Home / Science / Startup's AI Outperforms European Weather Giant
Startup's AI Outperforms European Weather Giant
1 Jun
Summary
- New AI weather model offers more frequent and accurate predictions.
- Model outperforms world-leading European government system.
- Company leverages unique data collection with 400 balloons.
Windborne Systems, a startup founded by Stanford students, has released WeatherMesh 6, an advanced AI weather forecasting model. This new version reportedly surpasses the accuracy of established systems, including those from the European Centre for Medium-Range Weather Forecasting (ECMWF).
The model provides hourly forecasts, a significant increase over the six-hour intervals of traditional systems. Its resolution reaches 3 km in regions with the highest data quality, such as Europe and the continental US. This performance is driven by Windborne's unique approach, which integrates data from its own network of around 400 weather balloons.
Unlike traditional physics-based models requiring supercomputers, Windborne's AI leverages deep learning and direct data assimilation from its sensors. This direct ingestion of data is cited as the primary reason for WeatherMesh 6's improved capabilities. The company has raised $25 million in venture funding, with its data already being utilized by entities like NOAA and the U.S. Air Force.
