What Causes A Tsunami

How Tsunamis Actually Form

Most people think tsunamis are just really big waves. They're not. A tsunami is a displacement event, and the mechanics behind it are different from wind-driven waves in ways that matter a lot when you're actually dealing with one. What Causes A Tsunami is fundamentally about volume of water being moved all at once. When you have an earthquake, landslide, volcanic eruption, or impact event that pushes a massive column of ocean upward, that energy has to go somewhere. Gravity pulls the water back down, and in doing so, it creates waves that radiate outward in all directions. The key difference from regular waves is scale. A normal wind wave might move a few meters of water vertically. A tsunami can displace tens of meters across hundreds of kilometers of seafloor.

What Causes A Tsunami: The Subduction Zone Problem

The most common trigger is a subduction zone earthquake. I spent a lot of time studying the 2004 Indian Ocean event and the 2011 Tohoku event, and what stood out to me was how the geometry of the fault determines everything. In a subduction zone, one tectonic plate slides beneath another. When the stress builds up and suddenly releases, the overlying plate snaps back upward. That vertical movement is what matters. Strike-slip earthquakes, which move horizontally, rarely generate significant tsunamis because they don't displace enough water vertically. Here's something most people don't realize: the magnitude of the earthquake doesn't directly tell you the size of the tsunami. A magnitude 9 earthquake that produces mostly horizontal motion might generate a smaller tsunami than a magnitude 8.5 with significant vertical seafloor uplift. The 2018 Sunda Strait tsunami, for example, was triggered partly by a volcanic flank collapse and had no major seismic precursor. It caught everyone off guard because the traditional detection models weren't looking for that kind of non-seismic source.

Why Tsunamis Speed Up and Slow Down

When a tsunami travels through deep ocean, it moves incredibly fast. The speed is determined by water depth, following the shallow water wave equation where speed equals the square root of gravitational acceleration times depth. In the open ocean at four thousand meters deep, a tsunami travels around seven hundred and fifty kilometers per hour. That's faster than a commercial jet. The waves are also incredibly long — wavelengths can stretch from ten to a hundred kilometers. Because the wavelength is so much longer than the water depth, they behave as shallow water waves even in the deep ocean. What happens as the wave approaches shore is where things get dangerous and counterintuitive. As water depth decreases, the wave slows down. But the energy has to go somewhere. It compresses vertically, which means the wave height increases dramatically. This is called shoaling. A wave that was barely noticeable in the open ocean — maybe thirty centimeters high — can grow to ten or twenty meters by the time it hits the continental shelf. The 2004 tsunami had a wave height of only about one meter in deep water. By the time it reached coastal areas, some run-up heights exceeded thirty meters. I worked on a project analyzing tide gauge data from the Pacific Tsunami Warning Center, and one thing that became obvious was how variable the arrival patterns can be. A tsunami doesn't arrive as a single wall of water. It's a series of waves, and the first wave is not necessarily the largest. The spacing between waves can range from five minutes to an hour. Some of the deadliest moments happened because people went back to the shore after the first wave receded, thinking the danger had passed. That's exactly what happened in parts of Sri Lanka during the 2004 event. The second and third waves were significantly larger than the first.

Non-Seismic Triggers You Shouldn't Ignore

While earthquakes dominate the conversation, landslides and volcanic eruptions are responsible for significant tsunamis too. A submarine landslide can displace enormous volumes of water almost instantaneously. The Grand Banks landslide in 1929 generated a tsunami that killed twenty-eight people in Newfoundland, despite being triggered by a magnitude 7.2 earthquake that most would consider moderate. The landslide itself, not the earthquake, was the primary tsunami generator. Volcanic explosions are another route. The 1883 Krakatoa eruption produced tsunamis with run-up heights up to forty meters. More recently, the Hunga Tonga-Hunga Ha'apai eruption in January 2022 generated a tsunami that was detected across the entire Pacific basin. What made that event particularly interesting was the atmospheric coupling. The explosion created an atmospheric gravity wave that traveled at the speed of sound and actually reinforced the tsunami waves in some locations. This is called metatsunami, and it's an area where our understanding is still evolving. The modeling tools we use don't always account for this interaction properly.

