In the field of weather forecasting, new technological innovations are quietly taking place. The DeepMind team under Google has developed an artificial intelligence weather forecasting program named GenCast, and its forecasting performance has already surpassed that of the current top weather forecasting system - the ENS forecasting system of the European Centre for Medium-Range Weather Forecasts (ECMWF). According to research, GenCast has improved the prediction accuracy for daily weather and extreme weather events by 20% compared to ENS.

The main advantage of GenCast lies in its fast and efficient forecasting ability. Traditional weather forecasting relies on complex physical models and needs to run on supercomputers for several hours. However, GenCast, after being trained on 40 years of historical meteorological data from 1979 to 2018, can complete a 15-day weather forecast in just eight minutes. It can predict global weather changes within a 28-square-kilometer area at 12-hour intervals.

In a comparative experiment, GenCast performed better than ENS in predicting tropical cyclones and their landing locations, especially in predicting extreme events. This ability has important reference value for industries related to energy and so on. Although currently GenCast is mainly used as an auxiliary tool for traditional weather forecasting, its accuracy and efficiency mark an important turning point in weather forecasting technology.

Currently, GenCast is one of the latest achievements of Google's continuous efforts in advancing AI technology for weather forecasting. Last year, Google also launched NeuralGCM, which combines AI with traditional physical models, and GraphCast, which provides a single best prediction. GenCast further enhances the reliability of forecasts by generating more than 50 kinds of weather forecasts and assigning probabilities to different weather events.

The meteorological community has shown recognition for this technological progress. A chief forecaster from the UK Met Office called it "exciting work", and a spokesperson for the European Centre for Medium-Range Weather Forecasts also recognized it as an "important advance". However, some experts have pointed out that although the performance of GenCast is encouraging, it is still necessary to pay attention to whether it has sufficient physical realism to deal with uncertainties in weather forecasting.

Although AI weather forecasting technology has shown great potential, experts say that there is still a long way to go before replacing traditional physical models, and further research is needed in the future to solve related scientific problems.