A critical overview of the (Im)practicability of solar radiation forecasting models

Abstract

To size solar energy facilities, the availability of solar radiation data on a horizontal surface is inevitable. Based on the vast applications of energy from the sun in electricity generation, quality research has been dedicated to developing models capable of estimating solar radiation for various purposes. Although tremendous successes have been recorded while developing these models, some researchers have questioned the relevance of such models for practical applications. Aside from presenting an overview of existing solar radiation forecasting models, this study presents a perspective and critical discourse on the practicability or otherwise of solar irradiation prediction models available in the literature. The existing models are classified as satellite-driven models, regression models, statistical models, artificial intelligence-driven models, and hybrid solar irradiation forecast models. This study also presents research gaps that need to be examined in the future. Insights from the review reveal that although the generalization capabilities of solar radiation forecast models specifically for locations without measuring instruments have turned out to be an uphill task, some promising researches are being implemented and improved to bridge the gap of generalization of models. The subject discussed in this paper will be valuable to researchers, practitioners, funders, and policymakers interested in developing and utilizing these models to foster the growth of the solar energy industry.