← Back to Sessions

Glitch Finder General: Building a Glitch Detection Scheme Using the Matrix Profile

Description: 

Seismic data is a valuable tool that can be used to sample a planetary body’s interior and reveal information about its tectonic activity. However, factors such as a limited number of seismic stations, noise, and even anomalous signals can complicate the task of seismic event detection. Factors such as noise or glitch-like signals can conceal or disrupt event waveforms, making it difficult to identify and characterize potential seismic events. In this work, we attempt to build a framework for “glitch” detection using the Matrix Profile (MP). The MP is a fast template matching method that does not require a priori templates, can be more efficient than traditional autocorrelation approaches, and unlike machine learning approaches, does not require training. Furthermore, previous work (Shakibay Senobari 2024) has shown that the MP is capable of detecting events below the noise level. For this reason, we apply the MP to the InSight seismic dataset to build a catalog of glitches for the martian seismic dataset.

The InSight seismic data set has a high signal-to-noise ratio that severely inhibits the detection of both seismic events and glitches, leading to the spatial and temporal clustering of these signals. Glitches can be concealed within true event waveforms and can even be misinterpreted as seismic signals, making their identification important. Before applying the MP to the full dataset, we applied the MP to the time periods between sols 246-259, 1009-1043, and 1211-1241. We use the Pan Matrix Profile (PMP) to select a set of subwindow lengths, which allows us to detect a variety of different signals. Preliminary results show we are able to identify glitches based on their high degree of similarity (MP correlation coefficient r≥0.99) and their “boxcar” shape in the resulting MP, even during noisier periods. As part of our future work, we will continue to curate this glitch catalog and refine our glitch detecting framework for application to other seismic datasets. Lastly, we will finetune the MP for seismic event detection in the martian seismic dataset.

Session: New Frontiers in Seismic Observations and Modeling with Innovative Methods and Emerging Data on Earth and Other Planets [Poster]

Type: Poster

Date: 4/17/2026

Presentation Time: 08:00 AM (local time)

Presenting Author: Norma A. Contreras

Student Presenter: Yes

Invited Presentation: 

Poster Number: 111

Authors