14:15 〜 14:30
[SCG45-27] Deep Learning-Driven Seismicity Catalog of the Cascadia Region
キーワード：Earthquake, Cascadia , Seismicity, Deep Learning
The Cascadia region, particularly the Cascadia subduction zone, is well known for its potential to generate destructive megathrust earthquakes. A comprehensive and complete seismicity catalog is crucial for enhancing our understanding of regional tectonics and assessing seismic hazards over both short- and long-term timescales.
Using advanced deep-learning pickers, we construct a detailed catalog of offshore seismicity based on seismic data from the four-year Cascadia Initiative amphibious array and three additional temporary Ocean Bottom Seismometer (OBS) deployments. Continuous OBS and land-based seismic data were analyzed using deep-learning phase pickers, OBSTransformer and EqTransformer, respectively. Preliminary earthquake locations were determined through three association methods—REAL, Gamma, and PyOcto—considering only events identified by all three. These locations were further refined using absolute location methods and double-difference relative relocation techniques. To enhance seismicity completeness, template matching was applied, particularly in regions with low seismic activity, such as the Juan de Fuca Plate and the Cascadia margin.
Our catalog consists of approximately 24k earthquakes, with magnitudes ranging from -2.81 to 5.21 and a completeness magnitude of 1.5. This comprehensive and complete earthquake catalog enhances our understanding of regional tectonics. The characteristics and implications of these newly detected earthquakes will be discussed, with a focus on their underlying mechanisms.
Using advanced deep-learning pickers, we construct a detailed catalog of offshore seismicity based on seismic data from the four-year Cascadia Initiative amphibious array and three additional temporary Ocean Bottom Seismometer (OBS) deployments. Continuous OBS and land-based seismic data were analyzed using deep-learning phase pickers, OBSTransformer and EqTransformer, respectively. Preliminary earthquake locations were determined through three association methods—REAL, Gamma, and PyOcto—considering only events identified by all three. These locations were further refined using absolute location methods and double-difference relative relocation techniques. To enhance seismicity completeness, template matching was applied, particularly in regions with low seismic activity, such as the Juan de Fuca Plate and the Cascadia margin.
Our catalog consists of approximately 24k earthquakes, with magnitudes ranging from -2.81 to 5.21 and a completeness magnitude of 1.5. This comprehensive and complete earthquake catalog enhances our understanding of regional tectonics. The characteristics and implications of these newly detected earthquakes will be discussed, with a focus on their underlying mechanisms.