Urban CO2 Emission Assessment based on High-Resolution Dispersion Simulations and MCMC based Inversion

Cities are significant contributors to global greenhouse gases. Accurately quantifying urban CO₂ emissions from atmospheric observations requires fine-scale modelling and rigorous inverse optimization of the emissions. Within the ICOS Cities project, we developed a high-resolution urban CO₂ emission estimation framework for Munich, coupling a detailed emission inventory with the computational fluid dynamics (CFD) based GRAMM-SCI/GRAL-ST-ROG model and with a novel inversion algorithm based on a Markov Chain Monte Carlo and Gaussian Process (MCMC-GP) approach.

The framework utilizes GRAMM-SCI to simulate mesoscale wind fields, which are subsequently refined by the GRAL-ST-ROG model. By integrating high-resolution datasets—including land cover, 3D building, and a self-developed tree cover dataset—with surface meteorological observations and Doppler wind lidar vertical profiles, the model generates wind fields at a 10-meter spatial resolution. The resulting wind fields are then combined with high-resolution emission inventories to drive the CO₂ dispersion simulation.

The simulated CO₂ concentrations were validated against Munich's mid-cost observation network. Furthermore, a new MCMC-GP algorithm was developed to facilitate spatio-temporal inversion across multiple emission sectors. This approach offers high flexibility, the capability to perform inversions under data-sparse conditions, and the ability to refine prior knowledge—such as spatial correlations and uncertainties—to ensure the method’s robustness.

This study presents a high-fidelity tool for quantifying urban emissions, supporting evidence-based policymaking to achieve climate targets.