Stress Orientation Confidence Intervals from Focal Mechanism Inversion

TL;DR

This paper evaluates the accuracy of confidence intervals in stress orientation from focal mechanism inversion, demonstrating that proper statistical methods can yield reliable uncertainty estimates and validate the method's effectiveness.

Contribution

It clarifies the reasons behind previous overestimation of confidence intervals and shows how directional statistics improve the assessment of stress field homogeneity.

Findings

Proper perturbation methods produce accurate confidence intervals.

Bias in synthetic data affects confidence interval width.

The method is validated as effective with appropriate statistical analysis.

Abstract

The determination of confidence intervals of stress orientation is a crucial element in the discussion of homogeneity or heterogeneity of the stress field under study. The error estimates provided by the grid search method Focal Mechanism Stress Inversion of Gephart and Forsyth (1984) have been shown to be too wide but the reasons for this failure have escaped elucidation. Through the use of directional statistics and synthetic focal mechanisms, I show that the grid search methodology does yield appropriate uncertainty estimates. The direct perturbation of the synthetic focal mechanisms introduces bias which leads to confidence intervals which become increasingly too wide as the amount of perturbation increases. The synthetic data also show at what point the method fails to overcome this bias and when confidence intervals will be too wide. The indirect perturbation of the focal mechanisms by perturbing the generating deviatoric stress tensor generates synthetic data devoid of bias. Inversion of these data sets yields correct confidence intervals. The Focal Mechanism Stress Inversion method is vindicated as a highly effective method, and with the use of appropriate directional statistics, its results can be assessed and homogeneity or heterogeneity of the stress field can be discussed with confidence.