Where the wave begins
The claim at the centre of this brief is straightforward but easy to misunderstand: the most dangerous tsunamis are generated not by wind or storms but by the sudden vertical movement of the seafloor itself. At a subduction zone, one tectonic plate is being forced beneath another, and the boundary between them can remain locked for long periods before slipping suddenly in what is called a megathrust earthquake. When that slip happens under the ocean, it can shove a huge area of seafloor either up or down within seconds, and the entire column of water above it moves with it. That abrupt displacement is what launches a tsunami, which then spreads outward across the ocean as a wave train.
Invisible in the deep ocean
In the open ocean this wave is almost unnoticeable from a ship or the shore, typically rising only about a metre above normal sea level, because its wavelength can stretch to roughly two hundred kilometres. What makes it dangerous is speed: in water several kilometres deep, a tsunami can travel faster than eight hundred kilometres an hour, comparable to a passenger jet. As it crosses into shallower coastal water, wave shoaling comes into play, the wave slows to well under a hundred kilometres an hour while its wavelength shrinks and its height grows substantially, sometimes taking several minutes to reach its full size once it arrives. It can behave less like a breaking wave and more like a fast-moving surge of water pushing inland.
Growing dangerous near shore
The historical record gives a sense of scale. The 1960 Valdivia earthquake in Chile, estimated at around magnitude 9.5, remains the largest instrumentally recorded earthquake and produced a tsunami that crossed the Pacific. The 2004 Indian Ocean earthquake, generated on the Sunda megathrust, killed or left missing at least 230,000 people across fourteen countries, making it one of the deadliest natural disasters on record. The 2011 Tōhoku earthquake off Japan produced waves that exceeded the height of coastal seawalls built specifically to hold back tsunamis, contributing to the Fukushima nuclear accident. These events, along with the 1964 Alaska earthquake, are treated as the reference cases for understanding megathrust-generated tsunamis.
What the record shows
Detection and warning have improved substantially since these disasters. Modern systems, including the Pacific Tsunami Warning System, use seafloor pressure sensors mounted on buoys to detect the passage of a tsunami directly, feeding that data into computer models that can estimate arrival times at distant coastlines within minutes of an earthquake being detected. Some coastlines also exhibit a warning sign of their own: because a megathrust rupture can first pull water away from shore before the main wave arrives, an unusual and rapid recession of the sea can itself signal an approaching tsunami, though this drawback does not occur for every earthquake or every coastline, depending on how the fault ruptured.
Warning systems and their limits
Even with better detection, physical defences have repeatedly been overwhelmed by the largest events. Japan built seawalls up to twelve metres high and floodgates reaching over fifteen metres in response to past tsunamis, yet the 2011 wave still exceeded these barriers in places. Similarly, the 1993 Okushiri tsunami produced waves around thirty metres high that washed over surrounding walls despite being slowed by them. This matters well beyond the science, because coastal planning, nuclear plant siting, and insurance and evacuation policy all depend on realistic assumptions about how large a tsunami a given stretch of coast might eventually face, assumptions that these events have repeatedly forced planners to revise upward.
What is still uncertain
This is worth understanding because it corrects a common intuition: a tsunami’s danger has almost nothing to do with what it looks like far offshore, and almost everything to do with the shape of the seafloor it eventually crosses. The mechanism, sudden seafloor displacement at a subduction zone producing a fast, low wave that steepens dramatically near land, is well established and consistently supported across many independent events. What is not settled is prediction: seismologists can identify which subduction zones are capable of megathrust earthquakes and roughly how large those earthquakes could be, but not when the next one will occur. For anyone living near a subduction zone coastline, that gap between understood mechanism and unpredictable timing is the whole point.