A tsunami in deep water is almost nothing to look at.
The whole wave might be a few inches from trough to crest, spread over a wavelength of a hundred miles or more, moving at the speed of an airliner.
A ship above it feels nothing at all.
All the energy is there, but it is distributed through the entire depth of the ocean rather than piled up at the surface.
What changes everything is the shelf.
Shallow water slows the front.
The wave piles into its own back
When the leading edge of a tsunami reaches shallower water it slows down, and the water behind it does not.
The wave compresses lengthwise and the energy has nowhere to go except upward, which is why a few inches offshore becomes many feet at a coastline.
How many feet depends entirely on the shape of the seabed and the coast, and those vary enormously over a few miles.
A wide gently shelving bay funnels and amplifies. A steep rocky shore lets the wave reflect straight back out again.
The same offshore wave can therefore give three feet at one harbor and ten at the next one along.
Which is why a single forecast number covering a whole coast is always an envelope.
A trench, a shelf and a very exposed shore
The Sanriku coast runs down the northeast of the Japanese mainland, facing a deep ocean trench where one plate is sliding under another.
Its shape is the problem. The coastline is a ria coast, drowned river valleys forming long narrow inlets that taper inland.
A funnel like that takes a wave and squeezes it into a smaller and smaller cross section, which is the classic recipe for a large run up.
This coast has the worst history in the country for exactly that reason, and everybody living there knows it, which shapes how the warnings are written before a single wave arrives.
So the warning system there is not calibrated for an average event.
It is calibrated for the worst one.
Ten minutes and a ten foot forecast
The quake struck at 16:53 local time on April 20 of 2026, about 60 miles east northeast of a port town in Iwate.
Magnitude was put at 7.4, with the two main agencies disagreeing on depth, one giving about 22 miles and the other about 6.
A tsunami warning went up for three prefectures, forecasting waves of up to 10 feet, and evacuation orders reached four more.
The gauges then reported what actually arrived. The tallest was 31 inches at one port, then roughly 16 inches at two more, 12 at another, and 8 apiece at two others.
Six gauges, six different answers, along one stretch of the same coast, from one wave.
Every advisory was lifted by just before midnight, seven hours after the shaking.
The gap between the forecast and the measurement was a factor of nearly four, and that gap is the entire design.
Why the overshoot is not an error
A tsunami forecast issued in the first minutes cannot be accurate, and pretending otherwise would be the actual failure.
What the agency has to work with at that moment is a seismic solution, nothing more. Magnitude, location and depth, with all three still uncertain, and no observation of a wave at all.
Converting that into a height means assuming how much seafloor moved and in which direction, and small errors in the fault model produce large errors in the wave.
So the forecast is deliberately biased high, because the two kinds of mistake are not symmetric, in the way that a major crossing or a new material is designed to a factor of safety.
Reporting at the time noted the alert was downgraded then lifted once real readings came in, which is the system working.
The cost is paid in belief
None of this is free, and the price is not measured in money.
Every warning that turns out to be four times too high teaches a coastline something, and what it teaches is that the number is theater.
That is the mechanism behind most tsunami deaths in living memory, which is not people who never heard a warning but people who heard one and decided, from experience, that it did not apply to them.
Forecasters know it, which is why the tone of the message matters as much as the figure in it, and why the wording tells people to go uphill rather than to expect a particular height.
Coverage of this event carried that wording straight into the headlines, repeating the instruction to evacuate ahead of any number.
The Sanriku coast was not overwarned by accident.
It was overwarned on purpose.
Read the whole thing?
Get the week's signal, not the noise
Our sharpest reporting on energy, climate and nature — free, once a week.
