Why Are Tsunamis So Difficult to Predict?

No one has infinite scientists or money.

How did everyone get caught completely unaware by the Dec. 2018 tsunami in Indonesia? originally appeared on Quora, the place to gain and share knowledge, empowering people to learn from others and better understand the world. You can follow Quora on Twitter, Facebook, and Google Plus.

Most tsunami warning systems are built around seismic sensors. The occurrence of a large earthquake near or under a body of water is what alerts the computers to alert people. This makes sense because most (I’ve seen numbers from 80% to 95%) measurable tsunami are caused by earthquakes.

The computers (and scientists, when there is time to loop them in) use the data from buoys and satellites to track the waves through the ocean. But the most effective way to do this is to have a model of the source event that the computer is working from. The initial warning is based mostly on rough estimate of quake size and location. Then the computer needs to check the relevant set of sensors around the expected tsunami time to refine the model and give a better warning to places further away.

This process fails in the case of landslides, even volcanic ones. The seismic signal from a landslide isn’t usually large enough to trigger a tsunami warning. The landslide itself is what causes the water displacement, not motion coupled with the surrounding earth.

So the first data the system sees that indicates anything is amiss is a pressure reading from offshore buoys. But one or two buoys aren't enough. One could be a glitch. Two doesn’t let you constrain direction, size or speed of a tsunami in three dimensions. You need multiple buoys, which are quite expensive and require expensive maintenance every year or few years. Even the Japanese, who have invested in earthquake and tsunami research and preparedness the way most countries invest in militaries and national defense, do not have a dense enough network of buoys along the coast to have recorded, recognized, and warned about a landslide-driven event such as this.

Indonesia is not a particularly wealthy nation. Since 2004, they have invested in buoys and scientists and evacuation routes and signs. But Indonesia has more coastline than any other country, and most of it faces some sort of tsunami hazard. Instrumenting the oceans and coasts of the entire country is a monumental task. It would be a huge undertaking by the standards of any nation, even those with higher GDPs and smaller populations to support. And even the best network in the world (Japan) is unlikely to have recognized this event before it struck shore because there was no big seismic event preceding it.

There is a chance that someone seeing the wave approach shore could have been given means to trigger an alarm system that alerted nearby beaches. But remember, tsunami waves are broad and only a few inches tall in deep water. They only rise up as the water shallows near shore. So it is only once the wave has struck somewhere that a human observer on the ground could trigger a warning. (At that point, presumably, the tide gauges are also alerting the computers at the heart of the sensor networks), but the warning is already too late for some people.

Knowing that Krakatau is at risk for explosive eruptions, scientists have made models over the years of the tsunami risks it poses. Now we will find out if their predicted travel times and estimated wave heights match the reality. But no one had the money or manpower it would take to monitor the volcano closely enough. There are hundreds of other volcanos in Indonesia, and a lot of seismicly active fault lines which make much bigger tsunamis.

If we had infinite money and scientists, we would have sensors all over that area, and we would replace them with drones whenever the lava, ash, or explosions took the sensors out. We would know whenever the mountain developed faults, and we would know within seconds that a flank collapse was starting. Then we would have our infinite computing power whip up a tsunami model based on that particular collapse, track the outgoing waves with our startlingly dense network of buoys and monitoring cables, and send a warning to shore that would still only give the population ~15–20minutes to find a safe spot (assuming our computers are very very fast).

No one has infinite scientists or money.

This question originally appeared on Quora. More questions on Quora:

* The Environment: How long did the Exxon Valdez oil spill (1989) take to clean up?

* Natural Disasters: How can an earthquake cause poverty?

* Current Events in Science: What is the largest natural disaster currently predicted to happen in 2018?

Photo Credit: NejroN/Getty Images