地球与行星物理学术报告通知（202623期）- Qinya Liu

报告人简介：

Dr. Qinya Liu is a professor of geophysics and the Teck Chair in Exploration Geophysics at the University of Toronto. She received her B.S. in Geophysics from the Department of Earth and Space Sciences and Department of SCGY, University of Science and Technology of China in 2000 and her Ph.D. in Geophysics from the California Institute of Technology in 2006. Prof. Liu was an assistant scientist at Caltech (2006-2007), and a post-doctoral researcher at the Scripps Institute of Oceanography, University of California, San Diego (2007–2008), before joining the University of Toronto in 2008. Her research focuses on full waveform inversion of dense array data for lithospheric structures, as well as advanced seismic source characterizations. Prof. Liu has published over 90 peer-reviewed articles in top-tier journals, including Science, Earth and Planetary Science Letters, Geophysical Research Letters, Journal of Geophysical Research: Solid Earth. She currently serves as an Associate Editor for Journal of Geophysical Research: Solid Earth

报告内容摘要：

Microseismic monitoring is of critical importance for characterizing induced earthquakes and understanding fault activation processes resulting from anthropogenic subsurface operations such as hydraulic fracturing. However, processing these specialized microseismic datasets with low-magnitude events and low signal-to-noise ratio (SNR) waveforms can be challenging with traditional methods. In this study, we present advanced machine-learning-based workflows for source characterization of microseismic datasets. Specifically, we apply deep-learning (DL) phase pickers to detect weak microseismic P and S arrivals through data-optimization strategies, and fine-tune a region-specific DL picker by integrating multi-model consensus and noise augmentation based on the public Tony Creek Dual Microseismic Experiment (ToC2ME) data. This targeted model adaptation drastically reduces the catalog magnitude of completeness and maximizes event detection. The DL-strategy is also applied to a surface-array recording of induced earthquakes in northern Montney play in the Western Canada Sedimentary Basin, enabling enhanced catalogs and high-resolution source-parameter evaluation. Using the open-source Moment Tensor Uncertainty Quantification (MTUQ) framework, full moment tensors were inverted for 567 microseismic events near an induced Mw 4.6 earthquake sequence.  This unified open-source workflow provides an efficient and robust microseismic monitoring solution for both industrial operations and regulatory agencies.