Estimating regional fossil fuel CO 2 concentrations from 14 CO 2 observations: challenges and uncertainties.

The direct way to estimate the regional fossil fuel CO surplus (ΔffCO) at a station is by measuring the ΔCO depletion compared with a respective background. However, this approach has several challenges, which are (i) the choice of an appropriate ΔCO background, (ii) potential contaminations through nuclear CO emissions and (iii) masking of ΔffCO by C-enriched biosphere respiration. Here we evaluate these challenges and estimate potential biases and typical uncertainties of C-based ΔffCO estimates in Europe. We show that Mace Head (MHD), Ireland, is a representative background station for the Integrated Carbon Observation System (ICOS) atmosphere station network. The mean ΔffCO representativeness bias when using the MHD ΔCO background for the whole observation network is of order 0.1 ± 0.3 ppm. At ICOS sites, the median nuclear contamination leads to 25% low-biased ΔffCO estimates if not corrected for. The ΔffCO masking due to C-enriched heterotrophic CO respiration can lead to similar ΔffCO biases as the nuclear contaminations, especially in summer. Our evaluation of all components contributing to the uncertainty of ΔffCO estimates reveals that, due to the small ffCO signals at ICOS stations, almost half of the C-based ΔffCO estimates from integrated samples have an uncertainty that is larger than 50%. This article is part of the Theo Murphy meeting issue 'Radiocarbon in the Anthropocene'.