2024 Undergraduate Research Showcase

Infrasound-Based Avalanche Event Classification: Leveraging Deterministic and Visual Techniques for Machine Learning Dataset Creation

Document Type

Student Presentation

Presentation Date

4-19-2024

Faculty Sponsor

Dr. Jeffrey B. Johnson

Abstract

This study focuses on the classification of avalanche events using a combination of deterministic methods and visual identification techniques applied to infrasound data collected in Little Cottonwood Canyon, Utah, from March 21, 2023, to April 15, 2023. Previous studies have predominantly relied on deterministic methodologies for avalanche event classification. In this study, we augment these methods by incorporating visual identification of events through correlograms, leading to the creation of a more accurate, labeled dataset. The dataset encompasses various events recorded during the specified period, including avalanches and other infrasound phenomena. To facilitate the creation of the labeled dataset, we developed a MatLab script that leverages deterministic classification results alongside visual identification. This script streamlines the process of distinguishing avalanche events from other occurrences, enhancing the efficiency and accuracy of dataset creation. The resulting labeled dataset serves as a valuable resource for future research endeavors, particularly in the development and training of machine learning models aimed at automating avalanche event identification.

This document is currently not available here.

Share

COinS