← Back to Sessions

Combining Horizontal Strain DAS and Local Seismic Stations in a Full Waveform Attribute Stacking Detector/Locator Algorithm: Verification Test for the Thorbjörn, Iceland, 2020 Unrest Episode

Session: Recent Development in Ultra-Dense Seismic Arrays with Nodes and Distributed Acoustic Sensing

Type: Oral

Date: 4/20/2021

Presentation Time: 05:45 PM Pacific

Description: 

In recent years, the development of automatic routines for detecting and locating micro-earthquakes based on the full waveform has rapidly advanced the field of micro-seismic studies. Overall, the routines are very computationally intensive and do not reach their full potential until the seismic wavefield is sampled by ultra-dense sensor networks. Migration-based detector / locator techniques as for instance implemented in Lassie (Pyrocko) have demonstrated their robustness in a wide variety of applications in seismology. In this work, we have extended Lassie to efficiently combine linear ultra-dense sensor arrays with sparse seismological networks.

We use the seismicity unrest episode in the Svartsengi fissure swarm close to Mt. Thorbjörn, SW Iceland, which started in January 2020 and was still ongoing in at the time of writing in January 2021, producing more than 5 earthquake swarms comprising thousands of individual events each. We were able to combine local and regional seismic networks with 6 months recording of a 17 km long DAS cable with a channel resolution of 4 m. The kHz DAS data were downsampled to 200 Hz and stacked every 64 m.

We compare two different approaches in how to include DAS recordings into the stacking process: (1) by simple extraction of representative virtual single-component stations through pre-stacking and (2) by additionally exploiting information about the observed apparent slowness along the cable. Although our framework is already parallelized and suited to process real time data, we further improved the computational efficiency of Lassie by implementing the stacking algorithm on GPU architecture. The inclusion of DAS data significantly reduced the magnitude of completeness and improved the localisation of events.

We discuss the implementation and the processing of the joined dataset and evaluate the performance in comparison to the location with seismic data only, and DAS data only.

Presenting Author: Sebastian Heimann

Student Presenter: No

Authors