HypoSVI: Hypocentre inversion with Stein variational inference and physics informed neural networks

Abstract

Summary We introduce a scheme for probabilistic hypocenter inversion with Stein variational inference. Our approach uses a differentiable forward model in the form of a physics informed neural network, which we train to solve the Eikonal equation. This allows for rapid approximation of the posterior by iteratively optimizing a collection of particles against a kernelized Stein discrepancy. We show that the method is well-equipped to handle highly multimodal posterior distributions, which are common in hypocentral inverse problems. A suite of experiments is performed to examine the influence of the various hyperparameters. Once trained, the method is valid for any seismic network geometry within the study area without the need to build travel time tables. We show that the computational demands scale efficiently with the number of differential times, making it ideal for large-N sensing technologies like Distributed Acoustic Sensing. The techniques outlined in this manuscript have considerable implications beyond just ray-tracing procedures, with the work flow applicable to other fields with computationally expensive inversion procedures such as full waveform inversion.

The 2020 Westmorland, California Earthquake Swarm as Aftershocks of a Slow Slip Event Sustained by Fluid Flow

“…Because the station distribution and local seismic velocity structure differ, we re-train the neural network following Ross, Trugman, Hauksson, & Shearer (2019) and Ross, Yue, et al (2019) and use the exact settings for the associator described in Ross and Cochran (2021). Once the association process is completed, we locate the events using HypoSVI (Smith et al, 2021), a variational Bayesian method. We use the Southern California Earthquake Center Community Velocity Model CVM-H (Shaw et al, 2015) and keep all tunable parameters the same as in Smith et al (2021).…”

Scientific Machine Learning Through Physics–Informed Neural Networks: Where we are and What’s Next

“…For earthquake hypocenter inversion, Smith et al [163] use Stein variational inference with a PINN trained to solve the Eikonal equation as a forward model, and then test the method against a database of Southern California earthquakes.…”

Scientific Machine Learning through Physics-Informed Neural Networks: Where we are and What's next

“…For earthquake hypocenter inversion, Smith et al (2021b) use Stein variational inference with a PINN trained to solve the Eikonal equation as a forward model, and then test the method against a database of Southern California earthquakes.…”