Grid-Search Techniques for Seismic Event Location
Abstract
We are developing a new algorithm for seismic event location to address particular problems of importance in monitoring the Comprehensive Nuclear-Test-Ban Treaty, including the assessment of location accuracy and increasing the number of seismic phases used for location. Our approach is formulated within a maximum-likelihood estimation framework and implemented numerically with grid-search and Monte Carlo techniques. It obtains globally optimal hypocentral estimates and non-ellipsoidal confidence regions that do not depend on the usual assumption of local linearity of the forward problem. The approach accommodates such complexities as 3-D travel time tables, non-Gaussian data errors, and general types of parameter constraints. Specific problems we are addressing are: (1) extension of our uncertainty analysis to account more realistically for the effects of modeling errors (errors in travel time tables), which traditionally have not been distinguished from random picking errors in the determination of location confidence regions; (2) incorporation of azimuth and slowness data, in addition to arrival times, into the maximum-likelihood framework; and (3) integration of phase association into the event location algorithm in order to perform these tasks simultaneously with a grid-search approach. We present examples of our event location algorithm applied to data from the International Monitoring System, illustrating the effects of nonlinearity on hypocentral confidence regions determined from sparse data sets.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 2000
- Accession Number
- ADA529939
Entities
People
- M. Nafi Toksöz
- William Rodi
Organizations
- Massachusetts Institute of Technology