Real-Time Processing

We are developing real-time analysis techniques that will enable rapid determinations (within minutes) of deformation following major earthquakes to complement seismological information. We use GAMIT/GLOBK processing techniques to estimate independent hourly solutions at the several cm-level horizontal precision and during the past year established an extension of the REDI system where estimates of postseismic positions are attempted when 10 minutes of data become available following an earthquake (Murray et al., 2002). We currently process 1-hour data batches available within 20 minutes of measurement from more than 20 continuously telemetered BSL and other stations providing hourly data. The hourly solutions have higher scatter than the 24-hour solutions (3-10 mm in the horizontal and 10-30 mm in the vertical), but our simulations suggest that displacements 3-5 times these levels should be reliably detected, and that the current network should be able to resolve the finite dimensions and slip magnitude of a M7 earthquake on the Hayward Fault.

We are testing a relatively new component of GAMIT that uses Kalman filtering techniques and improved ambiguity resolution methods to provide higher-precision kinematic positions. We are testing these rapid processing techniques to estimate higher frequency 1-Hz GPS displacements, which have been used to detect surface from large earthquakes and can potentially add valuable information about the seismic source. We are also developing methods to rapidly estimate finite-source models from coseismic GPS displacements and to use these models to predict strong-ground motions and improve ShakeMap depictions of these motions as rapidly as possible after an earthquake. The first step of this project has been the development of a new methodology to improve prediction of strong ground motion from a simple uniform-slip geodetic model of the source. This methodology is based on a simple assumption that the large slip should take longer to terminate. We use well known scaling relations between stress drop and slip velocity to develop a spatio-temporal slip model. We have tested this model on the 1992 Northridge earthquake and found that the predicted ground motions agree well with observed motions and with other models derived from combinations of seismic and geodetic data (Rhie et al., 2005).

Berkeley Seismological Laboratory
215 McCone Hall, UC Berkeley, Berkeley, CA 94720-4760
Questions or comments? Send e-mail: www@seismo.berkeley.edu
© 2005, The Regents of the University of California