Microsoft has announced the company’s new artificial intelligence model called Aurora It can predict the quality of air, storms, typhones and other atmospheric phenomena.
According to a report published this week in the Nature scientific journal and also reflected in a blog post, Microsoft has described the performance details of the model. According to the company, Aurora can predict atmospheric events far higher than traditional meteorological methods.
Aurora Microsoft’s advanced artificial intelligence model in the field of air forecasting
This model is trained with more than one million data hours, including satellite information, radar, meteorological stations, simulations and previous forecasts. It is also highly adjustable and can be optimized by adding new data to more accurate prediction of specific climate events.
Although artificial intelligence models are not a new phenomenon in the field of air forecasting, and companies such as Google Dipmind have introduced models such as Weathernext for years, Microsoft is one of the most powerful models in the field. This model can be used in the form of an effective tool for active research centers in the field of atmospheric sciences.
Microsoft has announced that Aurora has been able to predict the when the Doksuri storm arrives at the Philippines 7 days before it happened; This result has also been more accurate than some expert predictions. It also performed better than the US National Center for Storms in the forecast of tropical storms from 2022 to 2023, and has successfully predicted the occurrence of 2022 sandstorms in Iraq.
Despite the need for extensive computing infrastructure for training, the Aurora model is very optimistic and fast. Traditional systems require watches using superconductors, but it can predict in seconds. Microsoft has also announced that it has released the source code and weight of the model and is currently using a special version of the model in the MSN Weather app to provide hourly forecasts, including cloud status.
RCO NEWS




