Toward Ground Motion Forecasting

Tim Clements, Elizabeth S. Cochran, Sarah E. Minson, Nicholas J. van der Elst, Clara E. Yoon, Annemarie S. Baltay, Morgan T. Page, & Max Schneider

Submitted August 30, 2026, SCEC Contribution #15417, 2026 SCEC Annual Meeting Poster #TBD

Typically, earthquake forecasts project the expected rate or number of earthquakes in the hours to weeks following a large earthquake. While grounded in the statistics and physics of seismogenesis, earthquake forecasts do not account for local ground motion variability. Here, we develop a ground motion-based approach to earthquake forecasting. Rather than forecast the rate or number of earthquakes and apply a ground motion model, we directly forecast future ground motion at a particular location from recorded ground motion, effectively combining Earthquake Early Warning (EEW) and Operational Aftershock Forecasting. To do so, we develop a suite of Generate Unsupervised Aftershock Velocity Amplitudes (GUAVA) models using the Transformer (Vaswani et al., 2017) architecture. GUAVA models generate stochastic time series of ground motion intensity autoregressively at 5 Hz sampling rate with recorded ground motion time series as input. We train GUAVA on all continuous ground motion time series recorded by the Southern California Seismic Network in 2019 (~3 TB) and all recordings from the K-NET and KiK-net arrays in Japan in 1996 - 2026 (~85 GB for 20,566 events). The GUAVA training uses stochastic gradient descent to maximize the log-likelihood of the next ground motion amplitude A(t+1) given the previous 8,192=2^13 ground motion samples, effectively the last 1638.4 seconds of data at 5 Hz. When given a new ground motion time series as input, GUAVA generates suites of ground motion time series, complete with body, surface, and coda waves from aftershocks. We evaluate GUAVA’s shaking intensity accuracy and timeliness as an EEW and forecasting ability on the December 5, 2024, M 7.0 Offshore Cape Mendocino earthquake. GUAVA could be run in real-time using streaming seismic data to forecast ground motion from seconds to minutes after intense shaking occurs.

Key Words
shaking, ground motion, AI, forecast

Citation
Clements, T., Cochran, E. S., Minson, S. E., van der Elst, N. J., Yoon, C. E., Baltay, A. S., Page, M. T., & Schneider, M. (2026, 08). Toward Ground Motion Forecasting. Poster Presentation at 2026 SCEC Annual Meeting.

Related Projects & Working Groups
Earthquake Forecasting and Predictability (EFP)