A seasonal two-factor model for solar energy production: A climate extreme events analysis

In this paper, we propose a novel stochastic model for forecasting solar energy production, incorporating key climate-related uncertainties. Unlike existing approaches, which primarily rely on Gaussian or skew-normal processes, our model employs a skew-geometric Brownian motion with a time-dependent seasonal drift and an error term following a mixture distribution. Additionally, we integrate temperature variations, modeled as a non-homogeneous mean-reverting Ornstein–Uhlenbeck process, to account for their dynamic impact on photovoltaic efficiency. A distinctive feature of our model is the inclusion of a jump component of compound Poisson type, which explicitly captures the influence of extreme climatic events on solar energy output. By applying our methodology to data from 28 countries, we demonstrate that our model significantly outperforms two benchmark approaches in accurately predicting energy production under extreme conditions. This contribution provides a more comprehensive and realistic representation of solar power variability, improving risk assessment and decision-making in energy planning.