Uncertainty Analysis in Seismic Event Location

Abstract

Uncertainty in event locations derived from seismic data is caused by errors in the arrival times of picked phases, misidentification of seismic phases, and errors in the travel-time model used in the location process. The event mislocation induced by these error sources is affected by the number and spatial distribution of stations that record an event. This project is developing a statistical framework and computational techniques for accurately analyzing event location uncertainty. Our statistical approach is based on a maximum-likelihood framework, which defines an optimal location estimate to be that maximizing a likelihood function, and derives confidence regions in terms of hypothesis tests applied to likelihood ratios. An appropriate likelihood function is prescribed in terms of a probabilistic model of the various types of errors in seismic data. With appropriate computational tools, it is possible to implement a general class of error models that allow for non-Gaussian distributions, spatially correlated errors in travel-time tables, and other complexities that conventional location algorithms do not handle. Additionally, the assumption of local linearity of the forward problem (travel-time vs. location) can be avoided. We are developing such computational tools based on grid-search and Monte- Carlo simulation techniques. We have implemented our statistical formulation in a general event location algorithm that finds optimal location estimates from arrival time, slowness and azimuth measurements for regional and teleseismic phases, and computes the non-elliptical confidence regions which follow from a general error model and nonlinear analysis.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Oct 01, 2001
Accession Number
ADA529649

Entities

People

  • M. Nafi Toksöz
  • William Rodi

Organizations

  • Massachusetts Institute of Technology

Tags

Communities of Interest

  • Energy and Power Technologies

DTIC Thesaurus Topics

  • Algorithms
  • Calibration
  • Computational Science
  • Data Science
  • Distribution Functions
  • Earthquakes
  • Gaussian Distributions
  • Grids
  • Information Science
  • Monte Carlo Method
  • Probabilistic Models
  • Probability
  • Random Variables
  • Simulations
  • Standards
  • Statistical Algorithms
  • Statistical Analysis

Readers

  • Seismology
  • Statistical inference.
  • Systems Analysis and Design