Benchmarking deep learning methods for irradiance estimation from sky images with applications to video prediction-based irradiance nowcasting

To address the high levels of uncertainty associated with photovoltaic energy, an increasing number of studies focusing on short-term solar forecasting (i.e. nowcasting) have been published. Most of these studies use deep-learning-based models to directly forecast a solar irradiance or photovoltaic power value given an input sequence of sky images. Recently, however, advances in generative modeling have led to approaches that divide the nowcasting problem into two sub-problems: (1) future event prediction, i.e. generating future sky images; and (2) solar irradiance or photovoltaic power estimation, i.e. predicting the concurrent value from a single image. One such approach is the SkyGPT model, whose potential for improvement is shown to be much larger in the estimation component than in the generative component. Thus, in this paper, we focus on the solar irradiance estimation problem and conduct an extensive benchmark of deep learning architectures across the widely-used Folsom, SIRTA and NREL datasets. Moreover, we perform ablation experiments on different training configurations and data processing techniques, including the choice of the target variable used for training and adjustments of the timestamp alignment between images and irradiance measurements. In particular, we draw attention to a potential error associated with the sky image timestamps in the Folsom dataset and suggest a possible fix. By leveraging the three datasets, we demonstrate that our findings are consistent across different solar stations. Finally, we combine our best irradiance estimation model with a video prediction model and obtain state-of-the-art results on the SIRTA dataset.