This paper presents prospects of using NOx as a proxy to quantify fossil fuel CO2 contributions (ffCO2). The paper is mostly based on analysis of time series of NOx observations in Heidelberg. Although there are many aspects that complicate the NOx-based ffCO2 estimation, the authors do a good job in presenting the potential. The paper is overall clearly written, and I request only minor revision, specifically concerning:

1. The interpretation of contributions from traffic and/or heating in different seasons
2. The potential interpretation of night-time accumulation of NOx/CO/CO2 in summer, when the method works poorly due to short NOx lifetimes

Further comments are in the annotated pdf.

The manuscript entitled “Using NO_x as quantitative fossil fuel CO2 proxy in urban areas: challenges and benefits” describes the use of NO_x and ¹⁴CO₂ measurements in Heidelberg (Germany) to estimate the seasonally averaged ratios of local excess NO_x (ΔNO_x) to local excess fossil fuel (ff) CO₂ (ΔffCO₂) concentrations and to derive ΔffCO₂ concentrations from the NO_x measurements. The analysis presented in the manuscript is overall sound and sufficiently detailed. However, I believe the scope of this study is too narrow, and I doubt that the presented results are sufficiently significant to warrant publication of the manuscript in ACP. My specific concerns and some suggestions are outlined below.

Major points

1. The title of the manuscript is not quite appropriate, since the study addresses only one possible way to use NO_x observations as a ffCO₂ proxy, that is by using ¹⁴CO₂ measurements. A more common way does not require the availability of rather scarce ¹⁴CO₂ measurements but instead involves emission ratios from inventories. Hence, I suggest narrowing the title, for example as follows: “Using NO_x as quantitative fossil fuel CO₂ proxy based on ¹⁴CO₂ measurements in urban areas: challenges and benefits”.
2. It is quite obvious that variations of NO_x (linearly scaled or not) should correlate with variation ¹⁴CO₂ in vicinity of relatively strong local emission sources. It is also clear that this correlation can never be perfect due to atmospheric chemistry and the fact that the NO_x-to-CO₂ emission ratios differ for different source types. Furthermore, since the relationships between the NO_x- and ¹⁴C-based ffCO₂ are likely to differ (quantitatively) for various sites across the Europe and the world, the quantitative findings of this study are not necessarily relevant to any other sites. Thus the results of the study seem relatively trivial and the scientific message is unclear. Unfortunately, the authors did not show how the performance of the NO_x-based ffCO₂ record depends on the configuration of the local domain, even though they noted in the Introduction that “the choice of a suitable and common background for all species is of paramount importance”. A more detailed examination of this point (with dedicated test cases) could perhaps help justifying the study and generalizing its results.
3. The NO_x-based ΔffCO₂ concentrations are representative of an area of about 30x80 km (according to Fig. 2). Hence, given that ¹⁴CO₂ data are available in Europe from just a dozen or so operational sites, the proposed method can hardly help constrain the European carbon budget (unlike the original ¹⁴CO₂ observations; see, e.g., https://doi.org/10.5194/acp-25-397-2025). However, the study’s significance in this context could be increased if the authors used their ¹⁴CO₂ measurements to evaluate the inventory-based ΔNOx/ΔffCO₂ ratios. A similar analysis (but in the case of the ΔCO/ΔffCO₂ ratios) was conducted in a recent study authored by several co-authors of the given manuscript (https://doi.org/10.5194/acp-24-8183-2024).

Other points

L 41-42. I suggest clarifying here that the NO_x concentration excesses (ΔNO_x) are derived from in situ observations of both NO_x and ¹⁴CO2 concentrations. Otherwise, this statement looks misleading.

L 82-83. Were the samples taken evenly throughout the day or mainly during certain hours?

L 104: “This would lead to a higher uncertainty of the derived ΔNO_x-based ΔffCO₂ estimates (see Sect. 2.3)”. This statement is actually not proven in Sect. 2.3.

Fig. 2(b): Values of the travel time are missing in the plot.

L 123-124. The sentence is hard to understand. I suggest rephrasing.

L 209: “40%”: Is this uncertainty for individual grid cells? If so, should not the uncertainty in the simulated NO_x concentrations be much smaller than that (because random errors of emissions in different grid cells tend to compensate each other when these emissions mix in the atmosphere)?      

Fig. 4. In Sect. 3.1, the authors discuss the seasonally averaged concentration ratios, but Fig. 4 shows scatter plots for all valid flask measurements. Could the authors provide similar scatter plots for summer and winter separately?

L 395: “concentrations show a strong correlation with R² values over 0.8”. This statement appears to contradict the results reported on line 274 (for all summer samples … an R² value of 0.55 is found), especially since the next sentence refers to the seasonal ratios.