A building centric urban digital twin for resiliency planning

Digital twin technology, as an emerging modeling and simulation paradigm, holds significant potential for advancing flood risk management. However, the development of digital twins for flood mitigation relies on accurate and reliable building asset data as a foundational layer. Acquiring such data for numerous buildings remains challenging due to uncertainties in existing sources and methods.This study focuses on the built environment data layer of the digital twin, emphasizing the identification of spatial structural indicators critical for flood mitigation. It evaluates potential data sources and technologies, assessing their quality in terms of uncertainty, bias, and equity. The research begins by identifying key spatial structural indicators and corresponding capture methods, forming the basis for a framework to address data gaps, reduce uncertainties, and enhance completeness. This framework supports the creation of a robust urban digital twin to improve flood resilience at a city-wide scale.The study introduces Levels of Detail (LoDs) as a method for abstracting complex building information into essential attributes, simplifying data management and analysis within the digital twin framework. It categorizes spatial indicators essential for flood risk assessment at varying levels of complexity and employs Bayesian linear regression to evaluate the impact of LoDs on flood mitigation decisions. A case study on Hurricane Ida in Manville, New Jersey, demonstrates the feasibility of this approach, showing improved prediction accuracy with higher LoDs.To address the challenge of acquiring critical yet costly data, such as first-floor elevation (FFE), the study leverages inferencing and imputation techniques. Inferencing is applied by classifying buildings based on architectural typologies and utilizing data libraries to estimate missing FFE values. Additionally, geostatistical imputation is employed to predict the FFE of buildings where data collection is unavailable. These techniques significantly enhance the representation of urban structures in the digital twin, mitigating data acquisition challenges while ensuring accuracy across diverse urban and coastal environments.In summary, this research presents a systematic approach to developing a building-centric urban digital twin for flood resilience planning. By integrating spatial indicators, LoD abstraction, inferencing and imputation strategies, it enables accurate representation of urban structures, supports informed decision-making, and facilitates proactive measures to enhance resilience against flooding.