Quantifying anthropogenic CO2 emissions over urban areas from atmospheric observations and high resolution transport modelling

Urban areas are responsible for the majority of fossil fuel CO2 emissions - accounting for 70% of anthropogenic emissions on a global scale - and this number is expected to increase since the population of urban areas is rising. To manage and support mitigation efforts, there is a need for a better quantitative understanding of urban carbon budgets and for new methods for verification of anthropogenic emission estimates and their trends using atmospheric observations.

In this thesis, we focus on the area of Los Angeles (LA) as our test case and use OCO-2 observations over this urban area to estimate anthropogenic city emissions. A ΔXCO2 enhancement due to local emissions ranging between 0.7-2.8 ppm over the background is calculated from the observations for 2016. Using the same observational approaches, regional enhancements over large urban areas including New Dehli and Tehran are also estimated, revealing the need for an atmospheric transport model. A Lagrangian-dispersion modelling framework using the UK Met Office Numerical Atmospheric Modelling Environment (NAME) was applied to calculate surface footprints for satellite column observations from OCO-2. By combining column footprints with CO2 emission estimates from bottom–up anthropogenic Fossil Fuel CO2 emissions (FFCO2) and biogenic fluxes, we model the enhancement observed by the satellite (from both anthropogenic and natural fluxes) and their contribution to the observed signals over LA is estimated. Background concentrations are estimated from a global chemistry transport model (CarbonTracker) and NAME’s air back trajectories which are used to extract the inferred enhancement from the observations and compare with the estimated model enhancement. The average inferred anthropogenic enhancement for the 2016-2019 period was 0.55 ppm, which agrees well with the average model enhancement of 0.58 ppm. Some negative observed enhancements are predicted due to an overestimated background and show some correlation with high wind speeds. The capability of the model to calculate atmospheric CO2 concentrations with a daily resolution is demonstrated using OCO-2 synthetic data to prove the method for future satellites.