Digital Urban Twin

Abstract

Urban Digital Twins (UDTs) have emerged as important tools for supporting sustainable urban planning, renewable energy integration, and smart city applications. However, generating geomet-rically accurate and semantically rich urban models remains challenging due to limitations in air-borne LiDAR data and existing rooftop reconstruction workflows. This thesis presents a multi-stage geospatial methodology for developing an Urban Digital Twin of the Le Vigne district in Cesena, Italy, with a focus on rooftop reconstruction and photovoltaic suitability assessment. The proposed workflow integrates cadastral building footprints, institutional LiDAR datasets, DSM/DTM prod-ucts, Google Earth Pro measurements, Google Photorealistic 3D Tiles, semantic segmentation, and automated reconstruction using Roofer. Geospatial harmonisation was performed in QGIS, while dense photogrammetric point clouds were generated and processed using Blender and Cloud-Compare. The resulting LoD2 building models were semantically enriched and exported in CityJSON format. The findings indicate that sparse airborne LiDAR data are effective for general urban modelling but insufficient for accurately reconstructing complex residential roofs. Dense photogrammetric reconstruction significantly improved roof geometry, ridge continuity, and roof-type representation. The final Urban Digital Twin incorporated semantic attributes such as roof slope, roof type, and photovoltaic suitability, enabling rooftop solar assessment and urban energy analysis. The study demonstrates the potential of combining LiDAR and photogrammetric data to create interoperable Urban Digital Twins that support data-driven urban planning and smart city decision-making.