← Back to Sessions

On the Feasibility of Single-station Scalar Moment Estimation for Earthquake and Non-earthquake Sources

Description: 

Scalar seismic moment is the fundamental physically-grounded measure of source scale, yet its estimation is conventionally treated as contingent on network-scale observation and explicit source modeling. For small sources that cannot be observed across a well-instrumented region, these conditions are frequently absent. Sparse station coverage and uncertainty radiation characteristics impose additional limitations. Fortunately, the operational requirement is often an approximate measure of scale rather than a complete description of mechanism.

This work examines the extent to which scalar moment estimation is feasible when observation is reduced to a single station. We outline a conformer-based machine learning framework intended to infer scalar moment directly from broadband single-station waveforms. This is accomplished with an architecture that combines convolutional feature extraction with attention-based temporal modeling to isolate waveform characteristics diagnostic of source amplitude, while remaining tolerant to variability in source-time function, propagation effects, and site response. Training is formulated over a deliberately broad ensemble of synthetic and observed earthquake and non-earthquake signals, chosen to span diverse source types. We do not seek to recover full source mechanisms, moment tensors, or unique physical source model representations from single-station data. Emphasis is placed instead on exploring inference possibilities under conditions of extremely limited observations. The objective is to systematically explore the potential of single-station, ML-assisted scalar moment estimation as a practical direction for the characterization of earthquake and non-earthquake seismic sources in data-limited settings.

Session: Data-Driven and Computational Characterization of Non-Earthquake Seismoacoustic Sources - I

Type: Oral

Date: 4/16/2026

Presentation Time: 09:00 AM (local time)

Presenting Author: Benjamin L. Moyer

Student Presenter: Yes

Invited Presentation: 

Poster Number:

Authors