Earthquake Location With Grids -- Not Just for 3-D Velocity Models

D. H. von Seggern

The traditional method of earthquake location is known to be
unstable for events near the fringe or outside of a network 
(Buland, 1985).  Eigen-analyis of the matrix connecting the 
data with the solution parameters for simple three or four 
station cases reveals the problem of ill-conditioned 
matrices.  Grid search methods of location overcome this 
inherent instability and are robust in all cases.  Although 
most beneficial in applying 3-D velocity models (Nelson and 
Vidale, 1990), grid search methods are also beneficial in 
locating events with simple 1-D models.  This work applies 
grid location to the catalog of events produced by the 
Southern Great Basin Digital Seismic Network around Yucca 
Mountain, Nevada, over the past two years.  With travel-time 
grids based on a simple 1-D model, improved determination of
depth, lower rms residuals, and clearer spatial patterns of 
events have resulted, especially along the network boundary.
The grid method allows for easy computation of a realistic 
confidence region which is more representative of the 
uncertainty than the normal confidence ellipsoid computed in 
the traditional method of location.  The grid search method 
is comparable in speed to traditional algorithms, but at the 
cost of more memory utilization; and it will always produce 
a solution.