Potential of ¹⁴C-based versus ∆CO-based ∆ffCO₂ observations to estimate urban fossil fuel CO₂ (ffCO₂) emissions

Abstract. Atmospheric transport inversions are a powerful tool for independently estimating surface CO₂ fluxes from atmospheric CO₂ concentration measurements. However, additional tracers are needed to separate the fossil fuel CO₂ (ffCO₂) emissions from natural CO₂ fluxes. In this study we focus on radiocarbon (¹⁴C), the most direct tracer for ffCO₂, and the continuously measured surrogate tracer carbon monoxide (CO), which is co-emitted with ffCO₂ during incomplete combustion. In the companion paper by Maier et al. (2023a) we determined for the urban Heidelberg observation site in Southwestern Germany discrete ¹⁴C-based and continuous ∆CO-based estimates of the ffCO₂ excess concentration (∆ffCO₂) compared to a clean-air reference. Here, we use the CarboScope inversion framework adapted for the urban domain around Heidelberg to assess the potential of both types of ∆ffCO₂ observations to investigate ffCO₂ emissions and their seasonal cycle. We find that discrete ¹⁴C-based ∆ffCO₂ observations from almost 100 afternoon flask samples collected in the two years 2019 and 2020 are not well suited for estimating robust ffCO₂ emissions in the main footprint of this urban area with a very heterogeneous distribution of sources including several point sources. The benefit of the continuous ∆CO-based ∆ffCO₂ estimates is that they can be averaged to reduce the impact of individual hours with an inadequate model performance. We show that the weekly averaged ∆CO-based ∆ffCO₂ observations allow for a robust reconstruction of the seasonal cycle of the area source ffCO₂ emissions from temporally flat a-priori emissions. In particular, the distinct COVID-19 signal with a steep drop in emissions in spring 2020 is clearly present in these data-driven a-posteriori results. Moreover, our top-down results show a shift in the seasonality of the area source ffCO₂ emissions around Heidelberg in 2019 compared to the bottom-up estimates from TNO. This highlights the huge potential of ∆CO-based ∆ffCO₂ to verify bottom-up ffCO₂ emissions at urban stations if the ∆CO / ∆ffCO₂ ratios can be determined without biases.