Study of the variation of solar radiation using time series decomposition method for accurate forecasting

Abstract

Solar radiation is a key factor influencing energy generation, climate studies, and agricultural productivity. Accurate forecasting of solar radiation is crucial for optimizing renewable energy systems and mitigating climate-related challenges. This study investigates the variation in solar radiation using the time series decomposition method to improve the forecasting accuracy. By applying both time series multiplicative and additive decomposition models, the trend, seasonal and irregular (residual) components of solar radiation data can be analysed. This study focuses on identifying patterns that enhance forecasting reliability, particularly in regions with fluctuating climatic conditions. The results of the component analysis show that the two models have similar characteristic variation patterns. The results of the study in terms of MAPE and MAD showed that both models perform almost identically, with negligible differences between their values (i.e., MAPE = 12.8463 and 12.8474) for multiplicative and additive models and (MAD = 2.2776 and 2.2778) for multiplicative and additive models. Conversely, the additive model performs slightly better than the multiplicative model in terms of the mean squared deviation (MSD), although the difference is minimal (i.e., MSD = 7.7768 and 7.7768) for the multiplicative and additive models, respectively. These results demonstrate the effectiveness of applying time series decomposition methods for capturing solar radiation variability, providing valuable insights for energy planning and environmental monitoring and control management. The findings of this study contribute to the development of more precise solar radiation forecasting models, aiding policymakers, researchers, and industry stakeholders in decision-making processes to enhance environmental monitoring for hazard reduction and control management.