Focal mechanism estimation by classification

TL;DR

This paper introduces a support vector machine-based classification method for determining focal mechanism types and fault orientations from P-wave first motion data, avoiding prior assumptions about source mechanisms.

Contribution

It presents a novel non-parametric classification approach using spherical harmonic functions and correlation metrics for focal mechanism estimation.

Findings

Supports vector machine effectively classifies focal mechanisms.

Method distinguishes shear versus tensile dislocations.

Provides fault plane orientation estimates without prior assumptions.

Abstract

A classification technique for identifying focal mechanism type and fault plane orientation based on the polarity of P-wave "first motion" data is derived. A support vector machine is used to classify the polarity data in the space of spherical harmonic functions. The classification is non-parametric in the sense that there is no requirement to make a priori assumptions source mechanism. A metric of similarity potentially able to distinguish shear versus tensile dislocation without requiring estimation of the fault plane orientation is a natural consequence of this procedure. Going further, correlation functions between template source mechanism is derived, gives an estimate of fault plane orientation assuming a particular source mechanism.