Science Corner Sep 2025

by Alice Gabriel (Scripps/UCSD)

Sep 27, 2025

Real-time Bayesian inference at extreme scale: A digital twin for tsunami early warning applied to the Cascadia subduction zone

Tsunamis generated by great subduction earthquakes are among the most devastating natural hazards. The challenge is speed: reliable warnings must be issued within minutes of rupture onset, often when only sparse data are available. This is especially urgent for the Cascadia subduction zone in the Pacific Northwest, where coastal communities are at high risk during the next large earthquake and tsunami.

In a recent study, we present a tsunami digital twin framework that couples physics-based earthquake and tsunami simulations with real-time Bayesian inference. Running on leadership-class supercomputers, including El Capitan, the world’s largest publicly accessible system, the framework continuously assimilates new observations and updates tsunami forecasts on the fly. Within minutes, thousands of plausible scenarios are evaluated, yielding probabilistic predictions of tsunami height and arrival time.

This represents a fundamental shift: from static precomputed scenario catalogs to a dynamic and adaptive early warning system. Demonstrated for Cascadia, the digital twin shows how uncertainty quantification and extreme-scale computing can overcome longstanding barriers to rapid tsunami source characterization. The project brings together computational scientists and geophysicists and has been recognized as an ACM Gordon Bell Prize Finalist at SC’25 (https://sc25.supercomputing.org). Next, we plan to explore optimal experimental design to guide the placement of future seafloor sensors, including emerging technologies such as seafloor distributed acoustic sensing (DAS), and to extend our digital twin framework to other regions, starting with Japan’s offshore S-Net observatory.

Reference: S. Henneking, S. Venkat, V. Dobrev, J. Camier, T. Kolev, M. Fernando, A.-A. Gabriel, O. Ghattas (2025), "Real-time Bayesian inference at extreme scale: A digital twin for tsunami early warning applied to the Cascadia subduction zone", to appear in Proceedings of SC’25 International Conference for High Performance Computing, Networking, Storage and Analysis, ACM Gordon Bell Prize Finalist, arXiv preprint, https://doi.org/10.48550/arXiv.2504.16344

Figure: (a) Topobathymetric map of the Cascadia subduction zone (CSZ); (b)–(c) Snapshots of vertical seafloor uplift and seafloor velocity from a physics-based 3D dynamic rupture and seismic wave propagation computation of a magnitude 8.7 earthquake scenario spanning the full margin of the CSZ; (d) Representative block of a 3D multi-block hexahedral mesh of the CSZ, depicting bathymetry-adapted meshing.

This video visualizes data obtained from large-scale computations of an acoustic–gravity model implemented with the MFEM finite element library and performed on the El Capitan supercomputer. The 3D high-fidelity model computes the coupled ocean acoustic and surface gravity waves for a magnitude 8.7 earthquake scenario spanning the full margin of the Cascadia subduction zone that stretches 1000 km from northern California to British Columbia. These computations are fundamental to enabling a newly developed digital twin methodology for real-time warning of tsunamis. Specifically, this Bayesian inversion-based digital twin employs acoustic pressure data from seafloor sensors, along with 3D coupled acoustic–gravity wave equations, to infer earthquake-induced spatiotemporal seafloor motion in real time and forecast tsunami propagation toward coastlines for early warning with quantified uncertainties. Details of this work are available in Henneking et al., 2025, to appear in Proceedings of SC25.