Deep learning–based model optimization for rooftop area detection from satellite imagery and determination of urban-scale rooftop photovoltaic potential

Accurate estimation of available rooftop areas for PV power generation at the city scale is critical for sustainable energy planning and policy development. In this study, using publicly available highresolution satellite imagery, rooftop solar energy potential in urban, rural and commercial areas is estimated using deep learning models. In order to identify roof areas, high-resolution open-source images were manually labeled and the training dataset was trained with DeepLabv3+ architecture. The developed model performed roof area detection with high accuracy. Model outputs are integrated with a user-friendly interface for economic analysis such as cost, profitability and amortization period. This interface automatically detects roof regions in the bird's-eye view images uploaded by users, calculates the total roof area and classifies according to the potential of the area. The system, which is applied in 81 provinces of Turkey, provides sustainable energy projections such as PV installed capacity, installation cost, annual energy production, energy sales revenue and amortization period depending on the panel type, roof type and city selection. The results show that the DeepLabv3+ architecture and the Adam optimization algorithm provide superior performance in roof area estimation with accuracy 96.6% and loss rates 0.08%. Tests conducted on different roof areas in 81 cities yielded a maximum roof prediction accuracy (IoU) of 91.5% and an average of 77.94%. In economic analyses, the amortization period for rooftop PV users who have formed cooperatives reaches its lowest value at 7,6 years in areas where polycrystalline panels are used, while the longest amortization period for the same panel usage in rooftop areas reaches up to 10,4 years. These periods were calculated separately for monocrystalline and thin-film panel types. At the same time, for four different roof types associated with different panel types, the lowest amortization period of 7,6 years is achieved with the use of polycrystalline panels on two-sloped roof types, while the highest period of 12,6 years is achieved with the use of thin-film panels on flat horizontal roof types. Finally, in a study conducted for Şanlıurfa province based on an area of 1 km2, the total roof area was estimated to be 0.296 km2. Based on the estimated roof area value for this region, the annual PV energy production was calculated to be 60.13 GWh. It has been determined that in the region where the study was conducted, the amortization period for rooftop PV installations will decrease by 25% if energy users form a cooperative. As a result, this study offers an interactive user interface integrated with a deep learning model capable of high-accuracy roof PV area detection, enabling the assessment of sustainable PV energy potential at the city scale and facilitating economic analyses. This approach is a valuable tool for planning and decision support systems in the integration of renewable energy sources.