Cloud cover (CC) plays a key role in atmospheric radiative transfer processes, the surface energy budget, and the retrieval of aerosol and cloud products. Whole-sky imagers (WSI) provide high-frequency, hemispherical observations well suited for CC estimation. Here, we present a comprehensive analysis of the probability density function of the Blue-to-Red Ratio (PDF BRR ) using a five-year WSI dataset. We quantify how clouds and aerosol load and type shape the PDF BRR , identifying robust radiative signatures across clear-sky, partly-cloudy, and overcast conditions. Results show that cloud-clear discrimination is encoded in a limited set of PDF BRR statistics while also revealing conditions in which aerosol and cloud signatures become radiatively indistinguishable (‘confusion zones’). These findings define the physical limits of BRR approaches, demonstrating that the PDF BRR provides a consistent basis to resolve aerosol-cloud ambiguity under well-defined atmospheric conditions. Building on these insights, we introduce CLARIS (Cloud–clear Adaptive Retrieval using Image-based Statistics). CLARIS is developed as a foundational methodological tool to validate and consistently evaluate the identified BRR signatures through objective pixel-level classification. Validated against an expert-supervised Reference Image Database (RID), CLARIS correctly categorizes clear-sky and overcast scenes, achieving a 90% pixel-level agreement in clean partly-cloudy conditions and 87–91% in polluted partly-cloudy conditions, where aerosol-cloud ambiguity is stronger. Across the full dataset, CLARIS reaches 97% pixel-level agreement. Beyond CC detection, our findings highlight the diagnostic potential of the PDF BRR for characterizing radiative behaviors and transitions associated with aerosol-cloud interactions.
Valdelomar et al. (Mon,) studied this question.