Lopez, Gabriel , Ivorra, Benjamin , FERRADA MARTINEZ, PABLO DANIEL, Gueymard, Christian A.
No
Sol. Energy
Article
Científica
01/02/2026
001645262500001
2-s2.0-105024935564
This work introduces a novel methodology for the extension and upscaling of direct solar irradiance spectra measured with spectroradiometers, using Genetic Algorithms (GAs) to address the associated inverse problem. Acquiring accurate spectral data of solar irradiance is critical for various applications, including photovoltaic (PV) technology, environmental monitoring, and biotechnology. However, limitations in spectroradiometer instru mentation often restrict the availability of detailed spectral information. The proposed methodology specifically targets periods when the sun is unobscured by clouds, under which accurate retrievals and spectral extensions can be obtained irrespective of the cloudiness status of the rest of the sky. This approach leverages the SMARTS radiative transfer model to generate synthetic spectra under diverse atmospheric conditions and employs GAs to estimate key atmospheric parameters that align the simulated spectra with the observed measurements. Compared to traditional methods, the GA-based optimization significantly improves computational efficiency and estima tion accuracy. Unlike conventional approaches that rely on selected narrow spectral ranges, this methodology utilizes full-spectrum (350-1050 nm) observations, enabling comprehensive spectral upscaling, improving spec troradiometer calibration, and offering a framework for benchmarking other algorithms. With median overall errors typically below 1-2 % in the validation range, it achieves a close match between optimized and actual spectra. The experimental validation of the method, based on more than 2000 observed spectra near Huelva, Spain demonstrates robust performance across varying local conditions, enabling the accurate determination of three key atmospheric quantities: aerosol optical depth, precipitable water, and the & Aring;ngstr & ouml;m exponent. Application to real measurements further confirms the methodology's potential in identifying calibration issues in spectrora diometers. This methodology offers a powerful tool for expanding spectral coverage in solar energy and related fields, with the added benefit of scalability to diverse geographic and atmospheric conditions. The developed ap proach facilitates accurate solar irradiance modeling and has promising implications for advancing PV efficiency assessments, agrivoltaics, and climate research.
Direct solar spectral irradiance; Spectroradiometers; Radiative transfer modeling; SMARTS; Genetic algorithms; Inverse optimization; Atmospheric constituents; Global optimization techniques; Photovoltaics