HPC Real-time Tsunami Forecasting — Disaster Geoinformatics Lab. (ReGiD)

Real-time ambition

The “10-10-10 Challenge”

Estimate the tsunami source in roughly 10 minutes, compute inundation on a 10 m mesh in roughly 10 minutes, and deliver actionable information soon after an earthquake.

01 · Real-time assessment

Tsunami inundation and damage estimation

Using real-time earthquake information and supercomputing, we develop technology to estimate tsunami damage rapidly. The work combines source estimation, inundation computation, building exposure, and fragility functions in collaboration with university and industry partners.

Real-time computing10 m meshDamage estimationFragility functions

Real-time tsunami inundation and damage prediction using the SX-ACE supercomputer — Real-time tsunami inundation and damage prediction using high-performance computing.

02 · Large-scale scenarios

Big-data analysis for disaster mitigation

The JST CREST project brings together disaster simulation, disaster management science, mathematics, and information science. Large-scale numerical simulation and real-time data assimilation help explore the “scenario explosion” and identify measures that can reduce worst-case impacts.

Data assimilationScenario analysisDecision support

Controlling the scenario explosion in large-scale disaster simulation.

03 · International applications

South American tsunami prediction

For tsunami-prone countries in Central and South America, we combine real-time computation, remote sensing, GIS, and fragility functions to assess near-field and far-field impacts and support wide-area damage estimation.

Latin AmericaGISNear/far field

Wide-area damage assessment for South American countries — Wide-area damage assessment combining tsunami simulation and remote sensing.

04 · Urban tsunami physics

Simulation that reproduces building damage

In dense urban areas, buildings obstruct flow and may be washed away as inundation progresses. A time-evolving synthetic equivalent roughness approach reproduces the changing urban surface and estimates which buildings fail, when, and how many.

Building damageTsunami simulationEquivalent roughness

Observed and estimated washed-away buildings — Observed and estimated washed-away building distributions for the Sendai Plain.

05 · Next-generation methods

Lattice Boltzmann tsunami simulation

The Lattice Boltzmann method is a fully explicit scheme well suited to massively parallel computing. We explore its potential for faster urban run-up simulation while maintaining useful accuracy.

Lattice BoltzmannParallel computingTsunami simulation

Tsunami inundation simulation using the Lattice Boltzmann method.

06 · Uncertainty and early warning

Rapid worst-case tsunami estimation

Immediately after an earthquake, the initial source estimate is uncertain. By exploring many possible fault arrangements, we estimate a credible worst-case tsunami scenario in minutes to support rapid assessment and evacuation decisions.

Early warningUncertaintyWorst-case scenarios

Fault arrangements for worst-case tsunami estimation — Example fault arrangements used to explore high-amplitude tsunami scenarios.