NVIDIA StormScope EU Updates Weather Forecasts Every 10 Minutes for Energy and Insurance Sectors
Developers in Europe are deploying open AI models from NVIDIA to generate high-resolution weather predictions for energy, insurance and transport sectors.
Key points
- NVIDIA released StormScope EU, a regional model trained on European data for short-term forecasting.
- Organizations like AXA and TotalEnergies are using these tools to assess cyclone risk and solar output.
- The system updates predictions every ten minutes using real-time satellite observations.
European organizations are deploying NVIDIA Earth-2 open models to generate localized weather and climate predictions. These tools allow developers to customize global forecasts with regional data, supporting decisions for energy producers, insurers and transport operators who need precise local conditions rather than broad global trends.
The risk
Accurate local weather data is critical for operational planning across multiple industries. A global forecast may indicate a storm, but specific decisions about solar farm output, shipping routes or insured property require granular local details. NVIDIA Earth-2 provides a family of open AI models that developers can customize with their own data to address these specific regional needs.
Who is exposed
Several European entities are actively building on these capabilities. AXA uses FourCastNet 3 to study regional cyclone risks across simulated seasons. Vaisala Xweather combines Earth-2 with proprietary lightning data to enhance forecasting for government and maritime clients. TotalEnergies evaluates solar-radiation nowcasts to anticipate changes in power generation at photovoltaic sites.
Cost and compliance
The University of Manchester researchers adapted the CorrDiff model for air-quality modeling, training it on the Isambard-AI supercomputer and running it locally on DGX Spark systems. Space42 developed SARDiff4Height using CorrDiff to estimate surface heights from radar images. These applications demonstrate how teams can incorporate proprietary data and domain expertise to meet specific regional requirements without relying solely on external services.
What should you do?
Teams should evaluate whether their current forecasting workflows can benefit from open-source AI models that integrate with proprietary data. Managers should check if their infrastructure supports the computational requirements for running these regional models. Organizations should verify that the model outputs align with their specific operational needs for short-term prediction accuracy.
The background in plain language
AI weather models use historical data to predict future conditions. Open models allow organizations to train these systems with their own local data, improving accuracy for specific regions. This approach enables teams to move from generic global forecasts to precise local predictions, supporting better decision-making for time-sensitive operations like energy management and risk assessment.
What to do and how to stay safe: NVIDIA
- Review current weather data sources to identify gaps in local resolution.
- Assess internal compute resources for running open-source AI models.
- Test integration of proprietary observational data with open model architectures.
- Monitor vendor updates for new regional model releases and performance improvements.
Step-by-step guide: Browser security settings are the architectural walls of your digital life
General security guidance from the Vector Tribune newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
What is NVIDIA Earth-2?
It is a family of open AI models and tools that developers can customize with their own data for weather prediction and environmental research.
How often does StormScope EU update its forecasts?
The model predicts evolving cloud conditions at ten-minute intervals using EUMETSAT satellite observations.
Which organizations are using these models?
Users include AXA, Vaisala Xweather, the University of Manchester, TotalEnergies and Space42, each applying the tools to specific regional challenges.
