Improved method for linear carbon monoxide simulation and source attribution in atmospheric chemistry models illustrated using GEOS-Chem v9

Abstract

Abstract. Carbon monoxide (CO) simulation in atmospheric chemistry models is frequently used for source-receptor analysis, emission inversion, interpretation of observations, and chemical forecasting due to its computational efficiency and ability to quantitatively link simulated CO burdens to sources. While several methods exist for modeling CO source attribution, most are inappropriate for regions where the CO budget is dominated by secondary production rather than direct emissions. Here, we introduce a major update to the linear CO-only capability in the GEOS-Chem chemical transport model that for the first time allows sourceregion tagging of secondary CO produced from oxidation of non-methane volatile organic compounds. Our updates also remove fundamental inconsistencies between the CO-only simulation and the standard full chemistry simulation by using consistent CO production rates in both. We find that relative to the standard chemistry simulation, CO in the original CO-only simulation was overestimated by more than 100 ppb in the model surface layer and underestimated in outflow regions. The improved CO-only simulation largely resolves these discrepancies by improving both the magnitude and location of secondary production. Despite large differences between the original and improved simulations, however, model evaluation with the global dataset used to benchmark GEOS-Chem shows negligible change to the model's ability to match the observations. This suggests that the current GEOS-Chem benchmark is not well suited to evaluate model changes in regions influenced by biogenic emissions and chemistry, and expanding the dataset to include observations from biogenic source regions (including those from recent aircraft campaigns) should be a priority for the GEOSChem community. Using Australasia as a case study, we show that the new ability to geographically tag secondary CO production provides significant added value for interpreting observations and model results in regions where primary CO emissions are low. Secondary production dominates the CO budget across much of the world, especially in the Southern Hemisphere, and we recommend future model-observation and multi-model comparisons implement this capability to provide a more complete understanding of CO sources and their variability.

Source Contributions to Carbon Monoxide Concentrations During KORUS‐AQ Based on CAM‐chem Model Applications

“…Such characteristics make tagging CO feasible and tagged CO relatively reliable as a tracer of pollution plumes from regional to hemispheric scales. Tagged CO has been widely used in previous studies for various research purposes such as source attribution (Buchholz et al, 2016;Chen et al, 2009;Fisher et al, 2017;Granier et al, 1999;Liu et al, 2003;Park et al, 2009;Pfister et al, 2011Pfister et al, , 2004Protonotariou et al, 2013;Staudt et al, 2001) and inverse modeling (Arellano et al, 2004(Arellano et al, , 2006Heald et al, 2004;Pétron et al, 2004). Our goal in this study is to elucidate the regional sources contributing to observed CO concentrations within the troposphere over Korea during the KORUS-AQ campaign using the tagged CO algorithm that is implemented in the Community Atmosphere Model with chemistry (CAM-chem).…”

“…Sources apportioned by tagged CO are sensitive to many parameters such as emissions, transport, chemistry, and resolution in the CTM. These factors are important sources of model errors (Gaubert et al, 2016;Müller et al, 2018;Naik et al, 2013;Strode et al, 2015;Yan et al, 2016), which could lead to CO underestimation commonly seen in most global CTMs but have not been fully understood yet (Fisher et al, 2017;Shindell et al, 2006;Stein et al, 2014;Tilmes et al, 2015). To provide insights on the sensitivity of our findings on source contributions to the aforementioned factors, we conduct an ensemble of model simulations with different model configurations and report the range of estimates of the source contributions from these simulations.…”

Simultaneous shipborne measurements of CO<sub>2</sub>, CH<sub>4</sub> and CO and their application to improving greenhouse-gas flux estimates in Australia

“…For CO, biomass burning plays an important role, as the main driver of the CO seasonal and interannual variability across the Southern Hemisphere (Edwards et al, 2006a, b). Overall, total CO in Australia is dominated by non-methane volatile organic carbon (NMVOC) and CH 4 oxidation (Té et al, 2016;Fisher et al, 2017), with negligible influence from anthropogenic emissions (Zeng et al, 2015). While prior work has provided some constraints on Australia's greenhouse-gas sources, both these studies and others have shown lingering differences between modelled and measured concentrations, implying that some sources of greenhouse gases in Australia remain missing or underestimated (Fraser et al, 2011;Loh et al, 2015).…”

TCCON Philippines: First Measurement Results, Satellite Data and Model Comparisons in Southeast Asia

“…We use regional tracers from the CO simulation to identify sources of observed enhancements recorded in the TCCON data. Because we are interested in the sources of variability rather than absolute concentrations, we first correct a known low bias in the GEOS-Chem CO simulation, driven at least in part by a high bias in global OH [27]. We calculate an offset using a weighted least squares fit of the GEOS-chem output to the TCCON data, computed only on days when both model and observations suggest the site is sampling clean background air from the Pacific.…”

“…The offline CO sim ulation is described by Fisher et al [27] and was run at 2 • × 2.5 • horizontal resolution with 47 vertical levels, from January 2016 until the end of May 2017. Fossil fuel emissions globally were taken from EDGAR (Emission Database for Global Atmospheric Research) version 4.2 [28], with regional overwrites for Asia [29], Europe [30], Canada [31], the United States [32], and Mexico [33], along with biofuel emissions from Yevich and Logan [34], ship emissions from the International Comprehensive Ocean-Atmosphere Data Set (ICOADS) [35] and Aviation Emissions Inventory Code (AEIC) [36].…”