The Vancouver Island tsunami warning siren has never been heard in practice, but scientists know that the lull is temporary. A new study says artificial intelligence can sharpen the fleeting calls that determine whether visitors and local people can reach a higher position.
This work is Zero in Tofino, a surfing town filled with beaches and hotels all year round. Researchers have tested a variety of warning strategies and found that machine learning could save more people than the rules used by emergency crews today.
Professor Katsuichiro Goda, a professor of geoscience at Western University and Canada's research committee chair for multi-hazard risk assessment, led the project.
He and his team compared traditional statistical models with new Random Forest algorithms and neural network tools to see how long an official could wait before pressing the alert button.
Tofino's High Stakes
Tofino sits in a short drive from the Cascadia subduction zone, the 600-mile fault where Juan de Fuca Plate dives under North America.
A rupture of its boundary size 9 could send a 65-foot wall of water onto land in about 20 minutes.
Japan's S‑NET network packs 150 marine bottom sensors along deep-sea cables, but Vancouver Island relies only on four nodes for real-time data.
That imbalance is important as Tofino hotels, marinas and boardwalk shops outweigh risky assets by over $2 billion in Canadian dollars during severe flooding.
Why are minutes important?
Evacuation drills show that even fit hikers can take up to 17 minutes to climb from the waterfront to the designated hill. An alert too late jams a single highway from town and locks beach fans behind food stall traffic.
Because Canada does not have a local catalogue of historic tsunamis, this study gave computers thousands of simulated waveforms, rather than actual past events.
The team discovered that performance shaking wildly when the training set omits certain rupture styles.
Adding a synthetic event from Bayesian's “digital twin” model could close that gap. This is the method being tested on the Cascadia Coast.
Random forest vs. human judgment
“If latency is too short, the performance of the early warning model for tsunamis differs greatly in terms of success,” Goda said.
His favorite tool, Random Forest, stacks dozens of mini decision trees, weighing different seismic variables before voting at the safest moment to send text alerts.
In the test, the algorithm hits multiple linear regression by approximately 15% with successful evacuation, reflecting the results of an international trial linking the undersea gauge with machine learning predictions.
Tsunami warning siren
Japan's 150-station array beams pressure and seismic data along 3,600 miles of fiber optic cables, providing residents with clearer information on wave heights and arrival times.
Near tourist beaches like Cox Bay, near tourist beaches like Cox Bay, Vancouver Island's 4-sensol network limits the accuracy of early detection near tourist beaches like Cox Bay.
Goda claims that even a modest expansion, one additional sensor every 25 miles of the coastline will provide a hunger algorithm and reduce the interval between earthquakes and alerts.
Crosshair Tourist Economy
Peak season will boost Tofino's population from 2,500 to nearly 20,000, expanding the challenge of bringing people uphill before the waves arrive.
Business owners should note that accuracy and trust are closely linked, as cancelled bookings can cost hundreds of thousands of dollars.
Locals say the success of any alert system depends on whether people trust it or not. That trust depends on understanding how and why evacuation orders are made.
How workshops and school programs explaining the dynamics of the tsunami and how warnings work have helped raise awareness in Tofino.
These efforts will build a culture of preparation, and as seconds count, people are more likely to act quickly.
Towards a smarter evacuation
“To get started with more AI models, you need to get better results, but more data is being hit and only more data is improving performance,” Goda explained.
Western research suggests counterintuitive tactics. Wait a few extra seconds before alerting the message to include estimates of hard waves.
Past disasters show why precision issues are important. The investigation was conducted after it was found that the repeated tsunami of 2011 was unlikely to allow some residents to escape the small waves later.
The next fall, Goda group will join residents during Tofino's annual drill to test the AI assist alert generator in real time. If the prototype works, the same code can protect the resort from Bali to Havana, reducing the casualty gap between often overlooked and often overlooked coastlines.
This research has been published in Coastal Engineering Journal.
– –
Like what you read? Subscribe to our newsletter for compelling articles, exclusive content and the latest updates.
Check out us on EarthSnap, a free app provided by Eric Ralls and Earth.com.
– –
