Modeling and Prediction of Clear-Sky Solar Irradiance Losses Using Meteorological and Radiative Parameters: A Case Study in Ouarzazate, Morocco

Abstract. Understanding how to quantify loss potential due to clear-sky solar irradiance is essential for the development of enhanced Photovoltaic and Solar Energy Forecasting systems in Semi-Arid climates. In Ouarzazate, Morocco, four years (2021 - 2024) of daily weather and solar radiation observations collected by satellite and from the ground were analysed in this research through the analysis of clear-sky losses. Clear-sky conditions were isolated from all others by applying a Clearness Index threshold of ≥0.65 (KT), and Clear-sky Losses were derived from the relative differentials between All-Sky Shortwave Solar Irradiance (SSI) and Clear-Sky Shortwave Solar Irradiance (SSI). Correlation analysis between the Clear-Sky Loss and four key weather variables produced strong correlations. Three models were created to predict clear-sky losses: Linear Regression (LR), Second Order Polynomial Regression (2PR), and a Random Forest (RF) algorithm. The results indicated that the 2PR model provided the best fit (R² = 0.9965, RMSE = 0.0079) allowing for accurate comparisons of the 2PR model to the two better-known methods of predicting Solar Resource & Energy Potential.