Anthropogenic Carbon Emission Constraints from CO and NO2 Data Streams

Quantifying the anthropogenic sources of CO₂ and CH₄ is imperative yet challenging. We introduce a tractable multi-species data analysis framework that aims to enhance CO₂ and CH₄ signatures from anthropogenic activities through the synergistic use of satellite-derived measurements of combustion-related trace gases (CO and NO₂, CO₂ and CH₄) and global chemistry transport modeling. This is anchored upon the utility of NO₂ data to identify combustion activity (for CO₂) and CO to track corresponding plumes and to provide information on combustion efficiency (CE). These constituents (CO, NO₂, CH₄) are also strongly coupled chemically via OH (i.e., CO is a sink of OH while NO₂ is a source of OH within the tropospheric O₃ chemistry). Here, we present initial results on our joint analysis of CO₂ and CO data in constraining regional fossil-fuel CO₂ from Asia and Korea/Japan (ffCO₂) using simulated CO₂ and CO (including regional tags of ffCO₂ and ffCO) in the Community Atmosphere Model with Chemistry (CAM-chem) and ¹⁴CO₂, CO₂, and CO data taken during KORUS-AQ field campaign in May 2016. We use an ensemble of posterior CO₂ fluxes from available CO₂ inversion systems (e.g., CarbonTracker) and best emission scenarios based on the HTAPv2 inventory for CO to provide a modeling platform that best simulate the abundance of these species, enabling us to explore these synergies more effectively. Our CAM-chem simulations show reasonable agreement across observation platforms including NOAA CCGG, TCCON, OCO-2, and MOPITTv8. We find that ffCO₂ derived from ¹⁴CO₂ and simulated ffCO₂ tags show strong correlation (r=0.82). We also find that ffCO₂ from East Asia and rest of the world need to be scaled up (1.61 and 1.28, respectively), while ffCO₂ from Korea and Japan needs to be scaled down (0.84). This is supported by a complementary analysis on CO abundance (and ffCO tags) which reveals clearer signatures of modeled ffCO₂ plume transport, sectoral emissions, and CE. We find that dCO/dCO₂ can be better use in diagnosing inconsistencies in CE using ffCO and ffCO₂ tags. The observed and modeled dCO/dCO₂ (~13 ppb/ppm) over Korea is largely influenced by dffCO/dffCO₂ from Korea (6.7 ppb/ppm) with modest influence from middle and northern East Asia (~52-55 ppb/ppm), suggesting a higher CE over Korea than East Asia despite recent decreases in CO. To further support our findings, we also conduct a suite of Bayesian synthesis inversions with KORUS-AQ and OCO-2/MOPITT data. We augment the ffCO₂ emission (state) and observation vectors with ffCO emission and CO, respectively. We took advantage of CO₂:CO error correlations derived from KORUS-AQ data to incorporate off-diagonal elements in the a priori and observation error covariances. These correlations simply serve to transfer CO information from the CO data to ffCO₂. Our results show that there is a reduction in a posteriori error in ffCO₂ emissions when these correlations are incorporated. However, the a posteriori ffCO₂emissions are fairly sensitive to the magnitudes of these correlations relative to inversions using only ffCO₂ data as constraints, suggesting that such correlations need to be further accurately quantified. Our results have implications on the design of current multi-species inversion systems, especially those that have planned on using satellite column retrievals of CO (and/or NO₂) with CO₂ (and/or CH₄).