Weather intelligence technology startup Meteomatics AG today revealed that it’s partnering with Nvidia Corp. to enable more hyper-localized, up-to-the-minute weather forecasts with unprecedented accuracy, in what it says will be a game-changing move for sustainable energy companies.

The startup is already capable of delivering hyper-local weather forecasts several hours before the predicted conditions unfold, using physics-based weather forecasting models. But by leveraging Nvidia’s cloud-based application programming interfaces, it now has access to more powerful models that will generate extremely accurate weather forecasts in minutes.

At Nvidia’s GTC 2024 event today, Meteomatics said it will collaborate with the energy providers Engie SA, Louis Dreyfus Co. and TotalEnergies SE on a series of pilot projects that will showcase its ability to deliver weather forecasting with minutes instead of hours. Once those projects conclude, it then plans to make its AI-powered weather data delivery service available globally to companies in the renewable energy forecasting, insurance and trading industries.

The startup explains that energy companies are becoming more reliant on renewable energy sources such as hydropower, solar and wind in order to meet the growing demand for more sustainable power. But renewable energy sources are often dependent on daily weather conditions and susceptible to seasonal variations.

That’s a problem, because even a minor fluctuation, such as a little less sunlight than usual, can significantly impact how much energy is produced. As a result, energy providers need a more reliable and accurate way to forecast shifts in the weather conditions, so they can optimize their operations and ensure they’re generating enough power to fulfill demand.

By using Nvidia’s generative AI models, Meteomatics says it can deliver two-kilometer-scale weather forecasts based on real-time data with incredible accuracy, getting it right to within a 90-meter radius. The AI forecasts enhance Meteomatics’ existing weather forecasts, which are generated by a combination of physics-based models and data collected by its “Meteodrones” — which can fly up to six kilometers in altitude — plus other data sources such as radar, satellites and weather stations.

By integrating Nvidia’s AI, the company says, it can adjust its forecasts according to numerous factors, such as the difference in temperatures between the peaks and valleys of mountains, for example, to improve accuracy. The resulting downscaled weather forecasts are then made available through the Meteomatics Weather API.

The startup said the main advantage of integrating Nvidia’s AI modeling is it reduces the time it takes to calculate its forecasts. Whereas traditional modeling techniques require several hours, Nvidia’s AI can perform these calculations in a matter of minutes. Moreover, they’re more accurate too, with conditions on the ground reliably forecast to within 90 meters.

Nvidia Data Center Product Manager Dion Harris said Meteomatics’ software is critical, because almost every business in the world is impacted by the weather conditions. “Our work with Meteomatics will bring AI-powered weather forecasts to energy companies, and soon to other industries, more quickly and accurately,” he said.

Image: Freepik