PI: Fredrik Lindsten, Linköping University
co-PI: Natascha Kljun, Lund University
Gas advection and dispersion dynamics shape air quality, greenhouse gas fluxes, and emergency response capacity. Yet, current modeling approaches are incapable of inference from diverse sensory data, too computationally demanding for real-time use, and lack reliable uncertainty quantification. This project will develop new computational methods, including physics-informed generative machine learning models, that integrate sensor networks and atmospheric observations for computationally efficient inference and prediction of gas dynamics at high spatial and temporal resolution. The methods will be applicable to complex geometries, making them suitable for inference in urban environments, and will enable more effective climate change mitigation, urban air quality management, and rapid response to hazardous releases.
Project number: G8