This manuscript examines the link between European air stagnation and the large-scale circulation and synoptic-scale features such as blocking, Rossby wave breaking, and jet locations and strength. The analysis presented appears sound, and the manuscript does present some interesting results connection these different atmospheric features. However, I am not sure of the real value of this research. It is claimed that "understanding the development of stagnant conditions is therefore crucial for studying poor air quality" and this is the justification for this study.  However, I am not convinced of the importance for air quality,  as multiple recent studies show there is only a weak connection between air stagnation and air pollution events, and there is also evidence that some of the other atmospheric features examined are better predictors of air pollution events (and hence more crucial).  

The connections shown between stagnation and other circulation features is probably worth publishing (although I am not totally convinced), but before this manuscript is suitable for publication there needs to be a major revision of the abstract, Introduction, and conclusions that presents a more balanced view of importance of stagnation events on air pollution, and hence a more balance view of the potential impact of this study on understanding air quality and/or for the weather-climate dynamics community.

MAJOR COMMENTS

The abstract, Introduction and conclusions are full on statements that air stagnation is closely connected with air pollution, and that understanding stagnation events is key/critical for understanding air quality. However, very few of the referenced studies actually show a close connection between stagnation indices and air pollution (and many just assume there is a connection as it seems like there should be). In fact, several recent studies (that are not discussed in Introduction) actually show a weak, or at least inconsistent, relationship between air pollution and stagnation indices. The results of two of these (Kerr and Waugh (2018) and Garrido-Perez et al. (2019)) are buried in the conclusions (lines 461-462) but there are more studies: E.g., Huang et al (2018), Wang et al (2018), and Oswald et al (2015) show either weak relations between stagnation and pollution or that stagnation is not a strong predictor of pollution events. Huang et al and Wang et al are referenced but no mention is made that these studies cast doubt on importance of stagnation for air pollution events.  Given multiple studies all showing weak relationship many of the claims made in the paper on importance of stagnation are not justified, and these counter studies need to be discussed in the Introduction.

In addition to limited evidence for stagnation events being major driver for pollution events, there are many recent studies showing connections of the other atmospheric features examined here (e.g jet latitude, RWB, etc) with pollution (e.g. some of references on lines 460-465). If understanding air quality is the main focus then why not relate the large-scale or synoptic-scale features to air quality data directly, rather than indirectly through stagnation events? Put another way, the importance of an atmospheric feature in explaining stagnation events is not the same as the importance of this feature on air pollution (with some features probably being more important for air pollution than for stagnation events).

This last point makes me wonder why the analysis presented was not done relating the large-scale / synoptic scale atmosphere features to surface air quality. This would directly address the issue of air quality. 

MINOR COMMNENTS

Title:  As much of the paper focuses on blocking, ridges and Rossby-wave breaking, which are synoptic-scale and not large-scale features, the title needs to be modified. 

Line 119. Not sure what is meant by "averaged vertically (every 75 hPa ...". Do you mean averaged vertically between 925 and 700 hPa, with data every 75 hPa?   Also, later in the sentence I think should say "zonally between 0 and 60 W", as the 0-60 is important.

Line 124: Why 20E-60W as opposed to 0-60W used for eddy-driven jet?

Line 165: space missing "inGarrido"

Fig 2 and 3. Can you add a box to the maps to show the region being considered?  

Also, I think the histogram and lines in right panels make figures hard to read. Maybe just show the curves?

Line 245-250: I could not follow this discussion, and how it related to figure 4. For example, it is stated that Rossby waves have no impact for souther regions, but the RWB and other indices have a very similar variation in all panels in fig 4.

REFERENCES

Oswald E M, Dupigny-Giroux L-A, Leibensperger E M, Poirot R and Merrell J 2015 Climate controls on air quality in the northeastern US: an examination of summertime ozone statistics during 1993–2012 Atmos. Environ. 112 278–88

GENERAL COMMENTS: 



Maddison et al. examine the relationship between stagnation, defined by the Air Stagnation Index, and various dynamical features related to synoptic-scale circulation over Europe. By building a model using a MLR approach and stepwise procedure, they are able to explain a majority of the variance of stagnation frequency across different regions in Europe. They found regional heterogeneity and seasonality in the features (e.g., STJ, RWB) linked to stagnation.

