As it turns out, when it comes to the weather, the past is the best predictor of the future.
The AFP reported on how Microsoft recently tapped into an artificial intelligence system, named Aurora, trained solely on historical data to beat out conventional weather forecasting predictors in a piece published on TechXplore.
"For the first time, an AI system can outperform all operational centers for hurricane forecasting," exclaimed senior author of the study, Paris Perdikaris. Aurora's capabilities were outlined in the study published in Nature. Impressively, the AI-powered system requires far less computational cost than traditional methods.
Considering the shortcomings of conventional weather prediction models, AI is a promising frontier. Instead of using historical data, traditional models look at first physical principles like momentum and conservation of mass. That process is energy-intensive.
Using historical inputs, the researchers revealed that Aurora outperformed competitors like the U.S. National Hurricane Center in forecasting all of 2023's hurricanes. That success was extended to projecting out cyclones like the devastating Doksuri typhoon that hit Asia.
The AFP noted that at the time, forecasts dropped the ball and predicted it would end up north of Taiwan. Aurora, meanwhile, predicted it would strike the Philippines four days in advance.
Another area of success the study highlighted was 10-day forecasts. Per the researchers, Aurora bested the well-regarded European Centre for Medium-Range Weather Forecasts (ECMWF) model 92% of the time in global forecasts.
That performance, as well as Google's own exemplary results with an AI model called GenCast, is forcing the traditional weather prediction agencies to take notice.
"This is something we have taken very seriously," Florence Rabier, director general of the ECMWF, commented.
The ECMWF is working on and improving its own AI model that it unveiled in February of this year. Rabier revealed it was "1,000 times less expensive" than conventional physical models.
TCD Picks » Upway Spotlight
💡Upway makes it easy to find discounts of up to 60% on premium e-bike brands
|
Should the government be allowed to control how restaurants run their business? Click your choice to see results and speak your mind. |
Never 👎 This race to unveil the best weather prediction models while harnessing new technology could do wonders for disaster preparedness. A group of researchers in Australia has similar aspirations for aiding wildfire preparedness with an AI tool. An MIT team is using the tech to help predict flooding after torrential storms.
Another use case for AI and weather prediction is in agriculture. More accurate forecasts can play a major role in optimizing harvests.
Perhaps most encouraging of all, the researchers who developed Aurora think the model has even more potential to revolutionize weather forecasting.
"I believe that we're at the beginning of a transformation age in air system science," Perdikaris shared in a video that accompanied the study.
While the historical data obviously worked some wonders, the scientist believes that with some more inputs and data, it will be really unleashed.
"In the next five to 10 years the holy grail is how to build systems that can directly work with observations from remote sensing sources like satellites and weather stations to generate forecasts at high resolution anywhere we like," Perdikaris said.
Join our free newsletter for weekly updates on the latest innovations improving our lives and shaping our future, and don't miss this cool list of easy ways to help yourself while helping the planet.
Cool Picks
