The latest weather forecast models, AI, GraphCast and HRES

The latest computational model developed by experts from the Commonwealth Scientific and Industrial Research Organization in Brisbane uses sea surface temperatures to predict a year in advance when whales and turtles will likely become stuck in nets and other fishing equipment. Thanks to the work of Australians, fishermen will know when to stop fishing, e.g. crabs, to save marine life that is particularly threatened by human activity.

Thanks to the development of AI and technology, weather applications are becoming more and more detailed and tailored to the individual needs of users. For example, the British Met Office has improved digital weather maps and data entry to present the situation in a selected place in real time. Loganair Airlines cooperates with the Met Office – their planes are equipped with sensors monitoring environmental conditions during flights. The Windy app helps long-distance cyclists track wind direction and choose the optimal travel time.

DeepMind’s single-chip AI model was more accurate at simulating weather forecasts than the supercomputer-based traditional HRES forecast program. GraphCast uses algorithms that analyse a compilation of a vast climate and weather dataset called “ERA5”, dating back to 1950 and developed by the European Center for Medium-Range Weather Forecasts (ECMWF). GraphCast presents weather data as a single point for a square area on Earth related to weather conditions in neighbouring areas.