Google DeepMind has announced the availability of a new model of Artificial intelligence (AI) called GenCast, aimed at predicting weather and making advanced forecasts of extreme weather events to protect more lives, prevent damage and save costs.
Google’s AI development company has recognized that more accurate forecasts of natural phenomena that may endanger society, such as heat waves or strong winds, allow timely and cost-effective preventive measures to be taken.
LOOK: Jared Isaacman, the millionaire astronaut chosen by Donald Trump to lead NASA
To offer an accurate service in real time, it has developed GenCast, a new high-resolution AI model that can predict atmospheric conditions up to 15 days in advance, and which has recently been published in the journal Nature.
This model, which the company plans to make available to meteorology professionals soon, provides better predictions of both daily weather and extreme temperatures than the European Center for Medium-Range Weather Forecasts (ECMWF).
This novelty, based on GraphCast, comprises a set of 50 or more predictions, each of which represents a possible weather trajectory. Likewise, it is adapted to the spherical geometry of the Earth and learns to accurately generate the different probable weather scenarios of the future by indicating the most recent state of the weather in them.
The company has noted that, to train GenCast, it has provided it with four decades of historical weather data from the ECMWF’s ERA5 archive – an analysis of weather spanning from January 1940 to the present. This information includes variables such as temperature, wind speed and pressure at different altitudes.
This division of Google has raised the use of GenCast in situations such as hurricanes or typhoons, where “getting better and more advanced warnings about where they will make landfall is invaluable.” In this sense, their model offers superior predictions to the trajectories of these deadly storms.
Google DeepMind has also recognized that, in addition to improving weather predictions to protect more lives, avoid damage and save costs, models like this could play a fundamental role in other aspects of society, such as planning the use of renewable energy.
GenCast, whose code is public, is part of Google’s suite of next-generation AI-based weather models, aimed at improving the forecast of precipitation, flooding or extreme heat situations. These tools include NeuralGCM, SEEDS, and Google Research flood models.