Science.

All-sky imaging framework for assessing cloud-free line-of-sight conditions in free-space optical satellite downlinks. Special Issue: Advances in Free-Space Optical Communications.

Benchmarks several cost-effective DNI and DHI measurement systems against sun-tracker reference instruments across multiple sites worldwide. Quantifies where low-cost alternatives can replace traditional pyrheliometers in solar-energy projects.

All-sky imagers face an extreme intra-image dynamic range, with the circumsolar neighborhood orders of magnitude brighter than the diffuse dome. This study compares RAW-domain HDR merging of bracketed exposures against the camera ISP's built-in HDR pipeline for circumsolar radiometry, and shows RAW-HDR preserves the radiometric information operational pipelines need.

Self-calibration tool for all-sky imagers that recovers the geometric camera model from images of the sun and moon. Validated on 5 camera types at 3 locations with high accuracy. Released as an open-source Python package, with method, camera model, and test data.

Zero-shot and few-shot learning tasks for solar nowcasting from a model pre-trained at another location. Generalising beyond the training site.

Comprehensive survey of 72 open-source ground-based sky-image datasets for very-short-term solar forecasting.

Comparative study of local training, global training, and transfer learning for solar nowcasting from sky images.

Physics-informed stochastic video prediction that generates multiple possible future images of the sky with diverse cloud-motion patterns from a past sky-image sequence.

Multi-location satellite and sky-image dataset for solar forecasting and atmospheric science.

Literature review of computer-vision-based solar forecasting, focused on deep learning for cloud-cover observations from sky cameras and meteorological satellites.

A single ML framework that fuses sky images and satellite observations for intra-hour solar forecasting.

A novel deep-learning architecture for joint solar-irradiance and sky-image prediction.

Comparative study of data-augmentation and scene-representation methods, including polar coordinates. Applied to solar forecasting from sky images and satellite observations.

Processing method that consistently centres a polar representation on the incoming flow of clouds using optical flow.

Benchmarking of standard deep-learning models for solar-power nowcasting from sky images.

Sun-tracking algorithm based on sky images only.