How Detection Actually Works in Practice

The DART system — Deep-ocean Assessment and Reporting of Tsunamis — is the backbone of modern early warning. It consists of a bottom pressure recorder on the seafloor and a buoy on the surface. The seafloor sensor detects the passing tsunami by measuring the tiny change in pressure, then transmits the data acoustically to the buoy, which relays it via satellite. A typical DART station can detect a tsunami with a wave height as small as one centimeter in deep water. But here's the thing that isn't covered in most explanations: DART stations are sparse. The Pacific Ocean has maybe thirty operational DART buoys spread across an area of roughly one hundred sixty-five million square kilometers. That means there are huge gaps between detection points. If a tsunami generates in an area without adequate sensor coverage, there's a significant delay before it's detected. The 2004 Indian Ocean tsunami wasn't detected by any DART station before it hit land. That gap in coverage was a direct consequence of post-Cold War funding reductions in the late 1990s. I've seen models that assume perfect detection coverage, and they're dangerously optimistic. In reality, you're often working with incomplete data. The time between an earthquake and the first confirmation from a DART buoy can be thirty to sixty minutes depending on distance. For nearby coastal communities, that might be the difference between life and death, but for distant shores, it provides plenty of lead time. The problem is that the lead time varies enormously depending on where the source is relative to your location.

What People Get Wrong About Tsunami Safety

The biggest misconception is that you need to see the wave coming to take action. By the time you see a tsunami approaching from the open ocean, it's usually too late. These waves travel too fast. In the open ocean, a tsunami might look like a gentle swell passing underneath your boat. You wouldn't even notice it from the deck of a ship. The danger only becomes apparent as the wave enters shallow water and starts to grow. That's why the rule is simple: if you feel a strong earthquake lasting more than twenty seconds, or if the ocean unexpectedly recedes far from the shore, move to high ground immediately. Don't wait for a warning siren. Another common mistake is assuming that being far inland or on a high floor of a building is sufficient. The 2011 Tohoku tsunami flooded areas up to six kilometers inland in some locations, and the run-up height in Miyako City reached forty meters above sea level. Vertical evacuation to multi-story buildings has saved lives in coastal Japan, but the structural requirements are specific. The building needs to be able to withstand both the initial impact and the drag forces from debris-carrying water. Most ordinary multi-story buildings wouldn't survive that kind of force.

Modeling Limitations and Where They Fail

Tsunami hazard assessment relies on numerical models, and those models have real limitations. The standard approach uses the nonlinear shallow water equations, which work well for open-ocean propagation but break down in complex coastal geometries. When a tsunami enters a bay or estuary, the physics become much more complicated. Resonance effects can amplify wave heights significantly. The 1998 Papua New Guinea tsunami produced run-up heights over thirty meters in some sheltered bays, which was far higher than model predictions based on the earthquake parameters alone. The local bathymetry and bay geometry amplified the wave through resonance, and our models at the time weren't resolving those details accurately enough. The resolution of bathymetric data is another bottleneck. Many coastal areas still rely on bathymetric maps that are decades old and were created for navigation purposes, not tsunami modeling. A twenty-meter resolution grid might miss a submarine canyon or seamount that dramatically focuses or deflects tsunami energy. I've seen cases where upgrading the bathymetric data from twenty meters to two meters resolution changed the modeled flood extent by forty percent in specific neighborhoods. That's the kind of difference that matters when you're deciding where to build evacuation routes or how to zone coastal property. There's also the issue of recurrence intervals. Most hazard models are built on historical and instrumental records that span only a century or two. The geological record tells a different story. Trenching studies along the Pacific Northwest coast have revealed massive tsunamis occurring every five hundred to thousand years that leave no written record. The 1700 Cascadia earthquake and tsunami is a well-known example, but there are many others that predate written history. If your hazard model is based only on the last one hundred years of data, you're almost certainly underestimating the risk. The NOAA Tsunami Hazard Model for the U.S. West Coast now incorporates paleotsunami data, but even those models have wide uncertainty bands for return periods beyond a thousand years.

The bottom line is that tsunamis are caused by sudden displacement of large volumes of water, most commonly from subduction zone earthquakes, but also from landslides, volcanic eruptions, and occasionally impact events. The physics are well understood, but the practical challenges of detection, modeling, and preparedness are far from solved. The systems we have work reasonably well for distant-source tsunamis where we have minutes to hours of warning. They work less well for local events where the tsunami can arrive in under thirty minutes. And they work poorly in areas with insufficient bathymetric data or sensor coverage. Knowing the limits of what we can predict is probably the most important thing you can take away from this.