Applications of Subspace Seismicity Detection in Antarctica
Monday, 15 December 2014
Subspace detection can improve event recognition by enhancing the completeness of earthquake catalogs and by improving the characterization and interpretation of seismic events, particularly in regions of clustered seismicity. Recent deployments of dense networks of seismometers enable subspace detection methods to be more broadly applied to intraplate Antarctica, where historically very limited and sporadic network coverage has inhibited understanding of dynamic glacial, volcanic, and tectonic processes. In particular, recent broad seismographic networks such as POLENET/A-Net and AGAP provide significant new opportunities for characterizing and understanding the low seismicity rates of this continent. Our methodology incorporates three-component correlation to detect events in a statistical and adaptive framework. Detection thresholds are statistically assessed using phase-randomized template correlation levels. As new events are detected and the set of subspace basis vectors is updated, the algorithm can also be directed to scan back in a search for weaker prior events that have significant correlations with the updated basis vectors. This method has the resolving power to identify previously undetected areas of seismic activity under very low signal-to-noise conditions, and thus holds promise for revealing new seismogenic phenomena within and around Antarctica. In this study we investigate two intriguing seismogenic regions and demonstrate the methodology, reporting on a subspace detection-based study of recently identified clusters of deep long-period magmatic earthquakes in Marie Byrd Land, and on shallow icequakes that are dynamically triggered by teleseismic surface waves.