Estimation of fossil-fuel CO₂ emissions using satellite measurements of "proxy" species

Abstract. Fossil-fuel (FF) burning releases carbon dioxide (CO₂) together with many other chemical species, some of which, such as nitrogen dioxide (NO₂) and carbon monoxide (CO), are routinely monitored from space. This study examines the feasibility of estimation of FF CO₂ emissions from large industrial regions by using NO₂ and CO column retrievals from satellite measurements in combination with simulations by a mesoscale chemistry transport model (CTM). To this end, an inverse modeling method is developed that allows estimating FF CO₂ emissions from different sectors of the economy, as well as the total CO₂ emissions, in a given region. The key steps of the method are (1) inferring "top-down" estimates of the regional budget of anthropogenic NO_x and CO emissions from satellite measurements of proxy species (NO₂ and CO in the case considered) without using formal a priori constraints on these budgets, (2) the application of emission factors (the NO_x-to-CO₂ and CO-to-CO₂ emission ratios in each sector) that relate FF CO₂ emissions to the proxy species emissions and are evaluated by using data of "bottom-up" emission inventories, and (3) cross-validation and optimal combination of the estimates of CO₂ emission budgets derived from measurements of the different proxy species. Uncertainties in the top-down estimates of the NO_x and CO emissions are evaluated and systematic differences between the measured and simulated data are taken into account by using original robust techniques validated with synthetic data. To examine the potential of the method, it was applied to the budget of emissions for a western European region including 12 countries by using NO₂ and CO column amounts retrieved from, respectively, the OMI and IASI satellite measurements and simulated by the CHIMERE mesoscale CTM, along with the emission conversion factors based on the EDGAR v4.2 emission inventory. The analysis was focused on evaluation of the uncertainty levels for the top-down NO_x and CO emission estimates and "hybrid" estimates (that is, those based on both atmospheric measurements of a given proxy species and respective bottom-up emission inventory data) of FF CO₂ emissions, as well as on examining consistency between the FF NO₂ emission estimates derived from measurements of the different proxy species. It is found that NO₂ measurements can provide much stronger constraints to the total annual FF CO₂ emissions in the study region than CO measurements, the accuracy of the NO₂-measurement-based CO₂ emission estimate being mostly limited by the uncertainty in the top-down NO_x emission estimate. Nonetheless, CO measurements are also found to be useful as they provide additional constraints to CO₂ emissions and enable evaluation of the hybrid FF CO₂ emission estimates obtained from NO₂ measurements. Our most reliable estimate for the total annual FF CO₂ emissions in the study region in 2008 (2.71 ± 0.30 Pg CO₂) is found to be about 11 and 5 % lower than the respective estimates based on the EDGAR v.4.2 (3.03 Pg CO₂) and CDIAC (2.86 Pg CO₂) emission inventories, with the difference between our estimate and the CDIAC inventory data not being statistically significant. In general, the results of this study indicate that the proposed method has the potential to become a useful tool for identification of possible biases and/or inconsistencies in the bottom-up emission inventory data regarding CO₂, NO_x, and CO emissions from fossil-fuel burning in different regions of the world.