The manuscript presents a concrete model for the conversion of NO to NO2 in point source emission plumes - an effect that is well known in principle, but hard to quantify in concrete cases due to its dependency on multiple parameters. Consequently, this process has so far been ignored in satellite-based point source emission estimates.

The authors propose a simple parameterization for the NOx to NO2 ratio and provide concrete parameters for four selected point sources. 

It is also shown that ignoring the in-plume conversion can result in significant underestimation of point source emissions from satellite data.

Thus, the topic of the study matches the scope of ACP, and the study contributes an important scientific progress.

The paper is generally well written and the presented material and the conclusions drawn are mostly comprehensible. In some cases, the authors should provide more quantitative or concrete information, as specified in the detailed comments below.

My main concern is that the paper gives concrete recommendations for an improved NO2 to NOx upscaling for 4 point sources which are based on only 2 days model simulation each.

The representativeness of the selected days for simulations is discussed rather qualitatively.

As the conversion time has to be expected to depend on stability, the wind speed probably plays an important role - an aspect that is not explicitely investigated or discussed in the current manuscript. 

Thus, the authors should at least add one further model simulation and fit of f(t) for each location, such that for each location a high wind speed and a moderate wind speed case are investigated.

The discussion of the representativeness of the model simulations and the impact of stability/wind speed should be updated/added accordingly.

I thus recommend publication after (a) additional simulations have been added, (b) the discussion of the representativeness of the simulation results is updated accordingly and the impact of meteorological stability is discussed, and (c) the additional comments listed below have been accounted for.

Additional comments:

General: Ratios are denoted as "NO2-to-NOx", "NO2 to NOx", or (reciprocal) "NOx:NO2". Please be consistent throughout the manuscript.

Line 21: I don't see (yet) the step towards a "global uniform" application. The study demonstrated that conditions can be very different for the investigated point sources.

Line 42: I don't think that the NO2 data from satellite has a higher accuracy than for CO2. The challenge for CO2 is to accurately measure the (small) CO2 *excess* on top of the high background. 

Line 52: Note that some recent studies do not use an "one for all" factor of 1.32, but account for the spatial variability of the ratio: e.g. 

- Beirle et al., 2021 (already cited)

- Lange, K., Richter, A., and Burrows, J. P.: Variability of nitrogen oxide emission fluxes and lifetimes estimated from Sentinel-5P TROPOMI observations, Atmos. Chem. Phys., 22, 2745–2767, https://doi.org/10.5194/acp-22-2745-2022, 2022. 

Fig. 2 (b): Please explain the black lines (current legend is not helpful). For positive distance, the black line seems to be a Gaussian fit (eq. 2?), but what is the meaning of the black line for the first subplot with negative distance, and what is done with these values?

Line 101: Subscript "eff" should be in text mode.

Line 101: Please add a ref to section 4.3 here.

Lines 103-105: What time interval is used for the emission fit? Do the results depend on this choice?

Line 114: Please provide further details about how the chemistry was "tuned". Was this done once for all or did you need to tune MicroHH for each location?

Line 121: Please provide further details and list the used bottom-up emissions with reference to table 1.

Section 2.1.2 / Table 1: Please explain how the simulation periods have been selected - probably because TROPOMI indicated a clear NO2 plume?

As written above, I would encourage the authors to add additional cases for each location. Ideally one day with high vs. a day with moderate wind speed. In any case, u_eff should be added to Table 1, as this is an important characteristic for the plume properties.

Section 2.2.1: It is mentioned later (3.2), but it would have helped me understanding the data processing if a sentence like "TROPOMI data were selected where plume detection worked successfully (see section 3.2 and Fig. A3)" would appear somewhere in section 2.2.1.

Line 164: I don't understand this point: why should a variable (VCD dependend) precision be problematic for the calculation of line densities?

In the TROPOMI NO2 retrieval, there are additive terms (stratospheric correction) and multiplicative terms (AMF). Due to the latter the NO2 VCD precision is inevitably correlated with the VCD and can be substantially larger than 1e15 close to strong emitters. 

This has a direct impact on the uncertainty of the emission estimate: If the AMF is wrong by 20%, the derived emissions are off by 20% as well (if the rest of the algorithm is perfect).

