Abstract: Detection of small earthquakes in urban intraplate regions is highly challenging due to its dynamic noise levels, sparse seismicity, and poor instrument coverage. Traditional methods like the STA/LTA ratio and template matching are poorly suited for such environments, often misclassifying earthquakes with low signal-to-noise ratios (SNRs). To investigate a potential case of human-induced seismicity in the Chicago area, we applied an existing deep learning model and a newly developed Random Forest model to detect small earthquakes in an industrial corridor of the Chicago area. We also developed an unsupervised learning workflow to detect and cluster non-earthquake seismic events, producing a labeled dataset of 1000+ events that can be used to train future detection algorithms. We discuss the development of these two machine learning workflows, their performance in the Chicago area, and future directions for seismic monitoring in noisy, built environments.

Bio: Ann is a postdoctoral researcher in the Department of Earth, Environmental, and Planetary Sciences at Northwestern University. She recently defended her PhD dissertation at Northwestern on urban intraplate seismology and geoscience education in the Chicago area. Her research focuses on developing data-driven methods to detect and characterize seismic events in noisy, data-sparse environments. She is passionate about community engagement in STEM, having over four years of experience in organizing workshops, conferences, and internship programs that provide hands-on STEM experiences for the larger community.