Researchers from Switzerland and the Netherlands have developed a new framework for intraday spatiotemporal PV power prediction at the national scale, combining satellite-based deep learning and optical-flow approaches with physics-based numerical weather prediction models. This method allows for improved short-term power production planning, strategic action during energy surpluses or shortages, reduced electricity costs, and better balance of the energy grid. The satellite-based models outperformed traditional forecasting methods, with relative errors below 10% on 82% of days in 2019-2020, demonstrating potential for operational use in climate solutions.
https://www.pv-magazine.com/2026/07/28/open-access-framework-for-intraday-pv-power-prediction/