One researcher with a desktop computer will be able to provide precise weather forecasts using a new weather forecast approach, which is dozens of times faster and uses thousands of times the power of computing is less than traditional systems.
Weather forecast is currently created through a complex set of stages, each takes several hours to operate on allocated giant computers, which requires large teams of experts to develop, maintain and publish them.
Aardvark Weather provides a scheme to replace the entire process by training artificial intelligence on raw data from weather stations, satellites, weather, ships and aircraft from all over the world to enable them to provide predictions.
This provides the possibility of extensive improvements in prediction speed, accuracy and cost, according to Published search On Thursday in nature from the University of Cambridge, the Alan Toring Institute, Microsoft Research and the European Medium Range Weather Center (ECMWF).
Richard Turner, Professor of Automated Learning at Cambridge UniversityHe said the approach can be used to provide detailed predictions for specific industries or sites, for example predicting African agricultural temperatures or wind speeds for a renewable energy company in Europe.
This contrasts with traditional weather forecasting systems, as it takes a dedicated years of work system by large teams of researchers, while giant computers take hours to process measures from the real world in order to build prediction models.
“This is a completely different approach to what people have done before. Writing on the wall that this will turn things, the new way to do is.” He said the model will eventually be able to produce accurate predictions for eight days, compared to forecasting for five days at the present time, as well as local excessive predictions.
The achievement can “weaken prediction” by providing strong technologies for developing countries all over the world, as well as helping policymakers, emergency and industries that depend on accurate weather predictions.
Dr. Anna Allen, the main author of the paper, from the University of Cambridge, indicated that the results paved the way for better expectations of natural disasters such as hurricanes, forest fires and hurricanes, as well as other climatic issues such as air quality, ocean dynamics and marine ice predictions.
After promoting the newsletter
Aardvark depends on the recent research conducted by Huawei, Google and Microsoft that shows that one step from the prediction process known as the numerical Solver, which calculates how the weather develops over time, can be replaced by AI to produce faster and more accurate predictions. This approach is already published by ECMWF.
The researchers said that only 10 % of the input data required by the current systems can actually outperform Aardvark on the American national prediction system in some respects, and it was a competitor with weather services expectations in the United States.
