A research team led by the National University of Malaysia has developed a novel deep-learning method for PV suitability mapping.
“The novelty of our research lies in the development of a fully optimized and interpretable deep learning pipeline, combining TabNet, Optuna, and SHAP for solar PV site suitability mapping,” corresponding author Khairul Nizam Abdul Maulud told pv magazine. “This is the first study to apply this integrated approach at a regional scale across the Middle East, leveraging real-world PV deployment data alongside environmental, social, and governance (ESG)-aligned techno-economic variables. It significantly improves the accuracy, scalability, and explainability of spatial decision-making in PV planning, especially in data-scarce regions.”
The first step in creating their new model was to identify the suitability criteria. Based on a literature review and interviews with experts, that group has decided on 12 criteria. Six of them were technical conditioning factors, namely solar radiation, estimated power output, air temperature, cloud days, wind speed, and elevation. The other six were economic conditioning factors, namely slope, surface roughness, land cover, proximity to roads, proximity to cities, and proximity to grids.
The research team sourced 612 PV sites from the Global Solar Power Tracker database, along with 612 non-PV sites. Using ArcGIS, each was considered for the 12 technical and economic criteria. The dataset was then divided into a training subset (70%), a validation subset (15%), and a test subset (15%). The model itself was built with TabNet, Optuna, and SHAP. The first is a deep learning model optimized for tabular data, the second automates the selection of the best model settings, and the last was used to understand how the model makes its predictions.
“Results demonstrated the high predictive performance of the proposed approach across both validation and testing datasets, achieving classification accuracy (CA) and area under the receiver operating characteristic curve (AUC) values of 0.875 and 0.947 (validation) and 0.886 and 0.913 (testing), respectively, thus outperforming TabPFN, FT-Transformer, and seven ML models, including RF,” the academics stressed.
Applying their model to the entire Middle East, they found that approximately 5.8% of the region has very high suitability and 11.5% is highly suitable for PV energy development. The very high suitability areas are mainly concentrated in the coastal lands, central Anatolia, and parts of Saudi Arabia and Iran, while the highly suitable areas are found particularly in central and southern Turkey, the alluvial plain of Iraq, and the Nile Basin in Egypt.
“We were particularly surprised to find that proximity to infrastructure such as roads, cities, and electricity grids had a more significant influence on PV site suitability than solar radiation itself,” Maulud said. “This contrasts with conventional assumptions and is likely due to the uniform solar exposure across the Middle East. Additionally, surface roughness was found to negatively affect suitability, an underexplored factor in past studies.”
Maulud added that the research team is planning several follow-up studies. “These include incorporating additional exclusion and environmental risk factors at the micro-scale, integrating time-series Earth observation data to assess temporal changes in suitability, and expanding the model for hybrid systems such as wind-PV or agro-PV. We are also exploring transfer learning methods to improve model adaptability in other developing regions,” he said.
The novel methodology was presented in “Enhancing solar PV suitability mapping in the Middle East using an optimized deep learning framework,” published in the Alexandria Engineering Journal. Aside from the National University of Malaysia, scientists from Iraq's University of Thi-Qar and the United Arab Emirates' University of Sharjah have also participated in the study.
This content is protected by copyright and may not be reused. If you want to cooperate with us and would like to reuse some of our content, please contact: editors@pv-magazine.com.
By submitting this form you agree to pv magazine using your data for the purposes of publishing your comment.
Your personal data will only be disclosed or otherwise transmitted to third parties for the purposes of spam filtering or if this is necessary for technical maintenance of the website. Any other transfer to third parties will not take place unless this is justified on the basis of applicable data protection regulations or if pv magazine is legally obliged to do so.
You may revoke this consent at any time with effect for the future, in which case your personal data will be deleted immediately. Otherwise, your data will be deleted if pv magazine has processed your request or the purpose of data storage is fulfilled.
Further information on data privacy can be found in our Data Protection Policy.