Dear Authors,

the reviewers acknowledge the improvements of the revised version of your manuscript.
However, there are still some minor points which should be addressed, see below.
After addressing these points, the paper can be accepted for publication.

With best regards,

Thomas Wagner


1) In response to my reviewer comment, the authors added a definition of "persistency" in their analysis which is very helpful. Still, I find the definition lacking in technical details. The authors state that "it is checked how many days back in time that particular wind direction was continuously sustained". This must involve some threshold of acceptable deviation. Were winds allowed to deviate by up to 10 degrees? 50 degrees? 1 degree? or did they truly need to be continuously sustained, to the same decimal point? This is of course a trivial detail, but I don't understand why it can't be included in their manuscript to help readers understand their exact approach. And how sensitive is their identification of "two distinct modes" to whatever threshold they chose?

2) The authors have added a sensitivity test based on cloud screening which is certainly helpful. Still, in their manuscript methods they state "we allowed retrievals under partially cloudy conditions to be analyzed" (this is repeated in Section 4). This statement is misleading, since it implies to me that there is some filtering that has been done ("partially cloudy"). Surely some scenes may be fully cloudy. The authors need to state explicitly that they have not applied any filter for cloudy data, and that they test the impact of this later.

3) I actually agree with Reviewer 2 regarding the question about the assumption that NO2 observed from satellite is indicative of enhanced NO2 levels on the ground. But I also agree with the authors that this is not necessarily within the scope of their current paper. However, I am a little disturbed that the author response is that they "put faith" in the datasets and "hope" the tropospheric columns would capture the variability near the ground. It is not the responsibility of the satellite retrieval science teams to guarantee the tropospheric columns have any relevance to surface conditions. It is also not the responsibility of the referee to point to references that discuss a discrepancy, as requested by the authors in response. Rather, in my opinion, it is the role of the authors to convince us that there isn't a discrepancy, or argue that any discrepancy is not important to their conclusions (which may very well be the case). Perhaps a compromise would be for the authors to include a few comments/caveats to this point specifically, or to include references to other literature that supports any relevant assumptions.

4) I commend the authors’ approach to Figure 2, using monthly thresholds instead of seasonal thresholds. However, this figure is difficult to read. This could be corrected by simply including the absolute values of the 90th percentile for each month beside the legend labels. Also, why not label the colors by the month name, instead of a number (the rest of the plots refer to month names ("DJF", "SON", etc.)).

* The newly added sentences on page 11, line 51 do not look right to me - emissions are not given in ppb. Please check

* if I have not overlooked this information, the threshold for the cloud screening which was applied in the test case is not given. Please add.

* it would be really good to make Fig. 10 identical to Fig. 3 with respect to figure sizes and legend

* I do not understand the sentence on page 14, line 163 "By definition, NO2 anomalies during extreme events are similar in magnitude to climatological values over Scandinavia - please explain

* page 14, line 153: concentrations => columns