OBIA approach for the analysis of medium and high resolution satellite data for environmental monitoring: PV plant mapping

The global expansion of photovoltaic (PV) installations continues to grow in response to rising energy demand and the need for sustainable energy produc tion. However, large-scale solar farms require the use of large areas of land that are diverted from food production, the demand for which has increased dramat ically in recent years, owing to the growing world population. To monitor the rapid growth of solar farms, it is essential to collect data on the quantity, distri bution, and impact of these systems. Despite the many challenges in mapping PV systems owing to the variety of materials and complexity of layouts, many re searchers are trying to improve this process using remote sensing techniques combined with machine and deep learning. This study developed a comprehen sive framework for mapping and monitoring PV systems using open-source Sen tinel-2 (S2) satellite imagery and remote sensing techniques supported by ma chine-learning algorithms. The framework was applied to two study areas with different characteristics by adopting an Object-Based Image Analysis (OBIA) ap proach to improve image segmentation and classification. The preliminary phase included the analysis of the Normalized Difference Vegetation Index (NDVI) index variation within the Ground Truth (GT) polygons, improving the segmentation through the creation of image