Validation of surface solar irradiances estimates and forecast under clear-sky conditions from the CAMS McClear model in Benguerir, Morocco

Abstract

McClear is a fast model of surface solar irradiance under cloudless condition (clear-sky), based on the radiative transfer model libRadtran that exploits the clear sky atmospheric optical properties (notably partial and total aerosol optical depths - AOD - at different wavelengths, total contents of water vapor and ozone) provided by the EU-funded Copernicus Atmospheric Monitoring Service (CAMS). This model provides global, diffuse horizontal irradiances and direct normal irradiance at any place and any time since 2004 onwards, with 2 days of delay, with temporal resolution up to 1 min (interpolation). In this paper, we present a first validation of the McClear clear sky irradiance model for the specific climate of Benguerir, Morocco. McClear's validation was performed by comparing the 1 min average clear sky irradiance data (GHI: Global Horizontal Irradiance and DNI: Direct Normal Irradiance) from a high precision thermopile-based pyranometric station installed in Benguerir at Green Energy Park, and the values estimated by McClear for the same instants. The bias for global horizontal irradiance is 5 W m-2 (0.6 % of the mean observed irradiance). The root means square error (RMSE) is 17 W m-2 (2.2 %) and the correlation coefficient (CC) is 0.996. The bias for the direct normal irradiance is -17 W m-2 (-1.9 %). The RMSE is 58 W m-2 (6.7 %) and the correlation coefficient (CC) is 0.845. The same types of comparison have been done to assess the ability of McClear to forecast clear-sky GHI and DNI one day-ahead, using the provided forecasts of the state of the clear atmosphere by CAMS. The Bias, RMSE, STDE and correlation coefficient of such comparisons on the average 1-min profile of GHI and DNI from McClear are respectively for GHI - 0.8%, 3.3%, 3.3% and 0.991. For DNI -2.8%, 9.9%, 9.5% and 0.779.