Accuracy surpasses traditional weather forecasts! Google releases AI weather prediction model GenCast
Google has released an artificial intelligence weather forecasting model called GenCast, which surpasses traditional weather forecasts in accuracy. The model is more accurate than the ECMWF model 97.2% of the time, and its accuracy reaches 99.8% for forecasts beyond 36 hours. GenCast's forecasting speed has also significantly improved, capable of generating a 15-day forecast in 8 minutes, while traditional models take several hours. Google hopes to collaborate with the meteorological community to enhance predictions for agriculture and extreme weather events
According to Zhitong Finance APP, Google (GOOGL.US) has announced an artificial intelligence weather forecasting model named GenCast, which outperforms traditional weather forecasting models in accuracy.
The model forecasts 15-day weather more quickly and accurately than other weather forecasting systems, including the European Centre for Medium-Range Weather Forecasts (ECMWF) system. In a study published in the journal Nature, GenCast was found to be more accurate than the ECMWF model 97.2% of the time, with an accuracy rate of 99.8% when the forecast time frame exceeds 36 hours.
A blog post from DeepMind, Google's artificial intelligence research organization, stated: "Ensemble forecasting expresses uncertainty by making multiple predictions across different possible scenarios. If most predictions indicate that a hurricane will hit the same area, then the uncertainty is low. But if they predict different locations, then the uncertainty is higher. GenCast achieves an appropriate balance, avoiding exaggeration or underestimation of its confidence in the forecasts."
GenCast's forecasting speed is also significantly faster than traditional weather models. For example, a single Google Cloud TPU v5 can generate a 15-day forecast in 8 minutes while producing more predictions. In contrast, traditional physics-based ensemble forecasting on supercomputers takes several hours.
Google stated that improved forecasting benefits agriculture and helps predict extreme weather events such as typhoons and hurricanes.
Google said: "We are eager to collaborate with a broader meteorological community, including academic researchers, meteorologists, data scientists, renewable energy companies, and organizations focused on food security and disaster response."