Estimation Global Solar Radiation from Common Meteorological Data Using a Stepwise Multiple Regression Model in Abu Gharib, Iraq

Studies on energy efficiency in solar power plants typically focus on improving solar radiation system performance, panel efficiency, and energy production optimization. This study employed Stepwise Multiple Regression (SMR) to predict total solar radiation (TSLR) using thirteen daily average meteorological parameters from 2014 to 2024. The SMR method analyzed the influence of independent variables on TSLR in a step-by-step manner, identifying the most optimal prediction model based on the results. The analysis identified key predictors of TSLR using Pearson’s correlation, revealing strong positive correlations with evaporation (ET), clear sky (CS), and UV downward radiation, while negative correlations were observed with relative humidity (RH) and surface pressure (SP). A model incorporating eight variables was selected, achieving a high correlation coefficient (R = 0.952) and an adjusted R² of 0.907, demonstrating a robust fit. The model’s predictive performance was validated against 11 years of observed data, yielding strong metrics: R = 0.945, RMSE = 2.88 and MAE = 2.22. This study successfully excluded uncorrelated variables, improving model clarity and predictive efficiency. The predicted TSLR values closely matched observed data, achieving strong performance metrics: RMSE = 2.88, R² = 0.893, R = 0.945, MAPE = 14.7, and MAE = 2.22. These results demonstrate the model’s robustness and suggest its applicability for predicting TSLR in other regions of Iraq with similar climates, providing valuable insights for environmental and solar radiation studies.