Towards the systematic reconnaissance of seismic signals from glaciers and ice sheets – Part A: Event detection for cryoseismology

Abstract

Abstract. Cryoseismology is a powerful toolset for progressing the understanding of the structure and dynamics of glaciers and ice sheets. It can enable the detection of hidden processes such as brittle fracture, basal sliding, transient hydrological processes, and calving. Due to the diversity and often low signal-to-noise levels of glacier processes, the automated detection of seismic events caused by such processes can pose a challenge. We present a novel approach for the automated detection of events in glacier environments, the multi-STA/LTA algorithm, with a focus on capturing the many signal types recorded on ice sheet margins. This develops the use of approaches that use the ratio between short and long time averages (sta,lta) of signal amplitude as the means of event detection. Implemented in the open source and widely used ObsPy python package, the algorithm constructs a hybrid characteristic function from a set of sta, lta pairs. We apply the multi-STA/LTA algorithm to data from a seismic array deployed on the Whillans Ice Stream (WIS) in West Antarctica (austral summer 2010–2011), to form an event catalogue. The new algorithm compares favorably with standard approaches, yielding a diversity of seismic events, including all previously identified stick-slip events (Pratt et al., 2014), teleseisms, and other noise-type signals. We investigate a partial association of seismicity with the tidal cycle, and a slight association with ice temperature changes of the Antarctic summer. The new algorithm and workflow has the potential to yield systematic catalogues for further cryoseismology studies: conventional glacier seismology, and those tailored to pattern recognition by machine learning.

Environment‐Modulated Glacial Seismicity Near Dålk Glacier in East Antarctica Revealed by Deep Clustering

“…For the detection of impulsive signals, we employ a short-term average/long-term average (STA/LTA) algorithm (Allen, 1978) with the STA window set at 0.5 s and the LTA window set at 30 s. We use a trigger threshold of 10 and a detrigger threshold of 3. We choose a 0.5 s STA window here to identify impulsive and short-duration signals, while one should note that the choices of STA/LTA parameters might lead to different detection results (Latto et al, 2023). Signal detection is performed at each station, and the following analyses are performed on the individual detected signals.…”

Towards the systematic reconnaissance of seismic signals from glaciers and ice sheets – Part B: Unsupervised learning for source process characterisation

“…Cryoseismic events occurring during the 2010-11 instrument deployment on the WIS were identified (Latto et al, 2023) using an implementation of the multi-STA/LTA algorithm (Turner et al, 2021). Two catalogues are produced, a reference event catalogue and a trace catalogue, which list event information per seismometer.…”

“…A semi-automated approach is therefore desirable. The first step to automation, robust event detection, is addressed in Latto et al (2023), which describes an algorithm designed for diverse, low signal-to-noise microseismicity from glaciers. The next step is event analysis, where data-driven approaches present an appealing way forward (Bergen and Beroza, 2019).…”

“…A repeatable approach will afford the possibility to make comparisons between years for a given glacier environment, and compare different localities. The two companion papers presented as Latto et al (2023), and herein, respectively demonstrate a systematic workflow in 1) building a catalogue and 2) using a simple clustering method as a reconnaissance tool to better understand glacier dynamic and hydrological processes.…”

“…We present a workflow for employing semi-unsupervised cluster analysis to inform investigations of the processes occurring in glaciers and ice sheets. In this demonstration study, we make use of a seismic event catalogue previously compiled for the Whillans Ice Stream, for the 2010-2011 austral summer (outlined in companion paper, Latto et al, 2023). We address the challenges of seismic event analysis for a complex wavefield by clustering similar seismic events into groups using characteristic temporal, spectral, and polarization attributes of seismic time series with the k-means++ algorithm.…”