LUGANO, Switzerland: Swiss researchers are using artificial intelligence and vast amounts of NASA climate data to develop faster ways to forecast extreme weather and detect early warning signs of natural disasters.
Scientists at the Federal Institute of Technology Zurich (ETH) have transferred around 100 petabytes of publicly available NASA data to servers connected to Alps, one of the world’s most powerful supercomputers.
The data, comprising around six billion files, took about a year to transfer, according to Reto Knutti, a climate physics professor at ETH and head of its Centre for Climate Systems Modelling.
The researchers believe combining massive Earth observation datasets with AI and high-performance computing could reveal patterns that conventional forecasting methods may miss.
AI could transform disaster warnings
AI-based statistical models can process weather patterns far faster than traditional mathematical simulations of complex atmospheric and oceanic processes.
Knutti said a statistical model could potentially produce a global forecast covering several days in about a minute, allowing researchers to run forecasts repeatedly and identify emerging weather patterns.
ALSO READ: Anthropic Discloses Fourth AI Hacking Incident Missed in Earlier Review
Although AI models require substantial resources to train, they are relatively inexpensive to operate once developed, potentially allowing more frequent forecasts and improved early warning systems for natural disasters.
“Data is essentially everything,” Knutti said, while stressing that the next challenge is turning the enormous volume of information into useful knowledge.
Satellite data may help save lives
Researchers say AI could also improve monitoring of hazards that are difficult to predict, including landslides and glacier collapses.
Knutti pointed to Switzerland’s Blatten, where satellite data showed warning signs of a glacier collapse more than a year before the disaster. Authorities evacuated the village a week before the collapse, helping prevent mass casualties.
The researchers said similar monitoring could potentially have identified warning signs before the devastating glacier collapse on the Nepal-China border on August 26.
Knutti said systematic analysis of satellite data could help establish disaster prevention monitoring systems capable of providing earlier warnings and potentially saving hundreds or thousands of lives.
