Source Sector Attribution of CO 2 Emissions Using an Urban CO/CO 2 Bayesian Inversion System

Abstract

We assimilate multiple trace gas species within a single high-resolution Bayesian inversion system to optimize CO 2 ff emissions for individual source sectors. Starting with carbon monoxide (CO), an atmospheric trace gas with fairly well-known emissions, we use emission factors of CO and CO 2 ff (called R CO ) defined for each source sector to enable us to jointly use CO and CO 2 atmospheric mole fractions to constrain CO 2 ff sectoral emissions. We first show that our combined CO-CO 2 inversion is theoretically capable of estimating the relative magnitude of sectoral emissions for two, specially defined sectors over Indianapolis, while CO 2 -only inversions failed at quantifying sectoral emissions. When assimilating hourly mole fractions collected over 4 months, inverse sectoral emissions converge toward high-resolution CO 2 ff bottom-up emissions from Hestia. The emission ratios between the two sectors agree within 15% with Hestia across various inversion configurations. The assimilation of CO mole fractions preferentially improves flux estimates from traffic emissions, because the CO levels originating from the combustion engine sector are large relative to those from other economic sectors. In a further investigation, we find that including an additional third tracer sensitive to the other sectors only slightly improves the accuracy of the inversion compared to our current two-sector inversions with CO and CO 2 mole fractions. We finally examined the impact of errors in trace gas emission factors and quantify their relative impact on sector-based inverse emissions. We conclude that multispecies inversions can constrain sectoral emissions at policy-level uncertainties if trace gas emission factors are sufficiently well known at the city level.