DeepHeliosat (ongoing)

Advancing satellite-based solar resource modelling with deep learning techniques.

Vadim BECQUET's work paves the way for next-generation satellite-based SSI retrieval methods, harnessing deep learning and the enhanced resolution of third generation geostationary satellites for greater accuracy in solar resource assessment.

GeoFishEye (ongoing)

Simulated sky camera networks for spatial SSI.

The research work of Max ARAGÓN CERECEDES focuses on producing multi-view synthetic sky imagery for training and evaluating machine learning algorithms, with a special emphasis on sky camera networks for SSI modelling.

This work is of value for PV nowcasting, PV performance monitoring, and in the integration of all-sky irradiances into cloudscale weather models.

SSI Variability (ongoing)

Modelling, characterization and prediction of SSI variability.

Jose GOMEZ GOMEZ is exploring and developing methodologies for generating SSI data with high spatial and temporal resolutions.

This work enhances energy storage planning, smart grid management and solar power plant optimization. It also provides valuable data for insurance and finance sectors to improve risk assessment and financial modelling in solar projects.

FishSPN1 (ongoing)

Data fusion of information from a fish-eye camera and a global/direct/diffuse pyranometric sensor with no moving part.

Suzanne WEYAND aims to fuse the data of the low-cost fish-eye camera and the SPN1 pyranometer of the SciDoSol monitoring station for improved retrievals of solar radiation for both instruments.

Low-Cost All-Sky Imaging (completed)

Imaging for solar radiance estimation.

Valentin BAUER developed a low-cost all-sky imager using a Raspberry Pi and a compatible fish-eye camera. His project focused on estimating sky radiance at a specific location on Earth, leveraging affordable hardware and open-source sensor and software.

SSI Variability (completed)

Characterizing SSI data using wavelet transforms.

Pierre CHAPEL focused on the mathematical modelling of SSI variability. He used wavelet transforms to model the statistical nature of the SSI under different atmospheric conditions.

Solar Measurements in Polar Regions (completed)

SSI data in extreme environments.

Arthur PAOLINI's project focused on developing an innovative method to measure solar irradiance in Arctic regions using a fish-eye camera. This approach was designed to overcome the challenges posed by harsh Arctic conditions. In this work traditional pyranometers were replaced with a versatile, camera-based method for measuring solar irradiance.

Multi-Input Multi-Output Multi-Horizon Extreme Learning Machines with Embedded Reconciliation for Multi-Source Power Forecasting (completed)

Joint forecasting of heterogeneous energy sources.

Yoan Jheelan's project focused on developing a multi-input, multi-output, multi-horizon (MIMO-MH) framework that combines Extreme Learning Machines (ELM) with forecast reconciliation. The proposed MIMO-MH jointly predicts thermal, hydropower, solar PV, wind, and import generation over 1–24 h horizons, ensuring builtin coherence across both sources and lead times. The results of this method exhibit a 30–45% error reduction versus persistence, and outperform NeuralProphet and TimeGPT.

Diffuse sky radiance measurement using low-cost fish-eye cameras (ongoing)

Following-up on the internship of Valentin BAUER, a new version of the low-cost camera is in design phase, addressing previous issues detected in the field from the first version. A framework for data encoding and archiving is also in development, following the FAIR principles (Findable, Accessible, Interoperable, Reusable). And the conversion from raw images to the physical quantity of sky radiance is ongoing.

Impact of circumsolar irradiance and aerosol optical depth on surface solar irradiance estimates (ongoing)

Improvements in the clear-sky SSI will, in turn, enhance the all-sky SSI, as most satellite-based models treat these components as independent modules. The effects of correcting clear-sky modelled values using circumsolar irradiance, along with the influence of aerosol optical depth accuracy on clear-sky models and their subsequent impact on SSI, are being investigated to better understand and further improve clear-sky SSI estimates.