← Back to Sessions

Room: Holiday Ballroom 1

Date: 4/15/2025

Session Time: 4:30 PM to 5:45 PM (local time)

Building and Decoding High-resolution Earthquake Catalogs With Statistical and Machine-learning Tools 

Recent application of advanced earthquake detection techniques such as template matching and machine learning (ML) have produced exponential growth in the quantity of earthquakes listed in next-generation high-resolution catalogs around the world. These improved catalogs can include relocated seismicity on the order of tens of thousands to a few million individual events, making it challenging to use standard analysis and modeling tools such as the Epidemic Type Aftershock Sequence (ETAS) model or the nearest neighbor algorithm for de-clustering, to extract key features and forecast seismicity. This session welcomes contributions on recent efforts to build high-resolution earthquake catalogs using waveform-based and ML methods such as transformers or foundation models. We also solicit presentations on  innovative methods to decode these high-resolution catalogs using statistical analyses and advancing our understanding of earthquake interactions, swarms, fault geometries or localization processes, as well as predictive modeling approaches including neural and Bayesian point processes, deep Gaussian process models, and other generative models. We especially encourage submissions that compare new results with benchmarks, e.g. with respect to standard catalogs, or to model benchmarks such as statistical ETAS models or physics-based models such as Coulomb Rate-and-State (CRS) models to forecast seismicity. 

Conveners 

Xu Si, Georgia Institute of Technology (xsi33@gatech.edu)
Maximilian J. Werner, University of Bristol (max.werner@bristol.ac.uk)
Shixiang Zhu, Carnegie Mellon University (shixianz@andrew.cmu.edu)

Oral Presentations