Researchers around the globe are using artificial intelligence (AI) to assist with conservation efforts, forecast natural disasters, and mitigate the impact of climate change. These projects include using algorithms to predict wildfires in the western US, deploying AI and drones with thermal-imaging cameras to count koalas in Australia, and facial recognition capabilities to monitor grizzly bear movements in Canada. An initiative involving AI and the world’s fastest supercomputer could revolutionize the way governments and disaster management teams respond to coastal flooding in the wake of tsunamis.
Using AI to predict tsunami flooding
On Tuesday, Fujitsu announced that it had successfully developed an AI model to help with tsunami-related disaster efforts in partnership with the International Research Institute of Disaster Science, Tohoku University, Earthquake Research Institute, and The University of Tokyo. These capabilities are hinged on Fujitsu’s Fugaku which boasts the illustrious title of “world’s fastest supercomputer.”
SEE: TechRepublic Premium editorial calendar: IT policies, checklists, toolkits, and research for download (TechRepublic Premium)
Overall, the effort utilized Fugaku to create “several high-resolution tsunami simulations” as well as an AI model that used offshore tsunami simulations and coastal flooding for training purposes, according to Fujitsu. Following an earthquake, Fujitsu said the AI model could use collected “tsunami waveform data” to forecast coastal flooding conditions before the water makes landfall.
The team tapped Fugaku to create training data consisting of 20,000 potential tsunamis, according to Fujitsu, and the model uses a “two-stage configuration” to first forecast flooding conditions in “rough resolution” based on observed offshore tsunami waveforms before increasing this resolution using AI.
The company said this process could enable area-specific flooding forecasts, as well as predictions about the potential impact of “localized waves” on urban infrastructure such as roadways and buildings.
Predicting floodwaters with “ordinary PCs”
It’s important to note that the Fugaku-trained AI algorithm can be run on “ordinary PCs” in a matter of seconds and this makes it “much easier to build practical, real-time flood prediction systems, which previously required supercomputers,” according to the post. The technology was used to predict flooding related to the Nankai Trough earthquake and this confirmed that it was possible to gain tsunami scenario predictions using a “regular PC in a matter of seconds,” Fujitsu said.
SEE: Aussie AI: Researchers use AI-enabled drones to protect the iconic koala (TechRepublic)
“By further utilizing the large scale, high speed performance of Fugaku in the future, while simultaneously training the system with additional tsunami scenarios, the partners can help to realize AI that can offer predictions for unexpected tsunami and flooding predictions over a wider area,” the post said.