Microsoft improves weather apps with AI: Microsoft's AI technology leads to more accurate weather apps

REDMOND/MUNICH (IT BOLTWISE) – Microsoft's latest artificial intelligence models could revolutionize the accuracy of weather forecasts.

AI is increasingly being integrated into various areas of Microsoft and Windows, including the Weather app. The Weather from Microsoft Start team works with five different AI models from three different deep learning architectures to improve weather forecasts up to 30 days out. These models are already showing impressive results compared to data from the European Center for Medium-Range Weather Forecasts (ECMWF).

A new preprint of Microsoft research, published at Cornell University, shows that modern deep learning methods can predict the weather faster and more accurately than traditional numerical weather prediction (NWP) models. Microsoft's AI models beat the ECMWF's one-week forecast by 17 percent and the four-week forecast by 4 percent. These improvements are due not only to advanced algorithms, but also to the incorporation of decades of weather observations into the data analysis.

The development of this technology could lead to significant improvements in weather forecasting, especially for forecast periods exceeding ten days. Microsoft's 30-day forecast could soon be an important addition to the world's leading weather models.

Microsoft's weather apps and widgets, including those on Windows 10 and Windows 11, were ranked as the “most accurate weather forecasting service” in an independent study by ForecastWatch. The use of the new AI models has not yet been confirmed, but their development clearly shows the potential of this technology.

notice: Parts of this text may have been generated using artificial intelligence.

Please send any additions and information to the editorial team via email to de-info[at]it-boltwise.de