City2Twin: an open urban digital twin from data integration to visualization and analysis

Abstract. Urban Digital Twins have gained significant interest in the urban and geospatial fields, enabling interactive visualization and advanced analysis of cities across various domains. However, current implementation approaches are heterogeneous in terms of data and approaches. Furthermore, most implementations are based on specific needs. This project develops a comprehensive framework for Urban Digital Twins, focusing on data integration, storage, visualization, and analysis, all using open-source tools. Our approach integrates various data types, including 3D city models, dynamic air quality data, and external data imported from the client side, such as vector data, 3D city models, and point clouds. We conducted a series of experiments for each step and tackled various challenges. Many configurations are applied before integrating the 3D models, including ground reprojection, geometry type conversion, and format conversion. For the data storage and management, we performed several comparative tests between 3DCityDB and CJDB, which led us to choose CJDB for its simplicity and lightweight nature. A client interface built with the Giro3D framework (based on Three.js) connects directly to the 3D model database via a Flask server. Dynamic data is retrieved via external APIs and stored in a separate database following the SensorThings API standard, allowing time-series analysis. Our framework is standardized and designed based on open-source software, emphasizing the openness, transferability, reusability, and maintainability of Urban Digital Twin. The City2Twin project proposes significant improvements in data analysis, highlighting the importance of having a separate database for storing static and dynamic data, as well as the importance of direct interaction between the client interface and the 3D database for data updates and management. To the best of our knowledge, this work is the first Urban Digital Twin initiative that relies on the CityJSON format, which defines itself as more «developer-friendly».