Microsoft researchers have introduced a new machine learning tool designed to make weather forecasts more useful weeks in advance. Traditional physics-based systems and many newer AI models tend to lose reliability beyond about two weeks, but Microsoft says its approach improves predictions in the crucial two-to-six-week range.
The potential benefits are practical and wide-reaching. Better long-range forecasts can help farmers decide when to plant or irrigate, give emergency responders more time to prepare for extreme weather, and help communities plan around heat, storms, drought, or flooding with greater certainty.
Open science for better forecasting
The research was published in Nature, and Microsoft has released the tool as open-source software. That means weather scientists, public agencies, and other AI researchers can inspect, test, and build on the work rather than waiting for a closed commercial product.
Microsoft’s project joins a fast-growing wave of AI weather forecasting efforts from companies like Google and Nvidia. Together, these advances point toward a future where forecasts are faster, cheaper, and more actionable—especially in the timeframes where better information can make the biggest difference.