain is the primary cause of stress, but strain can be released
vations between three wells below about 1 km in a relatively
aseismically by creep (and other processes such as APE, see
homogenous rockmass within a few tens of kilometers of the
below). To recognize that earthquakes are imminent requires
vulnerable site. This would need one vertical receiver well
that stress is monitored at close to seismic generation depths
in the crust. Two recent developments in understanding and
and two deviated source wells.
Even this would predict only the time and magnitude,
instrumentation make such monitoring feasible.
but it would reliably indicate that stress was building up and
A new understanding of rock deformation shows that
that a larger earthquake (or in some areas an eruption) would
the splitting of seismic shear waves observed in almost all
happen sometime in the future. The longer the build-up
crustal rocks is controlled by the same stress parameters that
continued the larger would be the potential earthquake. We
control deformation. The (prefracturing) deformation
have called this procedure stressforecasting and would argue
mechanism of rock is the result of anisotropic poro-elasticity
that such a crescendo of increasing alarm could best mitigate
(APE) with fluid migration along pressure gradients between
hazard to life and property by giving time to prepare evacuaneighboring grain-boundary cracks and low-aspect ratio
292 SeismologicalResearchLetters Volume70, Number3 May/June1999
tion procedures, service shut-downs, and property safeguards. With a large earthquake expected, it would also
stimulate the search for and more realistic interpretation of
short-term precursory phenomena.
Until recently such monitoring would have been technically impossible, but suitable borehole recording and shearwave sources systems for such stress-forecasting sites are now
being developed. Drilling deep deviated wells in preferably
crystalline rock would be extremely expensive (cheaper, more
simple, source-to-receiver geometries have now been identified), but once the monitoring site had been established, the
monitoring itself would be comparatively cheap, depending
on the frequency of sampling required. It would be an appropriate insurance policy for a sufficiently vulnerable site.
Note that stress forecasting escapes the impossibility-ofearthquake-prediction conundrum because shear-wave splitting is examining the effects of stress on the rockmass and is
independent of the source of the impending earthquake.
Stress forecasting merely assesses when the increase of stress
in the rock will reach fracture criticality.
ADDENDUM
There has now been a successful "stress forecast" earthquake
based on the changes in shear-wave splitting observed at
three stations in a 70 km line in Iceland. Routine observations of changes in shear-wave splitting in one area of southwest Iceland allow the level of fracture criticality to be
inferred. In October 1998, it was recognized that the behavior of shear-wave splitting, at two stations 40 km apart, indi-
cated an increase of stress that was approaching fracture
criticality in real time before the earthquake had occurred.
Consequently, preliminary stress forecasts were issued to the
Icelandic National Civil Defense Committee on 27 and 29
October. On 10 November 1998, it was recognized that a
third station also showed changes in shear-wave splitting, and
a final stress forecast was issued that there would be a M_> 5
earthquake soon or a M _> 6 earthquake before the end of
February 1999. Three days later, on 13 November, a M = 5
earthquake was reported 2 km from the central station.
The uncertainty in the forecast is because both inferred
rates of increase and inferred levels of fracture criticality are
subject to errors due to scatter of the data, so at best an earlysmaller-magnitude to later-larger-magnitude window can be
defined. Industrial seismology suggests that controlledsource measurements would substantially improve the accuracy of such estimates.
Note that variations in shear-wave splitting can be used
to stress-forecast the time and magnitude of future large
earthquakes but cannot identify the location of the earthquake focus within the large rock volume where stress builds
up and variations in shear-wave splitting are observed. However, once the earthquake had been stress-forecast, Ragnar
Steffinsson of the Icelandic Meteorological Office correctly
predicted the fault on which the earthquake was to occur
from local geophysical and geological analyses. El
Stuart Crampin
Department of Geologyand Geophysics
University of Edinburgh
[email protected]
Seismological ResearchLetters Volume70, Number3 May/June1999 293