Assessing Solar Radiation Characteristics in Ibadan, Nigeria Using a Regression-Based Diffuse Fraction Model

Abstract

The availability and accurate estimation of solar radiation characteristics are essential for the efficient design and deployment of solar energy systems. This study assesses diffuse solar radiation in Ibadan, Nigeria, using a regression-based diffuse fraction model to improve solar resource estimation. A fouryear average of monthly mean global solar radiation data, obtained from the IITA meteorological station, was analyzed to develop a predictive model for diffuse radiation estimation. The multiple regression model was developed using Minitab 20.4 software and validated using Mean Absolute Bias Error (MABE), Root Mean Square Error (RMSE), and Durbin-Watson statistics to evaluate the model’s accuracy and autocorrelation. Results showed that the developed regression model had an R² value of 99.94% and an adjusted R² of 99.74%, indicating a strong predictive capability. The error analysis revealed an MABE of 0.0004 and RMSE of 0.0008, confirming the model’s high accuracy in estimating diffuse solar radiation. The Durbin-Watson Statistic (2.23079) suggested the presence of negative autocorrelation, which implies minimal overfitting or redundant information in the dataset. The study demonstrated that the regression-based approach effectively predicts diffuse solar radiation, provided reliable data for solar energy system design and optimization in Nigeria. These findings are crucial for solar photovoltaic (PV) deployment, solar thermal applications, and energy planning in Ibadan and similar locations. The ability to accurately estimate diffuse radiation enhances solar energy conversion efficiency contributes to optimal solar panel orientation, and improves energy yield calculations for off-grid and grid-connected PV systems. This research contributes to the advancement of solar energy utilization in Nigeria and supports broader efforts in sustainable energy development by offering a validated predictive model for diffuse radiation estimation.