Inventory-Free Inversion of Urban ffCO₂ Emissions Using Combined Observations from Sentinel-5P and DQ-1

The intensification of global climate change has created an urgent need for high-precision monitoring of fossil fuel carbon dioxide (ffCO₂) emissions. The Paris Agreement emphasizes that countries must be able to rapidly and accurately track changes in carbon emissions to support effective policymaking and implementation. Achieving this goal depends on building accurate and verifiable carbon accounting systems. Precise estimation of ffCO₂ emissions is essential for climate prediction and the formulation of mitigation strategies. Here, we present a city-scale ffCO₂ inversion framework that integrates active and passive satellite observations to improve emission quantification. Using satellite-derived NO₂ data and CO₂–NOₓ emission ratios, we first constructed spatial maps of urban ffCO₂ emissions. We then incorporated XCO₂ observations from the DQ-1 satellite’s ACDL instrument to estimate monthly ffCO₂ emissions for several major cities worldwide. Unlike conventional top-down methods that rely heavily on prior emission inventories, our approach derives emission information directly from satellite observations. This innovation substantially reduces uncertainties caused by the temporal delays and spatial biases inherent in traditional bottom-up inventories, offering a more reliable and timely means of monitoring fossil fuel CO₂ emissions.

The framework combines high-resolution NO₂ column observations from Sentinel-5P/TROPOMI with column-averaged CO₂ (XCO₂) measurements from the world’s first spaceborne CO₂ lidar, the DQ-1 Atmospheric CO₂ Differential Absorption Lidar (ACDL). TROPOMI NO₂ data are first used to derive gridded urban NOₓ emissions through a mass-balance approach that explicitly accounts for wind divergence, chemical lifetime, and vertical distribution. These NOₓ emissions are then converted into prior ffCO₂ distributions using city-specific CO₂-to-NOₓ emission ratios. Subsequently, DQ-1 XCO₂ along-track observations are assimilated within a Bayesian inversion framework driven by high-resolution WRF-STILT simulations to constrain total urban ffCO₂ emissions.

This study demonstrates the unique value of combining active CO₂ lidar and passive NO₂ observations for rapid, observation-driven verification of urban anthropogenic CO₂ emissions, and provides a unified framework for city-scale carbon monitoring under limited or uncertain inventory conditions.