Development of All-Sky Imager quality control methods and their evaluation in Eye2Sky nowcasting

Abstract

Accurate prediction of the energy production of photovoltaic systems requires precise weather forecasts of solar irradiance. Numerical weather prediction models and satellite-based forecast methods provide solar irradiance forecasts only with limited temporal and spatial resolution. All-sky imager systems can provide solar irradiance forecasts with high resolutions in the order of metres and minutes. The focus of this work is on the all-sky imager network Eye2Sky, a camera network located in the north-west of Germany, whose topology is unique due to its size and camera density.

This work addresses the question of how errors of all-sky imagers, e.g. soiling or interfering objects, do affect the interim and final results of the solar irradiance forecast. The physics-based forecast algorithm WobaS was the focus of this analysis. More precisely, the effect of errors on the cloud masking process, on the cloud base height estimation, as well as on the resulting irradiance maps of WobaS were analysed. Based on the findings of this analysis, methods for the detection and prevention of errors have been developed and validated. Finally, recommendations are given for the operation of all-sky imager networks with regard to quality control.