The Cascadia Megathrust: How a Full M9.0 Rupture Looks

Imagine the ground beginning to move under Portland at 2:46 a.m. Six minutes and forty seconds later, the shaking has traveled the full length of a fault more than a thousand kilometers long. By 3:52, Seattle is shaking hard. This timeline comes from a simplified physics model of a full Cascadia megathrust rupture: a magnitude 9.0 earthquake that breaks roughly 1,130 kilometers of the subduction interface in one continuous event.

The model is not an official hazard map. It is an educational exercise that shows, in concrete numbers, what the physics of such a rupture implies for the Pacific Northwest coast and places as far away as Japan. Everything that follows is drawn strictly from that model’s key results, its construction methods, and the historical and probabilistic references it cites.

The Rupture Itself

The fault is the Cascadia trench, traced from real bathymetry using AWS Terrain Tiles and divided into 22 segments. The rupture is modeled using Okada’s 1985 elastic dislocation equations, with an 11-degree dip and a fault width of 130 kilometers. In the full-rupture case, the entire length fails. In some places, the seafloor rises by up to 5.4 meters. Along the coast, the land drops between 0.5 and 1.6 meters. That vertical displacement drives the tsunami.

Because the rupture takes about 6 minutes and 40 seconds to complete, strong shaking doesn’t arrive everywhere at once. Portland feels the motion at 2:46; Seattle feels it more than an hour later, at 3:52. The difference reflects the time it takes the rupture front to travel the zone and seismic waves to reach each city.

The First Waves

Once the seafloor has been permanently deformed, the water above it begins to move. The model solves the linear shallow-water equations on a 1/60-degree grid near the coast and a coarser 0.15-degree grid across the open Pacific, again using real bathymetry. The results give precise arrival times for the first wave taller than one meter:

• Brookings: 9 minutes
• Coos Bay: 12 minutes
• Newport: 21 minutes
• Cannon Beach: 30 minutes
• Seaside: 31 minutes
• Ocean Shores: 35 minutes

These are not the peaks. At Seaside, in water 30 meters deep, the model records a peak height of 9.1 meters 43 minutes after the rupture starts. Run-up on land is higher still. The same physics that produces a 9.1-meter wave offshore will drive water farther inland and higher up slopes once it reaches the shore.

Probabilities from 10,000 Runs

A single deterministic scenario helps illustrate the process, but the real Cascadia zone does not always rupture its entire length. The model therefore runs a 10,000-member Monte Carlo ensemble. The model draws full ruptures and southern-only ruptures in proportions consistent with the turbidite record compiled by Goldfinger and colleagues in 2012. Slip is distributed in a lognormal, patchy pattern across the fault segments.

Across those 10,000 realizations, the statistics are clear. Southern Oregon and northern California experience a wave taller than one meter in roughly three out of every four runs. Washington sees waves over one meter in 46 percent of runs. At Seaside and Cannon Beach, the numbers are more severe: a wave exceeding 5 meters occurs in about 31 percent of the simulations; a wave exceeding 10 meters occurs in about 10 percent.

These percentages are not forecasts for any particular future event. They are the relative frequencies produced by the ensemble under the model’s assumptions—full versus partial ruptures weighted by turbidite history, and heterogeneous slip.

The Distant Wave

The same tsunami that arrives at Brookings in nine minutes continues across the Pacific. The model places the first arrival in Japan at 9 hours and 15 minutes. Historical records of the 1700 Cascadia earthquake, reconstructed from Japanese documents and from geologic evidence on the North American side, place the arrival at roughly 10 hours. The model’s travel time therefore matches the best available observational constraint for that earlier event.

The 1700 earthquake is known from two independent lines of evidence cited in the model documentation: ghost forests and coastal subsidence records described by Atwater and colleagues in The Orphan Tsunami of 1700. Japanese tsunami records analyzed by Satake and colleagues in 1996. Those sources fix both the approximate size of the 1700 rupture and the time of day it occurred, allowing the orphan-tsunami connection to be made.

Odds and Human Consequences

The Pacific Northwest Seismic Network estimates the probability of a full rupture at about 15 percent in the next 50 years. That figure is independent of the educational model; it is simply the reference probability the model authors invoke when discussing the likelihood of the scenario they have simulated.

For a sense of what such an event would mean on the ground, the model points to the FEMA Cascadia Rising exercise of 2016, which examined impacts across Oregon and Washington. The model does not reproduce that exercise’s casualty numbers; it simply notes the scenario as a reference for the problem’s scale. One concrete mitigation example is also cited: Ocosta Elementary School in Westport, Washington, completed in 2016 and designed as a vertical-evacuation structure.

How the Numbers Were Generated

Every quantitative claim above traces back to a specific modeling choice. The fault geometry begins with bathymetry from AWS Terrain Tiles, segmented into 22 pieces. We impose slip using the Okada elastic dislocation formula at an 11-degree dip across a 130-kilometer-wide interface. The resulting seafloor deformation becomes the initial condition for a linear shallow-water tsunami model run on nested grids of 1/60 degree near shore and 0.15 degree across the Pacific, both using real bathymetry. The Monte Carlo layer adds 10,000 realizations that mix full and southern-only ruptures according to the Goldfinger turbidite proportions and that distribute slip in a lognormal patchy fashion. We extract arrival times, peak heights, and exceedance statistics directly from those runs. We compare the Japanese travel time with the 1700 historical constraint. The 15 percent / 50-year probability and the Cascadia Rising casualty framing are external references supplied for context.

Limits of the Exercise

The authors are explicit: this is a simplified physics model intended for education, not an official hazard map. Use official tsunami inundation maps published by each state to determine real evacuation routes. The numbers shown here illustrate the physics of a full rupture and the statistical envelope of partial and full ruptures under one set of assumptions. They do not replace site-specific engineering studies or official guidance.

The model makes vivid the extreme compression of time along the southern Oregon coast, the longer but still short window farther north, the substantial chance of multi-meter waves at popular beach towns, and the fact that the same event will still be traveling across the Pacific nine hours later. Those are the concrete results of the calculation.
_{Source: Doomsday Math}

As Always, Stay Vigilant and Be Prepared

You play a critical role in your preparedness. By preparing yourself for the unexpected, you will become more self-reliant and a valuable asset to your community