Machine Learning for Rapid Magnitude and Hazard Characterization of Large Earthquakes with HR-GNSS

Project Description

If and when an earthquake has been characterized, can shaking and tsunami amplitudes be accurately forecast for critical locations (e.g. population centers or critical infrastructure sites) on short timescales that have utility for warning? This project builds on the research group’s work on large earthquake and tsunami forecasting, using machine learning, specifically M-LARGE.

Improvements will be made to FakeQuakes, a suite of algorithms that is used to generate scenario ruptures and to simulate realistic high-rate global navigation satellite system (HR-GNSS) data used to train machine learning (ML) algorithms, such as M-LARGE. To supplement M-LARGE, a new algorithm will be developed to detect earthquakes directly from the GNSS waveforms, independent of an external trigger, which can detect early onset signals in noisy GNSS waveforms and facilitate more rapid and accurate warnings to the public.

Project Lead

Diego Melgar, PhD, Assistant Professor, Department of Earth Sciences, University of Oregon

Additional Collaborators

Amanda Thomas, PhD, Associate Professor, Department of Earth Sciences, University of Oregon

Jiun-ting Lin, Graduate Student, NASA Earth & Space Science Fellow, University of Oregon

Sydney Dybing, Graduate Student, University of Oregon