Fujitsu develops an app for tsunami prediction system

For the past year, Fujitsu has been developing a tsunami prediction system using supercomputers and AI to try and identify areas that are at risk of flooding. Fujitsu is taking its developments further and now developing an app that will warn users in dangerous areas to seek safety. What challenges do tsunamis present, what has Fujitsu been working on, and what other natural disasters warnings could be streamed to user devices?

What challenges do tsunamis present?

Any coastal area that is in close proximity to a fault line (where two tectonic plates meet) is at risk of tsunami damage, whereby an underwater earthquake causes a swell of water to rush inland. Unlike tidal waves (caused by tides and wind), a tsunami is a large body of displaced water that moves inland with incredible force and can go inland for many miles.

A practical analogy of the difference between the two is that a tidal wave is akin to splashing in a bathtub, while a tsunami is a sudden change in water level after getting into a bath. The splashing is annoying and soaks whatever it hits, but the sudden entry into the tub sees enormous amounts of water pushing its way out and going extremely far away.

Thus, tsunamis are problematic as they can easily overwhelm coastal defence, have the force to rip up buildings and trees, and travel very far inland which causes widespread flooding and damage to infrastructure. Worse, there is very little warning of an incoming tsunami as while underwater earthquakes can be detected, it is not so easy to detect significant changes in wave height far out; it is only when a tsunami approaches a shoreline that the severity of the wave can be observed.

Once a tsunami hits, it can be extremely difficult to alert those in the path of the tsunami and even more difficult to determine which areas are at most risk. As such, the general advice where tsunamis hit is to seek high ground regardless of where you are, turn off all utilities, and if the wave is about to hit, grab onto anything sturdy (such as a pole, a tree, or a concrete pillar).

Fujitsu announces development of early warning tsunami app

Last year, Fujitsu announced that it was using a supercomputer and AI to develop a flooding prediction system with the hope of providing early warning capabilities to those in vulnerable areas. Simply put, the AI is trained with historical earthquake and tsunami data to predict the wave height of incoming tsunamis and then combine this data with realistic geological models to identify how the tsunami wave will move inland and determine which areas are at risk of flooding. 

Now, after a year of development, Fujitsu has announced that it is working on an app that it believes will give users warning of flooding in their area and when an incoming tsunami is expected to hit. Last March, Fujitsu reported that it was able to try out its flood prediction app during a tsunami evacuation drill in Kawasaki. Selected members of the public were invited to install the app on their devices, and they could see the number of minutes before the tsunami hit specific areas (shaded as blue squares).

But not only does the app show users the estimated time of arrival for the tsunami, but it can also alert authorities and local organisations if people have been left behind in danger zones. For example, an elderly hospital in a danger zone may have patients who have been unable to call for help, and the app would allow authorities to locate them.

While the initial trials of the app show promising results, it is still faced with numerous challenges. One of these is that tsunami prediction requires government permission to share with the public due to the potential loss of life. As such, the app cannot be released to the public even as a beta, meaning testing the app at scale is challenging.

What other natural disasters could early warning apps detect?

Tsunamis are devastating but not the only natural disaster that could benefit from apps and prediction modelling. One excellent use of such apps would be the prediction of wildfire spread; they are difficult to control and predict as they will spread to any combustible material that is in reach. But they are also highly dependent on prevailing winds and will often follow the direction of the wind. As such, it may be possible to predict early warning systems that identify key areas prone to wildfires and, if a wildfire is detected, determine the extent of the damage that can be expected. It could then alert those in the path of the fire when it may reach them and suggest evacuation routes based on the wind.

Another example of a natural disaster warning app would be tornado prediction. While news alerts exist for tornados, it would be more advantageous to alert people via an app on a smartphone if they are in a predicted danger zone. As tornados can strike unpredictably, it could be vital to give people precious moments to prepare, whether fortifying a home or evacuating.

Overall, natural disasters are virtually impossible to stop, and when they happen, even having just a few moments of warning can be the difference between life and death.