Grid operators across Europe face a growing blind spot. As rooftop and small-scale solar installations multiply, transmission system operators (TSOs) such as France's RTE find it harder to know how much electricity these distributed assets are actually producing at any given moment. Most small PV systems do not report production data in real time. The grid operator must guess.
That uncertainty is expensive. When solar output is unpredictable, operators must hold larger reserve margins: backup power capacity that can be called on within minutes if supply falls short of demand. Larger margins mean higher costs, more idle conventional capacity, and slower progress towards decarbonisation. As PV capacity grows, so does the problem.
The challenge is not limited to national grid operators. Energy companies managing distributed assets at commercial and industrial sites face a related problem: without accurate short-term forecasts of PV output, they cannot optimise when to store, consume, or curtail solar generation on site. Poor forecasts lead to wasted renewable energy, unnecessary grid purchases, and clumsy curtailment decisions.
[Image: Satellite-derived cloud tracking for solar nowcasting. Credit: Reuniwatt/EUSPA]
How EO can help
The Copernicus Atmosphere Monitoring Service (CAMS) provides satellite-derived solar radiation data at high temporal resolution, covering all of Europe with records going back to 2004. Combined with temperature and wind speed estimates from the ERA5 reanalysis dataset produced by ECMWF under the Copernicus Climate Change Service (C3S), these data make it possible to model expected PV output at any location, for any panel configuration, in near-real time.
Rather than relying on static annual averages, operators can use dynamic, intraday estimates of solar generation across their control zones. The result is a far more accurate picture of how much distributed solar is feeding into the grid at any moment, and better forecasts of what comes next.
Copernicus data also powers a different but complementary approach: solar nowcasting. By combining geostationary satellite imagery with CAMS atmospheric variables (aerosol depth, water vapour, radiation), forecasting services can track cloud movement in near-real time and predict PV output over the next minutes to hours. These short-horizon forecasts, refreshed every 15 minutes, feed directly into energy management systems that steer on-site assets such as batteries, heat pumps, and EV chargers. The practical effect: smarter curtailment decisions, better use of locally generated solar power, and less reliance on grid imports.
Key examples
Example 1: ARMINES and RTE (France)
Paris-based research institute ARMINES, working under a EUSPA proof-of-concept initiative, built an interactive GIS-based platform that delivers high-resolution solar mapping in near-real time. The tool goes well beyond conventional solar cadastres, which typically offer only annual averages of rooftop irradiation. It computes intraday PV output on tilted planes, covering not just rooftops but parking canopies, roadside spaces, and open ground near buildings.
A distinctive feature is the platform's ability to estimate shading effects on solar generation, a factor that static models routinely ignore. By drawing on CAMS radiation data and ERA5 climate variables, the tool captures real-world conditions (cloud cover, temperature, wind) that affect panel performance hour by hour.
For RTE, the outcome is practical: more precise forecasts of aggregated small-scale PV production across its control zone. More accurate forecasts mean smaller reserve margins, lower balancing costs, and a more stable grid.
Example 2: Reuniwatt and Luminus (Belgium)
French solar forecasting specialist Reuniwatt, also working under a EUSPA proof of concept, built a nowcasting service for Belgian energy company Luminus. The system converts geostationary satellite imagery into cloudiness maps and applies optical flow techniques to track cloud movement, generating PV power forecasts refreshed every 15 minutes. Reuniwatt integrates CAMS atmospheric data (aerosol depth, water vapour, radiation parameters) to sharpen its irradiance predictions, particularly in rapidly changing weather.
Luminus deployed the service across three sites in Belgium, feeding the forecasts into its SmartKit energy management system, which steers PV, batteries, EV chargers, and heat pumps. The pilot showed improved irradiance forecasts closer to local conditions, especially in poor weather, and better time precision for fast-moving cloud fronts. Luminus and Reuniwatt are now evaluating a commercial roll-out.
