SC5.16

QuakeMigrate: an open-source software package for automatic earthquake detection and location
Co-organized by CR8/SM9
Convener: Tom WinderECSECS | Co-convener: Conor BaconECSECS

QuakeMigrate is a new, open-source software package for automatic earthquake detection and location (https://github.com/QuakeMigrate/QuakeMigrate). Our software provides a means for seismologists to extract highly complete catalogues of microseismicity from continuous seismic data, whether their network is installed at a volcano, plate-boundary fault zone, on an ice shelf, or even on another planet. Rather than traditional pick-based techniques, it uses a migration-based approach to combine the recordings from stations across a seismic network, promising increased robustness to noise, more accurate hypocentre locations, and improved detection capability. Cloud-hosted Jupyter Notebooks and tutorials (https://mybinder.org/v2/gh/QuakeMigrate/QuakeMigrate/master) provide an overview of the philosophy and capabilities of our algorithm, and in this session we intend to provide a more hands-on introduction, with a focus on providing a general understanding of the considerations when applying a waveform-based algorithm to detect and locate seismicity.

QuakeMigrate has been constructed with a modular architecture, to make it flexible to use in different settings. We will demonstrate its use in detecting and locating basal icequakes at the Rutford Ice Stream, Antarctica, volcano-tectonic seismicity during the 2014 Bárðarbunga-Holuhraun and 2021 Reykjanes/Fagradalsfjall dike intrusions, and aftershocks from a M5 tectonic earthquake in northern Borneo, which was recorded on a sparse regional seismic network. In each case we will discuss the reasoning behind parameter selections, and the key factors in maximising detection sensitivity while minimising computational cost. We will end the session by exploring sample datasets provided by attendees, with interactive involvement as we tune parameters and use the comprehensive array of automatically generated plots to take a preliminary look at unseen data.