Estimation des composantes d'éclairement incliné par ciel clair à partir des luminances descendantes et montantes prenant en compte la réflectance du sol

Accurate prediction of downwelling and upwelling solar irradiances for various plane orientations is crucial for solar energy applications. Bifacial photovoltaic and solar tracking technologies significantly improve energy yields by capturing solar radiation on both the front and rear faces of photovoltaic modules, optimizing their orientations over time. These predictions are also vital in architecture and energy building, enhancing energy efficiency and ensuring visual and thermal comfort in buildings.For accurate predictions under all-sky conditions, it is essential to be able to provide first broadband direct, diffuse, and reflected solar irradiance components under clear skies, accounting for general ground reflectance. The objective of this PhD study is to explore the possibility of extending the capacity of the existing clear-sky irradiance model McClear, provided by the Copernicus Atmosphere Monitoring Services (CAMS). This model, which incorporates atmospheric parameters (ozone, water vapor, and aerosol properties) from CAMS, delivers fast and accurate downwelling global, diffuse horizontal irradiances, and direct normal irradiance under clear sky conditions but does not offer downwelling (diffuse) and upwelling (reflected) tilted irradiance.In this work, we propose a new model, McClear_Radiance, based on modeling downwelling and upwelling radiances, which are the sources for computing any irradiance component for arbitrary orientations. The McClear_Radiance model, extended from CAMS McClear, utilizes a LUT method using the radiative transfer model libRadtran. It relies on three key input parameters (diffuse fraction, aerosol type, and SZA) and it also incorporates ground reflectance through the Ross-Li modeling of the bidirectional reflectance distribution function (BRDF) for non-Lambertian surfaces, beyond albedo.Various validations and performance analyses have been conducted using two widely used empirical models as baselines to estimate tilted irradiances from direct and diffuse horizontal irradiances: the Perez Transposition model proposed by Perez et al. (1987) and a CIE-type radiance model by Perez et al. (1993a; 1993b). These studies were first performed in a simulated numerical environment using the RTM libRadtran as a reference for numerous clear-sky situations and various plane orientations, demonstrating significant improvement for the proposed McClear_Radiance model compared to the two baseline methods.“Real-world” validation of the McClear_Radiance model was finally conducted with high-quality concomitant global and diffuse horizontal, direct normal irradiance, and three global tilted irradiances (GTI, 20° South, 30° South, and 45° Southwest) from pyranometric sensors operated by DLR at the Plataforma Solar de Almería (PSA) in Southern Spain. Comparisons for the three orientations under clear sky conditions showed RMSE values of 20.1 W/m² (2.8%) and high correlation coefficients (≈ 0.998), consistent with the RMSE of clear-sky GHI from McClear (18.7 W/m², 2.9%), slightly better than the Perez Transposition model (23 W/m2, 3.2 %) and indistinguishable from the Perez Radiance models (20.1 W/m2, 2.8 %), considering measurement uncertainties.