Dear authors,

Thank you for revising the manuscript, which addresses the reviewers’ comments. In spite of one request, I do not think it is necessary for the reviewers to review the revised manuscript once again. Yet I do think the manuscript is in need of another close inspection and revision before it can be published in AMT. None of the issues with the manuscript should be considered ‘major’, but the current text is too long and in need of quite a few clarifications and corrections still. Below please find a list of corrections and suggestions. Please respond to these issues, and revise the manuscript accordingly.

Kind regards – Folkert

1. The claim that the paper presents an approach to quantify soil NOx emissions “globally with a spatial resolution of 0.25 deg” is overstated. I agree that the method can be applied anywhere in the world, but it is only practically useful over selected areas dominated by soil NOx emissions. Much of the paper focuses on the Sahel, as already pointed out by one reviewer. The other two regions discussed in any detail are Australia and southern Africa. Please tone down phrases mentioning the global character of this study, e.g. the “global with spatial resolution of 0.25 deg” aspect of this study in the Abstract and in the Conclusions (page 20, lines 10-11). A more realistic claim would be to state that you have applied the method over all continents, and that it provides meaningful constraints on soil NOx emissions over a number of specific regions that are dominated by soil NOx emissions.
2. The introduction (page 3, paragraph starting on line 10) does not mention at all “canopy reduction”, e.g. the phenomenon that nitrogen oxides emitted by soils are quickly deposited on available vegetation surfaces. Still this is an important “environmental condition” influencing the effective soil NOx emission to the atmosphere.
3. In section 2 and later in the paper, I was surprised not to see a discussion on soil temperature data. Controlled experiments and field data suggest that soil NOx emissions are positively correlated with higher soil temperatures, conducive to stronger microbial activity. This aspect should be discussed.
4. Another aspect not mentioned but likely playing a role in the analysis, is “smearing” beyond the 0.25 deg cell, or beyond the study area. Suppose grid cell i has its first rain event on day 0, but adjacent grid cell i+1 only has the peak on day 4. In this situation, the time series for grid cell i may be expected to have a local peak on day 0 with an additional enhanced NO2 contribution from day 4 (if downwind of cell i+1). In a general sense, such contributions will influence (raise) the background beyond contributions from sustained local soil NOx emissions alone. This should be acknowledged or further investigated (See specific issues below).
5. The suggestion to implement your findings in models is a good one. I strongly encourage you to provide modellers with some practical hints as to how to do this. Preferably you give an equation that represents a parameterization of the pulsing events. Such an equation takes into account the most important drivers of the pulsing events: length of dry spell, amount of precipitation of Day0, land cover type. It is important to also quantitatively relate the pulsing emissions to the non-pulsing soil NOx emissions. Models generally include parameterizations for those (Steinkamp and Lawrence, Hudman-papers), and these would have to be tweaked if a pulsing contribution is accounted for. If you could show how this could be done, this would add a lot of value to your paper.

Minor issues:
P2, l20: please also include by Vinken et al., ACP, 2014 as a study indicating that soil NOx emissions are regionally underestimated.
P2, L27-29: suggest to merge the sentence: “Findings … from soils” with the previous paragraph.
P3, L21: also the study from Bertram et al., GRL, 2006 should be mentioned here.
P5, L2-4: the AMF also depends on the satellite viewing geometry (solar and viewing zenith angles, relative azimuth angle).
P5, L13-15: please check carefully with the documentation on www.temis.nl or with Ronald van der A whether FRESCO or FRESCO+ was used in your GOME-2 and SCIAMACHY product versions. If FRESCO+, please include the appropriate citation.
P5, 21: swath should be indicated as 2600 km, not km^2.
P5, L28-29: the instantaneous NOx lifetime and emission diurnal cycle also contribute to the differences between SCIAMACHY, GOME-2 on one hand, and OMI on the other. This should be acknowledged. Generally NO2 in the early afternoon is shorter-lived than in the morning. This effect may (partly) offset by soil NOx emissions peaking in the afternoon rather than in the morning (higher temperatures).
P10, L15: 50% of possible data available per pixel because of reasons of spatial representativeness? For a detailed discussion of this aspect, please see Boersma et al., GMD, 2016. That paper discusses how uncertainty from spatial representativeness becomes the dominant uncertainty contribution if only few valid satellite pixels per grid cell are available. It would be worthwhile (for future proper use of the satellite data) to discuss the issue briefly and refer to the Boersma et al. [2016] paper.
P14, L14-15: the pixel size differences between GOME, SCIAMACHY, and OMI also lead to different CF distributions.
P14, L21: typo Fig. 9.d.
P14, L25L typo “are” should be ‘is’ (the influence is investigated).
P15, L12-16: in this section also smearing should be discussed. To me it seems unlikely that over a larger region like the Sahel, all your pixels have the first rain on the same Day0. If the first days of rainfall vary over a region, then the background for pixel i will certainly be influenced by advection of soil NOx peaking earlier or later than Day0. This should be acknowledged and preferably also be inestigated in more detail (e.g. by selecting only those pixels from the Sahel that have the exact same Day0. If the distribution of Fig 12(a) does not change appreciably for this subset, then smearing may be of little importance.
P16, L30-32: “NO2 VCDs stay enhanced after Day0 for a period of about two weeks”. In line with my previous comment, this may be a convolution of structurally enhanced soil NOx emissions (microbes activated by the first rains) and the result of smearing. Your paper would be more convincing if you could wiggle out the effect of smearing.
P17, L24: “with an approach similar to ours”. This sounds strange, since Hudman’s approach was published well before this paper. I suggest to revert the statement and keep it at that your approach is similar to theirs.
P19, L7-12: I had to read a couple of times before I understood that you arrive at the 21-44% estimate for the contribution of pulsing to total soil NOx emissions (10-21 GgN) by dividing your own pulsing estimates by the enhancement of the background (46.4). Please clarify this in 5.2.
P20, L12: Hudman should not be between brackets.
P20, L34: “NO2 VCDs show moderately enhanced NO2 VCDs” … this should be rephrased.


References
Bertram, T. H., A. Heckel, A. Richter, J. P. Burrows, and R. C. Cohen (2005), Satellite measurements of daily variations in soil NOx emissions, Geophys. Res. Lett., 32, L24812, doi:10.1029/2005GL024640.

Boersma, K. F., Vinken, G. C. M., and Eskes, H. J.: Representativeness errors in comparing chemistry transport and chemistry climate models with satellite UV–Vis tropospheric column retrievals, Geosci. Model Dev., 9, 875-898, doi:10.5194/gmd-9-875-2016, 2016.

Dear authors - Congratulations on a very thorough and well-written manuscript! It is accepted for publication in ACP - the editor.