Enhancing rapid earthquake location by integrating streaming DAS virtual stations

Distributed acoustic sensing (DAS) on submarine fibre optic cables complements seismic networks by providing real-time information from the seabed over long range. The seismology community has demonstrated impressive new capabilities and processing techniques with DAS seismic recordings, and collaborations have been established to systematically deploy the instrumentation. We focus on the integration with existing earthquake monitoring systems and show that edge computing and streaming from the DAS interrogation of submarine cables enhances rapid earthquake location when processed in conjunction with data from terrestrial seismometer stations.

The widely used SeedLink protocol enables data transfer in near real-time and has support for many acquisition and analysis systems. The protocol is optimized for applications in seismology and for use with standard seismometer sensors. Recently, SeedLink streaming for DAS systems has been developed in the SeisComP framework. Since the DAS instrument data rate can be orders of magnitude larger than a typical seismometer, we introduce the DAS virtual station concept to expose pre-processed data in a sampling compatible with that of traditional seismic networks. To benefit from the dense spatial sampling along the cable, we implement array processing as edge computing on the DAS instrument server. This enables noise suppression and wave analysis techniques that would not be possible if a sparse subset of the DAS single component channels were streamed directly. For example, a 150 km cable can be exposed as 15 virtual stations spaced 10 km apart, with each station representing the denoised and decomposed landward and seaward propagating wave phases in different velocity intervals. The associated data rate will be comparable to 15 seismometers and suitable for streaming even over low bandwidth connections from the cable landing station.

Streaming of DAS virtual stations via SeedLink seamlessly facilitates the use of any analysis tools and workflows using existing standard seismological formats and services; this also facilitates data handling and curation in seismological data centres. The integration with real-time data from seismic network stations of similar spatial sampling can also be conducted within standard frameworks. To understand the impact of the integration, we consider using the virtual stations in near real-time processing as a complement for rapid earthquake location. The location accuracy enhancement from incorporating the sparsely sampled virtual stations along the submarine cable can be significant due to the extended azimuthal coverage of stations in the ocean. For early detection and location this achieves the most important benefit of the DAS cable sensing, and with the noise suppression possible from the dense sampling operating as edge computing. Further details might be required in post event analysis, in which case the full DAS dataset can be transferred.