Abstract
Urban resilience requires the ability to anticipate, absorb, and recover from shocks and stresses. Digital twins (DTs) offer a promising platform by combining real-time data streams with predictive models to simulate urban systems. This paper proposes a framework for developing city-scale digital twins that integrate heterogeneous IoT sensor data, dynamic simulations, and machine learning-based predictive analytics. Using a case study of a mid-sized European city, we demonstrate how real-time data assimilation and predictive modeling can enhance situational awareness, support proactive decision-making, and improve resilience outcomes. The methodology involves: (1) deploying a network of environmental and infrastructure sensors; (2) implementing a cloud-based data pipeline for real-time ingestion and processing; (3) constructing a 3D urban model with semantic layers; and (4) embedding predictive models for flood risk, heat island effect, and traffic congestion. Results show that the DT achieves high-fidelity representation with average prediction errors below 12% for key metrics. During a simulated extreme rainfall event, the DT provided a 25-minute advance warning of flood-prone zones, enabling timely evacuations. Comparative analysis reveals a 34% improvement in response time compared to traditional approaches. The discussion highlights scalability challenges, data integration issues, and the need for stakeholder engagement. Our findings underscore the potential of digital twins to operationalize urban resilience through real-time data and predictive modeling, while calling for further research on standardization and governance.
Keywords
digital twin, urban resilience, real-time data, predictive modeling, smart city, IoT, data assimilation, simulation