← Back to Sessions

Unsupervised Clustering of Cryoseismic Events Recorded by Distributed Acoustic Sensing at Rhonegletscher, Switzerland

Description: 

Higher fidelity monitoring and better quantification of mountain glacier processes remain a key priority as climate change transforms glaciated regions into landscapes of rocks, debris, ice-debris complexes, and glacial lakes. Our capability to monitor these highly dynamic and hazardous conditions for local communities remains limited by a lack of data at high spatial and temporal resolution. Distributed Acoustic Sensing (DAS) provides a unique opportunity to capture detailed spatiotemporal variations within and around a glacier. We focus on a DAS experiment from Rhonegletscher (Switzerland), during which we recorded data continuously for one month in 2023 using a 9 km fiber-optic cable on the ice surface. This deployment extended from the accumulation zone to near the glacier terminus, providing a unique perspective on glacial processes across surface balance regimes. The data captured environmental noise (e.g., wind, precipitation events) and acoustic signals from a variety of glacial processes, including surface fracturing and basal icequakes. DAS monitoring results in large data volumes, rendering manual event picking and classification as well as real time data telemetry practically unfeasible. Instead, we used automated processing tools and unsupervised clustering for event detection and classification with the goal of generating a comprehensive, classified catalog of cryoseismicity in the Rhonegletscher. We generated a low-dimensional feature representation of the data from the automated processing tools and used this data as input for an unsupervised clustering algorithm. We compared the clusters’ signal characteristics and spatiotemporal distribution with potential forcing mechanisms, including comparison with environmental data such as temperature and wind to assist in the identification or correlation of source processes. Our automated pipeline for DAS data analysis provides a new, high-resolution perspective on Rhonegletscher's response to short-term environment forcing and a pathway for leveraging the massive data volumes produced by DAS installations for real-time monitoring of alpine environments.

Session: Applications and Discoveries in Cryoseismology Across Spatial and Temporal Scales [Poster Session]

Type: Poster

Date: 5/2/2024

Presentation Time: 08:00 AM (local time)

Presenting Author: Rachel

Student Presenter: Yes

Invited Presentation: 

Authors