White Box Comparison of Different Algorithmic Approaches to Event Detection and Association

Session: Leveraging Advanced Detection, Association and Source Characterization in Network Seismology [Poster]
Type: Poster
Date: 4/30/2020
Time: 08:00 AM
Room: Ballroom
Description: 

Seismic event detection and association algorithms generally consist of two steps: 1) generation of a set of event hypotheses and 2) deciding which event hypotheses and associations to keep. Various algorithms take different approaches to these steps and are often evaluated from a “black box” perspective, i.e. solely on the quality of the resulting bulletin that is produced. Although useful, this type of assessment doesn’t always uncover why one algorithm performs better than another. Instead, deeper insight is needed to realize significant improvements in automatic bulletin quality. In this presentation, we assess different detection and association algorithms from a “white box” perspective by analyzing the above two steps individually and relating how each of these steps affect the resulting bulletins. All results are evaluated against a reference bulletin built by careful manual analysis of 3-component data from the University of Utah Seismic Network.

Presenting Author: Stephen L. Heck

Authors