One of Microsoft’s latest AI models can accurately predict air quality, hurricanes, hurricanes and other weather -related phenomena, the company claims.
To one document published in the journal Nature and accompanying blog This week, Microsoft reports in detail Aurora, which the technological giant says it can predict atmospheric events with greater precision and speed from traditional meteorological approaches. Aurora, who has been trained in more than one million hours of data from satellites, radar and meteorological stations, simulations and forecasts, can be adapted to additional data to make predictions for specific weather events.
AI weather models are nothing new. Google Deepmind has been released a handful in recent years, including Stripwhich the lab claims to be hitting some of the best prediction systems in the world. Microsoft places Aurora as one of the leading artists of the race – and a possible benefit for workshops studying the science of weather conditions.
In the experiments, Aurora predicted Typhoon Doksuri’s land in the Philippines four days before the actual event, hitting some expert forecasts, says Microsoft. The model also improved the National Blind Center to predict the five-day tropical cyclones for the era 2022-2023 and successfully predicted in 2022 Iraq Sandstorm.
While Aurora required significant computing infrastructure to train, Microsoft says the model is extremely effective to run. It creates forecasts in seconds compared to traditional hours they receive using a supercomputer material.
Microsoft, who has made the source code and the model available to the public, says it incorporates Aurora’s AI model into the MSN Weather application through a specialized version of the model that produces hourly forecasts, including clouds.
