Kiriakidis et al. (2026) Data-driven forecasting of aeolian dust impacts on solar irradiance and PV output over the Eastern Mediterranean
Identification
- Journal: Scientific Reports
- Year: 2026
- Date: 2026-09-18
- Authors: P. Kiriakidis, Alexandros George Charalambides, A. Ioannidis, T. Christoudias
- DOI: 10.1038/s41598-026-71126-9
Research Groups
- Cyprus Institute (CyI)
- University of Cyprus (UCY)
Short Summary
This study develops a data-driven framework to forecast all-sky hourly global horizontal irradiance (GHI) and diagnose the radiative impact of mineral dust over the Eastern Mediterranean. The Extreme Gradient Boosting (XGBoost) model is trained on CAMS forecast fields and achieves substantial improvements over CAMS reanalysis, particularly under extreme dust loading.
Objective
- Investigate the effect of mineral dust on surface solar irradiance and photovoltaic (PV) output in the Eastern Mediterranean region.
- Develop a data-driven framework to forecast all-sky hourly GHI and diagnose the radiative impact of mineral dust.
Study Configuration
- Spatial Scale: Eastern Mediterranean region, with a focus on Cyprus.
- Temporal Scale: Hourly time series from 2023-2024.
Methodology and Data
- Models used: Extreme Gradient Boosting (XGBoost) ensemble model.
- Data sources: CAMS forecast fields, PV output data from 472 operational plants in Cyprus.
Main Results
- The XGBoost model achieves substantial improvements over CAMS reanalysis, particularly under extreme dust loading.
- The model reduces the systematic positive bias of CAMS reanalysis by approximately 46% relative to CAMS reanalysis under extreme-dust conditions.
- The model shows a consistent reduction in systematic error across all dust regimes.
Contributions
- This study provides a data-driven framework for forecasting all-sky hourly GHI and diagnosing the radiative impact of mineral dust over the Eastern Mediterranean region.
- The study highlights the importance of considering dust effects on PV output and grid management.
Funding
- Cyprus Institute (CyI) research grant.
- European Union's Horizon 2020 research and innovation program.
Citation
@article{Kiriakidis2026Datadriven,
author = {Kiriakidis, P. and Charalambides, Alexandros George and Ioannidis, A. and Christoudias, T.},
title = {Data-driven forecasting of aeolian dust impacts on solar irradiance and PV output over the Eastern Mediterranean},
journal = {Scientific Reports},
year = {2026},
doi = {10.1038/s41598-026-71126-9},
url = {https://doi.org/10.1038/s41598-026-71126-9}
}
Original Source: https://doi.org/10.1038/s41598-026-71126-9