Towards physics-based probabilistic estimates of earthquake ground motion using Gaussian processes

Estimates of seismic wave speeds in the Earth (seismic velocity models) are key input parameters to earthquake simulations for ground motion prediction. Owing to the non-uniqueness of the seismic inverse problem (i.e., inference of seismic wave speeds from recorded ground motion data), typically several velocity models exist for any given region. The arbitrary choice of which velocity model to use in earthquake simulations impacts ground motion predictions. However, current hazard analysis methods do not account for these sources of uncertainty.

I will present a proof-of-concept ground motion prediction workflow for incorporating uncertainties arising from inconsistencies between existing seismic velocity models. The analysis is based on the probabilistic fusion of overlapping velocity models using scalable Gaussian process (GP) regression. Specifically, I fit a GP to two synthetic 1-D velocity profiles simultaneously, and show that the predictive uncertainty accounts for the differences between the models. I subsequently estimate peak ground displacement using acoustic wave propagation through velocity models drawn from the GP. The resulting distribution of possible ground motion amplitudes is much wider than would be predicted based on the two input velocity models. This proof-of-concept illustrates the importance of probabilistic methods, and the applicability of probabilistic machine learning, for physics-based seismic hazard analysis.