Implications for deriving regional fossil fuel CO2 estimates from atmospheric observations in a hot spot of nuclear power plant 14CO2 emissions

Using Δ¹⁴C observations to infer the local concentration excess of CO₂ due to the burning of fossil fuels (DFFCO₂) is a promising technique to monitor anthropogenic CO₂ emissions. A recent study showed that ¹⁴CO₂ emissions from the nuclear industry can significantly alter the local atmospheric ¹⁴CO₂ concentration and thus mask the Δ¹⁴C depletion due to DFFCO₂. In this study, we investigate the relevance of this effect for the vicinity of Toronto, Canada, a hot spot of anthropogenic ¹⁴CO₂ emissions. Comparing the measured emissions from local power plants to a global emission inventory highlighted significant deviations on interannual timescales. Although the previously assumed emission factor of 1.6 TBq(GWa)^–1 agrees with the observed long-term average for all CANDU reactors of 1.50 ± 0.18 TBq(GWa)^–1. This power-based parameterization neglects the different emission ratios for individual reactors, which range from 3.4 ± 0.82 to 0.65 ± 0.09 TBq(GWa)^–1. This causes a mean difference of –14% in ¹⁴CO₂ concentrations in our simulations at our observational site in Egbert, Canada. On an annual time basis, this additional ¹⁴CO₂ masks the equivalent of 27–82% of the total annual FFCO₂ offset. A pseudo-data experiment suggests that the interannual variability in the masked fraction may cause spurious trends in the DFFCO₂ estimates of the order of 30% from 2006–2010. In addition, a comparison of the modeled Δ¹⁴C levels with our observational time series from 2008–2010 underlines that incorporating the best available ¹⁴CO₂ emissions significantly increases the agreement. There were also short periods with significant observed Δ¹⁴C offsets, which were found to be linked with maintenance periods conducted on these nuclear reactors.