Dear authors,

please find my review comments in the attached pdf file. 

I reviewed two previous versions of this manuscript that was submitted to another journal.  This new version has incorporated some of my recommendations in a cursory way, but unfortunately major scientific flaws remain, and I have no choice but to recommend rejection of the paper.  While this paper has very interesting data and results, the current presentation and interpretation is not of sufficient quality for publication. 

Because these are so many major issues, I found it difficult to go line-by-line, instead giving my overall view of the paper:

• This paper presents a set of atmospheric ∆¹⁴C measurements, from which fossil fuel CO₂ (CO₂ff) can be calculated. The authors have made some problematic assumptions in their CO₂ff calculation, particularly regarding the choice of background.  Figure 3 shows very clearly that winds are very different in summer and winter in this region, and therefore using a single background site is likely to be problematic.
• The ∂¹³C analysis is difficult, because there is not much separation between isotopic values of sources, the biogenic CO₂ ∂¹³C value is not well constrained, and the atmospheric signals are also small, which results in very large uncertainties of ~25% for the partitioning. Thus, the interpretations of changing source sectors made in Figure 5 are not valid.
• In the discussion of the CO₂ff spatial variability shown in Figure 1, the authors seem to equate the patterns of CO₂ff mole fractions to spatial patterns in emissions. Since mole fractions are influenced by emissions and atmospheric variability, there is no direct connection between mole fractions and emissions, and therefore this analysis is not valid.  The authors include FLEXPART footprints (Figure 3), but do not attempt to use them to relate emissions to the observations – this would be a sensible way to make such a comparison.
• The authors compare their CO₂ff mole fractions with other datasets measured in the same cities in earlier years, and claim that they can observe changes in CO₂ff emission rates. Again, mole fractions are influenced by emissions and atmospheric variability, and for this comparison, the specific locations of the measurements much also be taken into consideration.  Therefore this comparison is also not valid.
• The examination of CO:CO₂ff ratios is more compelling, and the authors could reformulate a paper that removes the problematic points and focuses on these results.