AI model for weather forecasting: Google’s DeepMind team announced an AI model for weather forecasting called GenCast that outperforms the world’s top weather prediction systems. In a paper published in the journal Nature, DeepMind researchers said GenCast outperformed the world’s top operational forecasting system, the European Center for Medium-Range Weather Forecasts (ECMWF).
“New AI models advance predictions of weather uncertainty and risk, delivering faster, more accurate forecasts up to 15 days in advance,” Google said in a statement.
According to the tech giant, GenCast represents a significant advancement in AI-based weather forecasting, building on previous weather models that are deterministic and provide a single best forecast of future weather.
In contrast, GenCast forecasts consist of an ensemble of 50 or more forecasts, each representing an expected weather trajectory. GenCast is a diffusion model, a type of generative AI model that powers recent rapid advances in image, video, and music generation.
“However, GenCast is unique in that it adapts to the Earth’s spherical shape and learns how to accurately generate complex probability distributions of future weather scenarios when given the latest weather conditions as input. ,” Google said.
By more accurately predicting the risks of extreme weather events, authorities can save more lives, avoid damage, and save money. “Think about tropical cyclones, also known as hurricanes and typhoons. Having better and more sophisticated warnings about where they will make landfall is invaluable. GenCast tracks the path of these deadly storms. Google provides better predictions about
The company will soon release real-time and historical forecasts from GenCast and previous models, allowing anyone to integrate this weather information into their own models and research workflows.
GenCast is part of Google’s growing suite of next-generation AI-based weather models, including Google DeepMind’s AI-based deterministic medium-range forecasts and Google Research’s NeuralGCM, SEEDS, and flood models.
