The new model, dubbed 'NeuralGCM,' has proven to be more accurate than other purely machine learning-based models for one- to 10-day weather forecasts. It also excels in projecting climate conditions over decades.
NeuralGCM offers computational power and timeliness advantages over traditional models, running relatively quickly on a laptop compared to hours on supercomputers processing thousands of lines of code.
Developed collaboratively by research teams including Google, DeepMind, MIT, Harvard, and European Center for Medium-Range Weather Forecasts, the model integrates machine learning with neural networks and physics equations.
The novel aspect of NeuralGCM is its combination of large-scale physics with AI components, in contrast to AI models from other companies that omit physics entirely, as explained by meteorology professor Aaron Hill.
[
Collection
]
[
|
...
]