A Spatially Resolved Clear-Sky Filter using Photovoltaic Modules as Cloud Detectors

Operation and maintenance of photovoltaic (PV) power plants is essential to mitigate performance losses. Time series of the temperature corrected performance ratio (PR) can be used to monitor system performance or as input to advanced algorithms that estimates degradation and soiling. However, these algorithms are sensitive to the signal-to-noise level in PR time series, and might struggle if there are much missing data. Traditional clear-sky filters often discard large amounts of data, particularly during shifting weather conditions. This study introduces a novel clear-sky detection method that retains more data by using PV modules as cloud detectors, yielding a spatially resolved clear-sky index. The method proposed here is compared to the one by Reno and Hansen which is included in the open-source python library PVAnalytics. The results show that the new filter retains more data, especially on partially cloudy days, leading to more reliable performance metrics for the PV system.