Innovative Forecasting Redefines Accuracy And Frequency
WindBorne Systems, a weather technology startup founded by Stanford alumni in 2019, has unveiled its sixth-generation forecasting model, WeatherMesh-6. The AI-based system provides hourly forecasts with a resolution of up to 3 kilometres across Europe and the continental United States, positioning it among the latest generation of AI-powered weather forecasting tools.
Enhanced Precision Through Advanced Data Ingestion
WeatherMesh-6 combines deep learning models with data collected from approximately 400 weather balloons operating across 15 locations worldwide. According to Chief Product Officer Kai Marshland, the system can deliver forecast accuracy comparable to a traditional one-day forecast at a five-day horizon. The improvement stems largely from advances in data assimilation and the integration of sensor readings into transformer-based forecasting models.
Follow THE FUTURE on LinkedIn, Facebook, Instagram, X and Telegram
From Weather Balloons To Data-Driven Forecasting
WindBorne originally focused on collecting atmospheric data through its proprietary weather balloon network. Following advances in AI weather forecasting beginning in 2022, the company expanded into model development, using its data collection infrastructure as a foundation for building forecasting systems. Chief Executive Officer John Dean has argued that access to proprietary weather data provides the company with a competitive advantage as the sector becomes increasingly crowded.
Overcoming Traditional Forecasting Limitations
Traditional forecasting systems rely heavily on physics-based simulations that require significant computing resources and often update only a few times per day. WeatherMesh-6 produces updated forecasts every hour, allowing faster refresh cycles than many conventional forecasting models. The company is competing with both national weather agencies and technology firms developing AI-based forecasting systems, including Google DeepMind.
Safety Innovations And Strategic Industry Impact
WindBorne has also introduced operational changes to reduce risks associated with its balloon network. After an incident involving a United Airlines aircraft and one of its weather balloons last year, the company began using the ADS-B aviation surveillance system to track nearby air traffic and adjust balloon operations when necessary. Valued at $85 million in 2024 and backed by $25 million in venture funding, WindBorne supplies weather data to organisations including NOAA and branches of the U.S. military.
Looking To The Future
Although the company continues to use datasets from organisations such as ECMWF and NOAA, Dean said WindBorne has become less dependent on external forecasting inputs as its own models improve. “When we started doing data assimilation, we were still very heavily reliant on ECMWF. I predict today, if we removed ECMWF’s initial conditions, we would actually still do pretty good,” Dean said. The company plans to continue investing in forecasting models and data infrastructure rather than expanding into consumer-facing products.
Correction And Clarification
WindBorne uses ADS-B data to monitor aircraft movements and adjust balloon operations accordingly. However, the company has not installed ADS-B transponders directly on its weather balloons.
