Artificial intelligence revolutionizes the field of meteorology
A new forecast system of the weather based on artificial intelligence (AI) has the potential to revolutionize the field of meteorology, generating forecasts dozens of times faster than the models rolled on supercalculating and using only a fraction of their calculation power, according to a study published on Thursday, Nature magazine, reports Live Science on Friday.
The new system, called Aardvark Weather, is based on a learning algorithm and uses the gross data obtained from satellites, weather stations, vessels and weather balloons to make forecasts, without relying on atmospheric models.
« The weather forecast systems that we all rely on have all been developed over the decades, but it took only 18 months to build something that is competitive even with the best of these systems and which uses only a tenth of data on an ordinary office, » said Richard Turner, an engineer at the University of Cambridge.
Current weather forecasts are generated by entering data into complex physical models, a multi -stage process that requires several hours of dedicated operation.
The new AI -based approach can provide major advantages in terms of costs, speed and accuracy of weather forecasts. Instead of needing a supercomputer and a dedicated team of researchers, the Aardvark Weather system can generate a forecast using the power computer power in just a few minutes.
Thus, using only 8% of the observational data needed for a traditional forecast system, Aardvark Weather has performed the US Global Forecast System (GFS) system and had comparable results with the forecasts issued by the United States Weather Service.
