The
San Andreas Fault (SAF) passes through the small town of Parkfield, California,
which is situated roughly halfway between Los Angeles and San Francisco.
Parkfield has experienced strong (at least M6) earthquakes six times
between 1857 and 1966. These quakes, have an average repeat interval of 22 years
(24, 20, 21, 12 and 32 years). Excluding
the larger and more extensive 1857 earthquake, they have all occurred on almost
exactly the same part of the fault. Furthermore,
the 1934 and 1966 quakes have very similar-looking seismographs, and each was
preceded by a M5 foreshock 17 minutes before the main shock.
Another similar earthquake was expected to occur at Parkfield by around
1987, but it still hasn’t happened, and the gap is now 36 years.
(For
more information on the history of earthquakes at Parkfield see: http://www.johnmartin.com/earthquakes/eqpapers/00000075.htm)
In the mid
1980s the USGS and several California universities initiated an intensive
seismic monitoring program at Parkfield. The
program now includes the following instrumentation:
12 creep
meters (to measure slow aseismic slip on the fault) |
2
electronic distance measurement instruments (to monitor displacement) |
12 GPS
stations (to monitor displacement) |
8
dilatational strain meters (to assess strain build-up in rocks) |
3 tensor
strain meters (to assess strain build-up in rocks) |
12
short-period seismometers |
10
bore-hole seismometers |
30
strong motion sensors (to measure the ground motion associated with a
large earthquake) |
a 2.2 km
deep borehole with various instrumentation |
a
proposed 4 km deep borehole with various instrumentation |
Amongst numerous other studies, earth scientists are monitoring water levels in wells and analyzing data from satellites to assess ongoing ground displacement.
(For more information on the research at Parkfield see: http://www.scec.org/instanet/01news/es_abstracts/langbeinES1.pdf)
This
unparalleled research effort is being conducted for two main reasons.
Firstly, the relatively simple geometry of the SAF at Parkfield allows
for a clear understanding of strain accumulation and release on the fault.
Secondly, the apparent regularity of the historic earthquakes at
Parkfield makes this an ideal site for testing the “time-predictable
recurrence model” developed in the 1980s (Shimazaki and Nakata, 1980).
Some of the
data gathered at Parkfield over the past few decades have been recently analyzed
by geophysicists from Stanford University. Their goal is to understand why there is now a 36-year gap
between major earthquakes at Parkfield. Murray
and Segall (2002) have estimated the rate of strain accumulation on the
Parkfield segment of the SAF, and they conclude that the most of the strain
released by the 1966 quake had re-accumulated by 1981, and that there is a 95%
probability that another large quake should have occurred by 1987.
In carrying out this analysis Murray and Segall recognized that some of
the strain at Parkfield could have been relieved by the nearby M6.5 Coalinga
quake of 1983, and that this could have delayed Parkfield by about 2 years.
On the other hand, they also calculate that two small earthquakes in the
Parkfield area in 1992 and 1994 (around M4) actually increased strain on the
Parkfield rupture zone, essentially countering the delaying effect of the
Coalinga quake.
The only
explanation offered by Murray and Segall is that local variations in pore-water
pressure may have affected the tendency for failure on the Parkfield segment –
although they have no means of measuring this parameter.
Murray and
Segall go on to argue that the 36 years worth of strain that has now accumulated
at Parkfield is substantially more than that which had accumulated prior to the
previous six large earthquakes, and therefore if the segment fails soon (eg. in
2002 or 2003) the resulting quake will likely have a magnitude between 6.6 and
6.9 – which would be significantly more damaging than any of the past five
Parkfield earthquakes.
References
Murray
J and Segall P, Testing time-predictable recurrence by direct measurement of
strain accumulation and release, Nature, V. 419, p. 287-291
(September 2002).
Shimazaki K and Nakata T, Time-predictable recurrence model for large earthquakes, Geophysical Research Letters, Vol. 7, P.279-282 (1980)