All Sky Cameras

DNICast Advances

- Detection of cloud type, cloud height and speed in specific areas of the image.

- Both subsystems and the whole nowcasting system has been validated.

- Temporal and spatial aggregation significantly reduces deviations, as for field averages and 10 min averages. A RMSE below 10% is reached.

Forecasting of cloud induced variability can be achieved by all-sky imager (ASI) based system. The low spatial and temporal resolutions of satellite derived forecasts make it difficult to predict data with a spatial resolution corresponding to subfields (500mx500m) or even single collectors (10mx10m). The nowcasting methods developed in DNICast use the inputs of off-the-shelf surveillance cameras and at least one ground measurement station for Direct Normal Irradiance (DNI). Without available ground measurements, modelled clear sky irradiance data can be used. Every 30 seconds, four ASI images are taken. In these images, clouds are detected and the 3-D positions of all visible clouds are derived. By tracking clouds over multiple timestamps, cloud velocities are determined and used to predict future cloud movements. With the 3-D shapes and positions of all clouds known, their shadows are projected on a ground model. If available, reference ground measurements are used to determine cloud transmittances. With transmittances known, the shadow maps are transformed to irradiance maps. The cameras allow to predict DNI maps for an area of 8x8 km2. Spatial and temporal resolutions for these maps are 25 m2 for every 30 seconds for predictions up to 15 min ahead. The working principle is illustrated in Figure 1.



WobaS system_demonstrator.png
Figure 1: Working principle of the DNICast-based WobaS system. All-sky imagers are used to derive future irradiances in high spatial and temporal resolutions.

References

Bijan Nouri, Pascal Kuhn, Stefan Wilbert, Christoph Prahl, Robert Pitz-Paal, Philippe Blanc, Thomas Schmidt, Zeyad Yasser, Lourdes Ramirez Santigosa, and Detlev Heinemann. 2017. Nowcasting of DNI Maps for the Solar Field Based on Voxel Carving and Individual 3D Cloud Objects from All Sky Images. In SolarPACES. Santiago de Chile.

Blanc, Philippe, Pierre Massip, Andreas Kazantzidis, Panagiotis Tzoumanikas, Pascal Kuhn, Stefan Wilbert, David Schüler, and Christoph Prahl. 2017. "Short-term forecasting of high resolution local DNI maps with multiple fish-eye cameras in stereoscopic mode." AIP Conference Proceedings no. 1850 (1):140004. doi: 10.1063/1.4984512.

Kuhn, Pascal, Bijan Nouri, Stefan Wilbert, Christoph Prahl, Nora Kozonek, Thomas Schmidt, Zeyad Yasser, Lourdes Ramirez, Luis Zarzalejo, Angela Meyer, Laurent Vuilleumier, Detlev Heinemann, Philippe Blanc, and Robert Pitz-Paal. 2017. "Validation of an all-sky imager based nowcasting system for industrial PV plants " EUPVSEC 2017, also accepted for publication in Progress in Photovoltaics.

Kuhn, Pascal, Stefan Wilbert, David Schüler, Christoph Prahl, Thomas Haase, Lourdes Ramirez, Luis Zarzalejo, Angela Meyer, Laurent Vuilleumier, Philippe Blanc, Jean Dubrana, Andreas Kazantzidis, Marion Schroedter-Homscheidt, Tobias Hirsch, and Robert Pitz-Paal. 2017. "Validation of spatially resolved all sky imager derived DNI nowcasts." AIP Conference Proceedings no. 1850 (1):140014. doi: 10.1063/1.4984522.