CO2 emissions estimate from Mexico City using ground- and space-based remote sensing

The Mexico City Metropolitan Area (MCMA) has become the most populous urban region in North America and in the top five megacities worldwide (around 22 million inhabitants). To quantify the urban CO2 emissions of MCMA, a dense network of  Fourier Transform Infrared (FTIR) spectrometers with 6 portable EM27/SUN and 1 IFS 125HR were deployed within and around the MCMA  tracking gradients in atmospheric column CO2 concentrations (XCO2) from October 2020 to May 2021 (part of the French-Mexican MERCI-CO2 project). During these 7 months, twenty  XCO2 images (Snapshot Area Mode)  were collected by the NASA Orbiting Carbon Observatory (OCO-3) mission over the MCMA. By comparing  ground- and space-based observations, we found a positive XCO2 difference with a mean (± one standard uncertainty) of 1.16 (± 1.14) ppm between OCO-3 and FTIR column measurements, most probably caused by their respective calibration procedures. XCO2 gradients observed between the urban plume and its surroundings  show a good agreement between OCO-3 and the FTIR stations with a correlation coefficient (R) of 0.92,  decreasing significantly when comparing intra-city gradients (R=0.24). 

In a second phase, we assimilated these two types of dense column-integrated observations (FTIR network and OCO-3 SAM observations) separately to optimize the anthropogenic emissions from the MCMA and the biogenic CO2 fluxes in and around the city limits, in addition to the CO2 background concentrations. The X-Stochastic Time-Inverted Lagrangian Transport (X-STILT) model driven by the Weather Research and Forecasting (WRF) at 1-km resolution was used here to relate our atmospheric observations to surface fluxes and background conditions. An analytical Bayesian inversion technique was used here to robustly update the prior estimates at 1-km and 1-hour resolution. High-resolution prior biogenic CO2 emissions were computed with the light-use efficiency model CASA and our prior backgrounds from the global CAMS CO2 atmospheric inversion (version: v21r2). Prior anthropogenic CO2 emissions at 1 km are coming from a Mexico-specific inventory UNAM and from the global inventory ODIAC. These two inventories contain large discrepancies over MCMA (~40 %), while UNAM provides more detailed  information of point sources across our inversion domain. We performed sensitivity control experiments to quantify the effects from different prior fluxes and different covariance parameters. Furthermore, we combine OCO-3 and FTIR together in the same inversion process, a promising step toward providing a verification to  combine multiple data streams over a specific city.