Accuracy Evaluation of Horizon Shading Estimation Based on Fisheye Sky Imaging

Abstract

Measuring horizon shading and modeling available irradiance at a prospective site is necessary for accurate estimation of the energy yield of photovoltaic (PV) systems, as well as the expected operation and availability of PV powered products. Fisheye sky imaging is a relatively simple approach to characterize the surrounding horizon, however it's accuracy in estimating the shading loss on short timescales is not well quantified in the literature. In this work we evaluate the shade irradiance loss estimation accuracy of a horizon shading model based on fisheye sky images compared to actual local irradiance measurements from the site of interest. The results show that small errors in he horizon line estimation can lead to high irradiance estimation errors, for short timescales. However, these are mostly averaged out if the sampling period is 15 minutes or higher. The horizon shading maps obtained both with a commercial shading analysis tool, as well as self-calculated from raw fisheye images, tend to overestimating the direct beam shading and small relative errors are observed. Furthermore, ground reflected irradiance has a great influence on vertical surfaces, which was not accounted for in this work.