solar-forecasting

This paper conducts a controlled study on advection-aware graph nowcasting for distributed solar ramp forecasting, finding that accurate cloud-motion features are as important as graph structure and introducing a self-supervised estimator that reduces forecast error.

This paper proposes SolCloudLLM, an LLM-based multimodal framework that fuses sky images and time-series data for solar power forecasting, achieving significant performance improvements especially in few-shot and cloudy conditions.

FarSky is a generative forecasting framework using task-aware latent-space coupling with latent diffusion to produce deterministic and probabilistic intra-hour solar irradiance forecasts, achieving up to 11 percentage points improvement in skill and better ramp event detection.