3D digital twin reconstruction of street flooding

Accurate and cost-effective 3D digital reconstruction is crucial for applications such as urban planning, flood monitoring, and other applications. However, existing methods often rely on expensive Light Detection and Ranging (LiDAR) scanning or manual modeling, highlighting the need for an alternative scalable approach. Recurrent nuisance flooding poses a growing risk to vehicle traffic in low-lying urban areas, especially during high tides and heavy rainfall, yet remains under researched due to the lack of labeled data. This work presents a novel method for generating synthetic, location specific flooding by leveraging open-source geospatial data and 3D modeling in Blender software. We construct realistic digital environments by importing georeferenced 3D models from OpenStreetMap into Blender, focusing on flood-prone areas of Norfolk, Virginia as a case study. These 3D digital twin models include terrain, road layouts, and building geometries, enabling the generation of detailed, location-specific realistic scenes. In Blender software tool, we eliminate mesh artifacts, simulate floodwater behavior, model traffic flow, and configure illumination to closely match real world conditions. The resulting simulations offer promising visualizations of urban flooding under varying water levels. This approach supports the creation of annotated datasets for flood detection and analysis, enabling scalable, data-driven research in urban resilience and disaster preparedness.