Chapter 5. Solid-Earth Sciences
Earthquake Prediction Research
 Servicing a portable seismic station in eastern Nepal. | Much of the innovative research at CIRES on prediction of earthquakes
was led by Max Wyss, who left CU in 1991 to take up a position at the
University of Alaska, Fairbanks. Variations in numerous geophysical parameters
were examined for precursory patterns using global and regional
(Hawaii, California, Turkey, Alaska) observations. In particular,
Wyss and R. E. "Ted" Habermann, (who became and remains a NOAA
CIRES fellow) focused attention on seismicity rate decreases (quiescence)
prior to large earthquakes. As a crucial part of these studies, they developed
quantitative tools for recognizing and understanding apparent seismicity
variations caused by changes in seismic networks and data
processing procedures. They demonstrated that many seismicity rate
variations proposed as precursors are artificial and can be explained by
common network changes. Wyss also investigated crustal uplift and subsidence
as revealed by tide-gauge data and systematic variations in
b-value as plausible precursors.
Kisslinger and Engdahl also carried out prediction studies, mostly
based on observations with the Central Aleutians Seismic Network. A
number of phenomena that have been suggested as precursory to an imminent
earthquake were investigated. These included systematic rotation
of focal mechanisms of local events before a stronger event, the isolation
of possible asperities on the subduction thrust surface on the basis of the
distribution of background seismicity and stress drops, variations in attenuation
based on measurements of coda-Q, investigations of localized
changes in seismic body-wave velocities prior to a strong event, and seismic
quiescence. Of these, quiescence as detected with the data from an
adequate local network emerged as most promising, but no consistently
successful precursor has been identified.
Recent work by Rundle and others in the areas of non-linear continuum
mechanics, chaotic behavior and the physics of complex systems
may lead to advances in prediction technology, or at least to a better understanding
of approaches to overcoming the difficulties. Rundle is approaching
earthquake forecasting through the analysis of neural
networks. Projects by his group have emphasized the development of
techniques for understanding the space-time patterns and correlations
that appear in many high-dimensional complex non-linear systems, including
patterns of seismicity. The patterns are quantified in terms of
eigenvectors of defined autocorrelation operators.
Measurements of tilt in the Long Valley Caldera, California,
and of creep on the Hayward fault in the San Francisco
Bay region, both directed by Bilham, are now part of early
warning systems in these active zones. Data from these
CIRES monitoring arrays are updated every 10 minutes on
the following web site: http://quake.wr.usgs.gov/research/deformation/monitoring/index.html
Next > Investigations of the Earth's Interior
|