Calibração de dados satelitais de irradiação solar com base em medições in situ: uma abordagem de regressão linear e clusterização de GHI e DIF

Abstract

This paper presents a methodology for the calibration of solar irradiance data from satellites, using in situ measurements as a reference. The study proposes an innovative approach that combines linear regression models with clustering techniques to adjust the Global Horizontal Irradiance (GHI) and Diffuse Irradiance (DIF) components. Initially, the data are subjected to rigorous pre-processing, which includes the removal of missing and negative values, in addition to the correction of inconsistencies, ensuring the quality of the basis used for modeling. Then, statistical models are developed that correlate the field measurements with the satellite estimates, segmenting the data according to the solar zenith angle to capture specific behaviors in different conditions. The results show significant gains in the accuracy of the estimates, demonstrating that the integration of in situ measurements with satellite data can reduce the biases inherent in the estimates and increase the reliability for applications in photovoltaic system sizing and climate studies. It is concluded that the proposed approach is robust, viable in different geographic and atmospheric scenarios, and contributes significantly to the improvement of solar irradiation estimates.