Application of AI revolutionize weather forecasts

A researcher with a desktop can provide accurate weather forecasts by using the new prediction method with new AIs dozens of times faster and use less calculated capacity thousands of times more than conventional systems, Guardian Report. The current weather forecast is created through a series of complex stages, each period takes a few hours to run on supercomputers, need a large team of experts to develop, maintain and deploy.

Aardvark Weather provides a detailed plan to replace the entire process by training AI based on raw data from meteorological stations, weather balloons, ships and aircraft around the world to make predictions. This provides significant improvement of speed, accuracy and expense, according to research published on March 20 in the journal Nature from the University of Cambridge, Alan Turing Institute, Microsoft Research and European Mid -Range Weather Forecast Center (ECMWF).

Richard Turner, a machine professor at the University of Cambridge, said the above method can be used to quickly provide forecasts for specialized industries or specific locations, for example, temperature prediction for African Agriculture or wind speed for renewable energy company in Europe. That contrasts the traditional weather forecast system, in which the creation of a custom system requires many years of development from the researcher, and supercomputers takes hours to handle actual measurement results to build a predictive model.

This is a completely different method from what everyone has done before, according to Turner. He said the new model can create an accurate forecast 8 days in advance, compared to the current 5 days as well as make superior localization predictions.

Dr. Scott Hosking, Director of Science and Environmental and Sustainable Achievements at the Alan Turing Institute, said the breakthrough who could support policy makers, emergency response plans and industry depending on the exact weather forecast. Dr. Anna Allen, a team leader at the University of Cambridge, emphasized that he will pave the way to better forecast natural disasters such as storms, wildfire and tornadoes, as well as many other weather issues including air quality, ocean mobilization and marine prediction.

Aardvark Weather was born based on Huawei, Google, and Microsoft’s recent research in the process of weather forecasting, which is evolutionary weather calculation over time, which can be replaced by AI to make faster and more accurate predictions. This method has been deployed by ECMWF. The team said only 10% of input data required by the current systems, Aardvark Weather can work better than the US National GFS forecast system in some sides and compare to the US weather forecast.