Automatic Wave Group Identification on Deep Seismic Refraction Data Using SMF Clustering

Abstract

This letter proposes algorithms for the automatic identification of direct (Pg) and head (Pn) wave groups from first break readings of a deep seismic refraction profile. The process employs a split-and-merge fuzzy iterative prototype clustering technique to extract line segments from first arrivals. A constant velocity value is used as a threshold to identify Pg and Pn groups. Synthetic and real data were used to evaluate the algorithm with satisfactory identification results.