Accurate identification of clear-sky and cloudy periods from hyperspectral radiometric data is essential for atmospheric and remote sensing studies. Traditional approaches that rely on single-wavelength spectral ratios often fail under broken-clouds due to rapid irradiance fluctuations and aerosol interference. This study introduces a new method based on the spectrally integrated diffuse-to-direct irradiance ratio (IDR), which is derived from modeled hyperspectral components and provides a robust indicator of sky conditions. The model is developed using 15-year hyperspectral measurements of solar downwelling irradiance (ES(λ), mW m-2 nm-1) and sky radiance (LS(λ), mW m-2 nm-1 sr-1) collected in a fixed coastal station located at the Marsdiep tidal inlet in the Dutch Wadden Sea. The model is validated against reference clear-sky periods identified using established clear-sky models and detection algorithms based on simultaneous Global Horizontal Irradiance (GHI) measurements. Results show that IDR thresholds of ≤ 0.35 for clear skies and ≥ 0.70 for overcast conditions achieve high agreement (95% and 92%, respectively) with reference sky classifications. The IDR effectively accounts for the influence of rapidly changing cloud cover in intermediate conditions, ensuring robust classification of clear-sky and overcast periods. The classification is further supported by spectral analysis using Principal Component Analysis, which confirm the stability of diffusive component to ES(λ) features and the interferences of LS(λ)/ES(λ) ratio across different sky conditions. This work highlights the importance of the diffuse irradiance component role in hyperspectral sky classification. Future research should validate IDR with direct diffuse irradiance measurements to enhance the proposed model under diverse atmospheric conditions. |
