Work

Continuous fossil fuel CO2 emission estimates based on 14C-calibrated urban eddy-covariance fluxes of co-emitted species

Year: 2026

Type: article

Abstract: Atmospheric tall-tower measurements of radiocarbon ( CO 2 14 ), carbon monoxide ( CO ), and nitrogen oxides ( NO x ) show great potential for determining fossil fuel (ff) CO 2 emissions. While CO 2 14 relaxed eddy accumulation (REA) enables F ff CO 2 estimation during individual hours, F CO and F NO x , measured directly and continuously by eddy covariance (EC), can be used as proxies for continuous F ff CO 2 once local F proxy / F ff CO 2 ratios ( R proxy ) are known. We combined for the first time half-hourly EC F CO and F NO x , measured along F CO 2 at a tall tower in Munich from July 2024 to February 2025, with co-located F ff CO 2 from 76 hour-long CO 2 14 REA measurements. First, we examined the general applicability of CO - and NO x -based F ff CO 2 estimation by analyzing the C 14 -based R proxy . Then, we applied the C 14 -calibrated 〈 R proxy 〉 to the F proxy time series to obtain continuous ff CO 2 and non-fossil (nf) CO 2 fluxes, and analyze the F CO 2 composition. We demonstrate the potential of the purely observation-based partitioning approach by comparing R proxy and half-hourly F ff CO 2 and F nf CO 2 with modeled fluxes obtained from a footprint-weighted state-of-the-art Munich emission inventory and biospheric flux model. Although variability in R proxy limits precision and accuracy of half-hourly, proxy-based F ff CO 2 , average daytime fluxes could be estimated with about 20% uncertainty. On average, F ff CO 2 accounted for 70% of F CO 2 during daytime. The absolute modeled F ff CO 2 was 46% higher due to increased modeled heating emissions in the cold months. Observation-based and modeled F nf CO 2 indicated substantial human respiration and minor biospheric flux contributions. The multi-proxy EC and REA measurements enabled unique analysis and independent evaluation of urban ff CO 2 and nf CO 2 fluxes.

Source: Atmospheric Environment X

DOI: 10.1016/j.aeaoa.2026.100524