Journal of Natural Disaster Science

HYPOCENTER DETERMINATION USING A GENETIC ALGORITHM

Yoko MURAMATSU
and
Ichiro NAKANISHI
Department of Geophysics, Kyoto University

(Received 11 June, 1997 and in revised form 27 February, 1998)

Abstract

A genetic algorithm is applied to hypocenter location. One problem in the locating of hypocenters is how to get rid of local solutions which exist in the parameter space. The problem is more serious when the hypocenter is outside of the seismic network because the geometry of the stations is biased. With genetic algorithms, there is little danger of dropping into local solutions or relying too much on the initial guess. Less calculation is needed to reach the final solution than with other grid search methods. First numerical experiments are made with synthetic data sets, and the solutions compared with those obtained with a matrix inversion method. Next, the method is used with real data sets. The hypocenter is determined accurately and stably by this method even when it is outside the seismic network. The amount of calculation required to obtain the result for a hypocenter far from the network is somewhat more than that for the result inside the network. Our solution can be considered the final solution, or at least, a good initial guess for the matrix inversion method.

Key words

hypocenter location, genetic algorithm (GA), L1-norm residual, non-linear inversion