New model to predict cloud movements, improve grid integration for renewables

From pv magazine Spain

Spain's Meteo for Energy offers weather forecasts and energy production models for photovoltaic, solar thermal, and wind power generation. It uses cloud cameras and satellite image predictions based on Meteosat images, as well as predictive artificial intelligence models.

The company used cloud cameras for the nowcasting of cloud transients and uses Meteosat satellite image predictions to make short-term predictions about solar radiation, in order to integrate solar production into the continuous market. Precipitation can be displayed in real time, along with the forecasting of suspended dust to prevent soiling.

AI predictive models combine weather data with other information to generate highly accurate forecasts of weather conditions and expected energy production under different conditions, to facilitate better integration into the grid.

A Meteo for Energy spokesperson told pv magazine that the company's predictive models have an accuracy of 5% to 8% normalized mean absolute error (NMAE) for 48-hour meteorological models. The satellite can reportedly increase accurary by around 12% for a window of two hours.

“Meteosat satellite images and Deep Learning play a key role in ensuring a better integration of solar energy into the electrical grid, as this type of technology allows the photovoltaic and solar thermal sector to predict the movement of the clouds to improve their operation and reduce their diversion costs,” the spokesperson said.

Meteo for Energy said the entire process offers a minute-by-minute prediction of the two main factors that affect the integration of solar energy: cloud movement and the opacity index for each image coordinate.