I thought that the paper was well-written, and I appreciate that they tested other stagnation indices (Lines 411ff), as this has been a question of mine for quite some time. I do, however, think that the focus on stagnation is problematic and doesn’t really give information about pollution events. As discussed in my comments below, I believe this paper needs a direct assessment of the link between air pollutants and the large-scale dynamical features. If these comments are heeded, I believe the paper has the potential to be relevant for forecasting pollution events with short lead times. 

SPECIFIC COMMENTS:



• The stagnation index is problematic as it doesn’t correlate well with actual pollution events [Huang et al., 2018; Kerr & Waugh, 2018; Wang et al., 2019; Garrido-Perez et al., 2019] and it’s boolean, based on fairly arbitrary thresholds. The authors mention some of the known issues with the stagnation index in the Conclusions (Lines 461-462), but this comes across as a brief afterthought. In light of these issues, some of the findings contained in the study are not that groundbreaking or useful, especially for actually understanding the dynamical drivers of *pollution* events. 

For example when you describe stagnation frequency and where stagnant conditions are most likely to occur (e.g., Lines 30ff, Figure 1a, Lines 211-212), I’m not surprised that stagnation frequency is lower in mid-latitude Europe (near the latitude where the jet is located and where there are mid-latitude storms) versus in the southern part of your domain, where conditions might be more dry and where upper-level winds are not as great as in the mid-latitudes. Moreover, you also make statements like such as “air stagnation in Europe is shown to be strongly influenced by the large-scale circulation” (Line 432). Since stagnation, in essence, is just a measure of the local winds and precipitation, it doesn’t seem at all surprising that local weather conditions are influenced by the large-scale circulation. 

I believe that a more relevant study and one that would be impactful and important for the community would be to directly look at the relationship between pollutants and the dynamical features rather than the relationship between stagnation and the dynamical features, which may or may not have any direct relevant for surface-level pollution events. I think you could more or less repeat your methods and approach using the dynamical indices but subbing in perhaps PM and/or O3 from CAMS/ERA5. For example, you mention “stagnation is around 30% more likely to occur when there is a block or ridge present” (Lines 243-244). I would like to see a similar statement but for O3 or PM (e.g., “Extreme PM is X% more likely to occur when there is a block or ridge present”). As previously mentioned, I think this would be of wide interest to the air quality/atmospheric chemistry community. 

I noticed the other reviewer pointed out a similar comment, and hopefully by addressing this comment you could kill two birds with one stone….or, more accurately, you could shut up two reviewers with one additional analysis :)

References: Wang et al. https://journals.ametsoc.org/view/journals/bams/99/1/bams-d-16-0301.1.xml 

• Equations 1-3. You mention that variables such as BI_{region} and RWBI_{region} are monthly time series of blocked days. What do the distributions of these (and other) variables look like? Do they have many values = 0 and a few other values? I’m wondering if it’s appropriate to use linear regression to derive the coefficients in the case that the distributions are highly skewed. Plotting some examples would easily lend insight to this question. 

TECHNICAL CORRECTIONS:



• Figure 2a,b,e - It might be helpful to include latitude labels on these maps so readers can match them with the latitudes shown in c, d, f, and g. Additionally, can you clarify the quantities being shown in the shading in these subplots? For example, you mention in the figure caption that Figure 2b shows the “ridge index (shading)” but in the text you mention “an ~20% increase in the climatological frequency of subtropical ridging over the region” (Lines 178-179). So shouldn’t the quantities in shading be referred to as a departure from the index? Or am I misinterpreting? 

• Figure 2a,b,e - On a related note to my above comment, the changes in climatological frequency are generally at most +/- 20% but often quite a bit less than this value. Would, for example, a 5 or 10% increase in frequency even be statistically significant? Can the authors denote statistical significance here (and elsewhere, when applicable)? 

• Figure 3c - I understand that the RI departure plot is based on an index that considers three separate zonal sectors. However, it looks a little strange to me to have a negative box over part of Southern Europe and then a sharp cutoff to weak positive values. I realize that this affect is an artifact of the index, but what does it imply about its suitability? 

• Figure 6 - I think readers might benefit from having the region acronyms (e.g., SE, SW, CEU, etc.) labeled on the maps in case they forgot where these regions were.