Substantial overestimation of satellite-derived rooftop solar energy potential in multidimensional urban environments

Solar photovoltaic (PV) systems have emerged as one of the most effective technologies for converting sunlight directly into electricity through the photovoltaic effect. This rapid growth demonstrates the crucial role PV systems play in the global shift toward renewable energy. These systems exhibit remarkable versatility, capable of being deployed at diverse scales—from small residential setups that empower individual households to large utility-scale solar farms that contribute significantly to national grids. However, Complex multidimensional structure in urban environments have substantial impacts on solar PV energy harnessing. The intricate interplay of buildings, infrastructure, and urban geometry creates shading patterns and reflections that significantly affect the actual solar energy yields. However, satellite-derived estimates of PV potential often ignore these urban complexities, leading to substantial overestimations.

 

To tackle this issue, this study aims to propose a robust and cost-effective framework for quantifying the extent of overestimation by integrating high-resolution geostationary remote sensing imagery with LiDAR-based urban morphology data. First, we propose a hierarchical strategy for accurate large-scale solar position computation by sampling from the Solar Position Algorithm. Subsequently, the original global horizontal irradiance is decomposed into its primary solar constituents—beam, circumsolar, and isotropic—using solar position parameters. The digital surface model derived from LiDAR data simulates the effects of urban shading and sky occlusion on solar irradiance. The digital surface model derived from LiDAR data simulates the effects of urban shading and sky occlusion on solar irradiance. Ultimately, this method will enable the generation of accurate high-resolution solar energy potential maps and facilitate an analysis of the spatiotemporal characteristics of solar energy distribution patterns.

 

We use Hong Kong as the testbed, given its characteristic high-rise, high-density urbanization with multiple detailed data sources. Our framework is validated using eight in-situ ground measurements, showing a reduction in RMSE from 1.510 to 1.230 and an improvement in MAPE from 50.52% to 35.73%. Focusing on rooftop areas, our findings reveal that Hong Kong's overall solar energy potential in 2020 is 79.08 billion kWh, compared to 94.20 billion kWh estimated from direct satellite observations—a discrepancy of 19.11%, which highlights a significant overestimation. Our high-resolution maps have immense utility for urban planning and sustainable development, providing a precise tool for optimizing solar energy deployment in dense urban environments. These insights will aid in fostering more efficient and equitable energy solutions, contributing to the sustainable growth of urban areas.