Reliable forecasts of photovoltaics (PV) power are required for power marketing and grid operation. However, these forecasts are of low quality in many situations. In particular, the development of snow cover and fog is difficult to predict. The aim of this project is to improve PV forecasts in situations with fog or snow on PV installations. To this end, satellite-based detection and nowcasting, statistics-based very short-term forecasting, are to be improved and employed as a basis for better PV forecasts.

Our contributions to the project:

• Detection of fog and snow from satellite data using phyics-based techniques and machine learning
• Nowcasting of fog dissipation based on environmental conditions using statistical techniques / machine learning

Publications & Resources