We use a cyclical embedding to encode these time features discarding the year. We adapted the majority of the baselines using the Time Series Library (TSlib (Wu et al., 2023)), The training of the baselines took place on a single RTX8000 GPU over the course of 100 epochs. During training, a batch size of 64 was consistently employed. Plateau, which gradually decreased the learning rate by a factor of 0.5 after a patience of 10 epochs. For the hyperparameter tuning of the baselines, we employed the Orion package (Bouthillier et al., We report the MAE and RMSE for the easy and difficult splits (presented in the main paper) along with the number of data points for each split.

We use a cyclical embedding to encode these time features discarding the year. We adapted the majority of the baselines using the Time Series Library (TSlib (Wu et al., 2023)), The training of the baselines took place on a single RTX8000 GPU over the course of 100 epochs. During training, a batch size of 64 was consistently employed. Plateau, which gradually decreased the learning rate by a factor of 0.5 after a patience of 10 epochs. For the hyperparameter tuning of the baselines, we employed the Orion package (Bouthillier et al., We report the MAE and RMSE for the easy and difficult splits (presented in the main paper) along with the number of data points for each split.

Nonetheless, the inherent variability of solar irradiance poses a significant challenge for seamlessly integrating solar power into the electrical grid. While the majority of prior research has centered on employing purely time series-based methodologies for solar forecasting, only a limited number of studies have taken into account factors such as cloud cover or the surrounding physical context.In this paper, we put forth a deep learning architecture designed to harness spatio-temporal context using satellite data, to attain highly accurate day-ahead time-series forecasting for any given station, with a particular emphasis on forecasting Global Horizontal Irradiance (GHI). We also suggest a methodology to extract a distribution for each time step prediction, which can serve as a very valuable measure of uncertainty attached to the forecast. When evaluating models, we propose a testing scheme in which we separate particularly difficult examples from easy ones, in order to capture the model performances in crucial situations, which in the case of this study are the days suffering from varying cloudy conditions. Furthermore, we present a new multi-modal dataset gathering satellite imagery over a large zone and time series for solar irradiance and other related physical variables from multiple geographically diverse solar stations.