Thus, for a meaningful error propagation, the individual (VCD dependend) precision values have to be taken into account. 

Line 181: The "mixture of both" (mixture of correct concentration profile of plume and background?) is confusing. Please reword.

Line 188: This effect would mean that the observed SCD increases with distance, and thus (if not properly accounted for in the AMFs) the VCD as well. I.e. this effect interferes with the NO to NO2 conversion. Please give an estimate on the AMF change due to profile changes within the plume transport and discuss how far this could affect the estimated conversion rate r. 

Line 238: Please provide more quantitative information here and name the main reasons for the large differences:

- assumed emissions should be added to Table 1 and can be refered here

- "meteorological conditions" are basically u_eff which should be added to Table 1 as well

- while solar irradiance is of course important in general, I would not understand if this would explain the observed differences here, as for all cases cloud free conditions around noon are considered.

But if you find this to be actually different, please list the numbers as well.

- please specify the assumed O3 concentrations for the four point sources.  

- are the VOC concentrations very different for the four locations, and does this explain the far higher values for Matimba?

Section 3.2: The authors compare their derived f(t) with the often-used value of 1.32. For Jänschwalde or Lipetsk, f0 is close to 1.32, and the remaining discrepancy is the effect of NO to NO2 conversion in the plume.

But for Belchatow and particularly for Matimba, there are two reasons for the differences using 1.32: in addition to the ignored change of the in-plume NOx/NO2, also the final ratio is quite different.

As mentioned above, some recent studies account for the general place-dependent value of the NOx/NO2 ratio. In order to separate the impact of the wrong "background" value from the ignored in-plume NO to NO2 conversion, it would be very interesting to see comparisons of f(t) with the (still constant, but locally adjusted) value of f0, in addition or instead of 1.32.

Figures 6 and 7: shaded bars have different meanings in both figures - this should be made consistent.

Lines 300-302: Please comment on the needed effort for running the required simulations and the potentially required local "tuning".

Section 4.1: The selection of TROPOMI overpasses where a clear downwind plume shows up implies stable wind conditions. I would expect that the fitted f(t) is thus representing this special case, and average NO to NO2 conversion is probably quicker for less stable conditions.

This "selection bias" should be discussed somewhere.

Line 308: Again, I don't see photolysis as critical component here, as the selection of cloud free observations during noon results in high photolysis rates for all considered sites. If I am wrong please provide some quantitative statements on this.

Line 309: "Trace gas" would be O3, NOx and VOC, or something else as well?

Line 319: The authors should at least provide one more simulation per location with different conditions (wind speed) in order to better assess the question of representativity.

Lines 365-366: Please discuss how much of the discrepancy is due to ignorance of in-plume conversion, and how much is due to wrong "background" pss (which can easily be avoided) by the comparison of f(t) with f0, see comment to sect. 3.2.

Section 4.3

Taking u_eff just from the source location is a rough simplification. In 4.3, the authors discuss the difficulties of calculating appropriate effective wind speeds downwind due to plume mixing.

Apart from this, also the wind speed at a typical altitude might just change downwind. I would expect this to be rather common. As far as I understand, this effect is currently ignored completely.

Please add a discussion of this effect and its impact on the conversion of distance to time and the applied fits.

Line 393: Note that in Beirle et al. (2019), the correction factors of 1.35 and 1.98 are not intended to correct the total (tropospheric) VCD, but the VCD *excess* only (the enhancement of the VCD caused by the local point source), as the divergence of the flux automatically removes the local background. This is discussed in more depth in section 3.3 in

Beirle, S., Borger, C., Jost, A., and Wagner, T.: Improved catalog of NOx point source emissions (version 2), Earth Syst. Sci. Data, 15, 3051–3073, https://doi.org/10.5194/essd-15-3051-2023, 2023. 

Line 423: I would expect that stability / wind speed is a key component in this list.

Fig. A1: 

- I would expect that conditions are quite different during night. Due to lack of photolysis, f0 should approach 1. I would propose to skip (or at least to mark) nighttime in this plot.

- Why is there a gap of several hours for Lipetsk?

- For Matimba, time 6-12 UTC, the NOx/NO2 seem to *increase* with time - how can this be???