A Satellite-Driven Model for Monitoring Urban Material Metabolism, Embodied Emissions, and Carbonation

Urban systems are central to global material consumption and carbon emissions. However, systematically understanding urban metabolism remains a challenge due to the reliance on aggregated, top-down data which fails to capture fine-scale urban dynamics. To address this challenge, we developed an integrated, bottom-up framework and reconstructed the 30-year metabolic history of Xiamen, China. Our approach leverages a custom machine learning workflow on multitemporal, open-source satellite imagery to create a dynamic 3D Building Dynamics (3D-BD) model. This model accurately maps key building attributes (footprint, height, type, and construction/demolition year), achieving an R² of 0.81 for height estimation and F1-scores of 0.89 and 0.81 for detecting construction and demolition years, respectively. This high-resolution database then drives a GIS-MFA and LCA model to quantify material stocks, flows, embodied emissions across key lifecycle stages, and the often-overlooked carbon sink from cement carbonation. Results for Xiamen revealed a 5-fold increase in building stock, with material inflows peaking in the early 2000s while demolition outflows steadily rise, signaling a shift to urban renewal. Over the study period, cement carbonation provided a significant, distributed carbon sink, offsetting an estimated 6.6% of the building stock's total lifecycle embodied emissions. This transferable framework provides a more complete accounting of the urban carbon budget, offering a powerful tool for guiding sustainable planning and net-zero transitions.