Extending CityGML for Urban Solar Potential Estimation: A Semantically Enriched Model Informed by UAV-Derived Analysis

Abstract. Urban solar potential is often estimated using digital surface models (DSMs) and surface-based tools. These work well for rooftops but fall short when it comes to vertical surfaces, shadow dynamics, and simulation-ready attributes. In this study, we explored these limitations using a UAV-derived point cloud of the NIMBB building in UP Diliman. We generated a DSM and ran ArcGIS Pro’s Raster Solar Radiation tool to estimate rooftop solar potential under standard atmospheric assumptions. While the output highlighted high-potential zones on flat roof areas, it entirely excluded facade and ignored surface-level variables like panel orientation or shading over time. These are factors that influence real-world solar performance. These limitations point to the need for a more structured, object-based approach that can support detailed, surface-specific attributes and semantically link energy values to building components. In response, we proposed a conceptual extension to the CityGML data model. We introduce new classes such as SolarPanelInstallation, ShadowCastLog, and SolarPotentialAnalysisResult to represent the physical, contextual, and temporal dimensions of solar analysis. This model was designed to follow CityGML’s modular structure and can be integrated into semantic modeling workflows. The proposed model bridges the disconnect between 3D urban geometry and energy simulation, providing a clearer path for incorporating meaningful attributes into solar suitability assessments.