The main objectives of this study were to develop the indicatri es of reflected solar radiation from different natural surfaces and to show comparisons between values sensed in space of emergent radiation to ground values obtained from accounting for anisotropic reflection and estimating the effect of the intervening atmosphere. Thus, this study demonstrated that a prior knowledge of the angular distribution of reflected radiation allows determining the true hemispherical reflected radiation from a narrow field of view instrument such as found on a flying platform. Measurements for determining the indicatrices were made from a tower-mounted Nimbus MRIR and, in one case, with a hand-held TIROS five-channel radiometer. Anisotropy of reflected radiation was found for all surfaces examined and increased with decreasing solar angle. Different surfaces showed different degrees and patterns of fonvard and backscatter. A clearly defined anti-solar point was found for plowed field, various agricultural crops and vegetated desert surfaces, while snow, the Alkali Flats, and Bonneville Salt Flats showed a broad pattern of backscatter. As a consequence of surfaces exhibiting well-defined antisolar points the anisotropic correction factors relating normal reflectances to 2π reflected values were less than 100 percent for solar angles greater than approximately 60°. All surfaces examined showed anisotropic correction factors increasing with decreasing solar angles. The albedo over the White Sands dune field decreased with decreasing solar angles due to large shadow patterns which are produced at low solar angles. The sand dunes values were derived from aircraft measurements. Comparisons were made between estimated emergent radiation from the top of the earth 's atmosphere accounting for anisotropy of the ground reflection pattern and estimates of atmospheric attenuation to values of reflected radiation obtained from the MSS subsystem of the ERTS program for the lava beds region and White Sands area in New Mexico. Also comparisons were shown between the estimated emergent radiation from the earth's surface for the same features and spectral bands to the values sensed in space. Under high albedo conditions as found in the White Sands area there was a decrease in emergent radiation to space while with low surface albedo, such as the lava beds region, the extra-terrestrial radiation increased from the ground values.

Ce document présente la mise en place d’un système de mesure des éclairements énergétiques spectraux solaires pour tout type de temps, sur 4 plans. Les 4 spectromètres mesurent au total 900 spectres/min et produisent chacun un spectre/min moyen et son écart type. Entre 2014 et 2015, 700 000 spectres ont été enregistrés sur un domaine compris entre 400 et 1000 nm avec un pas ≤1 nm. Un échantillon de 145 000 spectres représentatifs du climat lyonnais a été sélectionné pour une analyse statistique. Pour ce faire, l’échantillon a été réduit par partitionnement à 1175 spectres. Son domaine spectral a été étendu de 280 à 1500 nm à l’aide du RTM SMARTS. Une ACP de cet échantillon extrapolé a permis d’en réduire la description à 3 composantes et ainsi de réviser le modèle des illuminants D de la CIE. Enfin, la relation entre composition spectrale et paramètres environnementaux ou colorimétriques ouvre une voie vers des modèles statistiques de génération de spectres du rayonnement solaire. / This manuscript presents the design and setup of an all-weather spectral irradiance measurement system on 4 planes. The 4 spectrometers measure a total of 900 spectra/min to produce every minute, a mean spectral irradiance and its standard deviation. Between 2014 and 2015, this system recorded 700,000 spectra, for wavelengths ranging between 400 and 1,000 nm with a step ≤1 nm. A sample of 145,000 spectra representative of the Lyon climate was selected for statistical analysis. For this purpose, the sample was reduced in size by partitioning it in 1,175 spectra. Its spectral domain was extended to 280-1,500 nm by extrapolating the spectra with curve fitting using the SMARTS2 RTM. A PCA of the extrapolated sample reduced its description to only 3 components; hence, allowing a revision of the CIE’s illuminant D series. Finally, the relation between spectral power distribution and environmental or colorimetric parameters opens a way towards statistical models for generating solar spectra.