Cornell Engineers Build a 'Digital Twin' to Model Manhattan's Air Quality

Cornell engineers have developed a "digital twin" framework that creates a real-time virtual representation of urban carbon dioxide conditions, a tool the researchers say could help city planners monitor emissions, identify hotspots and evaluate potential interventions before implementing them in the real world. The project, published in the journal Environmental Modelling & Software, used Manhattan as its first test case and was led by H. Oliver Gao, director of the Systems Engineering Program and the Center for Transportation, Environment, and Community Health at Cornell's Duffield College of Engineering.

A digital twin is a real-time digital replica of a physical system that continuously incorporates real-world data into its computational models. The team's "Sustainable Urban Digital Twin" used four layers of architecture working together: a physical layer that collects data from large databases, a digital layer that organizes and models that data, a brain layer that uses Bayesian modeling and machine learning to predict outcomes, and a service layer that suggests actions and provides tools such as visualization for decision-makers.

"We chose Manhattan as our pilot site because it is a dense and complex urban environment that would be a challenging test for our digital twin," said Yishuo Jiang, an Ezra Postdoctoral Fellow in the Systems Engineering Program and co-lead author of the study. "We had access to data from across the city that recorded carbon dioxide concentration, air temperature and humidity every five minutes, so we chose carbon dioxide as a variable to test our prototype." Using the digital twin, the researchers integrated data from different sources into one platform, monitored carbon dioxide levels, quantified uncertainty in the data, and generated maps showing how conditions varied across the city.

The researchers caution that the current platform should be viewed as a research prototype, and that additional work is needed before it can support comprehensive management of pollutants such as particulate matter, nitrogen oxides or ground-level ozone. They envision expanding the platform to other pollutants and other cities, and adding AI-assisted decision support. "We see this work as an early step toward a new generation of urban intelligence systems," Gao said. "Our long-term vision is to build such systems across cities and urban domains, turning data and models into actionable intelligence for healthier, more sustainable and more resilient communities."