
The use of propagation path corrections to improve seismic event location in western China
Author(s) -
Allen H. Cogbill,
Lee K. Steck
Publication year - 1998
Language(s) - English
Resource type - Reports
DOI - 10.2172/650323
Subject(s) - relocation , grid , aftershock , geodesy , earthquake location , seismology , nonlinear system , geology , amplitude , magnitude (astronomy) , induced seismicity , computer science , physics , quantum mechanics , astronomy , programming language
In an effort to improve ability to locate events in western China using only regional data, the authors have developed propagation path corrections to seismic travel times, and applied such corrections using both traditional location routines as well as a nonlinear grid search method. Thus far, they have concentrated on corrections to observed P arrival times. They have constructed such corrections by using travel time observations available from the USGS Earthquake Data Reports, as well as data reported by the ISC. They have also constructed corrections for six stations that are a part of the International monitoring System. For each station having sufficient data, they produce a map of the travel-time residuals from all located events. Large-amplitude residuals are removed by median filtering, and the resulting data are gridded. For a given source location, the correction at a particular station is then interpolated from the correction grid associated with the station. They have constrained the magnitude of the corrections to be {le} 3 s. They have evaluated the utility of the calculated corrections by applying the corrections to the regional relocation of 10 well-located Chinese nuclear tests, as well as a single, well-located aftershock in nearby Kyrgyzstan. The use of corrections having magnitudes > 2 s is troubling when using traditional location codes, as the corrections amount to a nonlinear perturbation correction, and when large may destabilize the location algorithm. Partly for this reason, the authors have begun using grid search methods to relocate regional events. Such methods are easy to implement and fully nonlinear. Moreover, the misfit function used to locate the event can very easily be changed; they have used L{sub 1}- and L{sub 2}-norm misfit functions, for example. Instances in which multiple local minima occur in a location problem are easily recognized by simply contouring or otherwise displaying the misfit function