The topic of relating observed surface snowfall rates to local enhancements in radar reflectivity is relevant, interesting, and well within the scope of ACP. However, in ACP’s Aim & Scope it is mentioned that “articles with a local focus must clearly explain how the results extend and compare with current knowledge.” It is not clear how the local focus (NE U.S.) could be extended/applied to non-orographic snowfall in other parts of the word, so the paper would benefit from some additional context.

This work presents relatively extensive and comprehensive analyses of snowfall data spanning 11 winter seasons and 264 storms days, leading to some novel findings about the relationship between the spatial distribution of enhanced radar echoes and associated snowfall amounts at the surface. The primary conclusion is that—contrary to more limited case studies focusing on extreme snowfall events in the northeastern U.S.—high rates of snowfall do not necessarily correspond to locally enhanced reflectivity features (i.e., mesoscale snow bands) in a large dataset such as this. In fact, only about one out of every four hourly occurrences have heavy surface snow rates when these enhanced reflectivity features pass over a site. This conclusion is indeed substantial.

The methods are laid out in a way that is mostly clear and easy to follow, although there are a few areas highlighted in my comments below added clarity is needed. All assumptions appear reasonable and results are thoroughly laid out and explained in a good amount of detail, leading to the conclusions in a logical manner. The paper is well-organized and well-written, overall.

Specific comments:

Abstract: It would be helpful to include one or two sentences about the purpose of this study. Perhaps something like ‘relationship between multi-bands and frontogenesis is not clear’ from lines 30-31.

Line 63: I’m not sure what you mean by ‘better than a factor of 2’? Is this based on some prior work that you could cite here?

Line 67, 78: Is there no lake-effect snow in upstate New York? Any reason to believe that KALB and KGBM wouldn’t be impacted by lake-effect snow (especially KGBM)? What is the reasoning for excluding lake-effect snow and how important is it that the dataset isn’t contaminated by these events?

Line 83: What is difference b/w 29 ASOS stations and 14 stations of GHCNd? Are the 14 GHCNd stations simply for determining dates/times of non-orographic winter storm events? What measurements were included? Just daily snow depth?

Line 91: Do you have a reference for ‘do not have a heated rim and thus are subject to capping’? It seems this could be a significant problem that could use more discussion here. For example, are there times that this could be happening and resulting in an underestimation of the snow accumulation? Is it possible that it could happen and report lower-than-actual amounts of snow (rather than no snow at all)?

Line 95: Is a collection efficiency of less than one something that you correct/account for? If so, how? If not, why is it not necessary?

Line 114: Is the interpolation technique described in detail elsewhere? I see later on that you cite Tomkins et al. (2022) for details. Can you add a few words to describe the technique? Is it linear interpolation? Or is it too lengthy/complicated to mention here?

Lines 321-322, 325, 327: It has been a little while since I’ve looked at a VAD wind profile. When you say ‘lots of wind shear illustrated by the wind barbs’—please describe what the direction is showing (e.g., any vertical component or horizontal only?). Are there higher wind speeds in the CFAD in 13e that aren’t seen in 13a? I think a little bit more description of what we are looking at for wind direction would help here.

Line 324: For the triangle icons it would be helpful to see a version of one of these images with a 1:1 aspect ratio—perhaps rotated to fit and place in the supplemental? Or just a subset perhaps. This would be nice to reference and easier to see how this would change streamer angles as you describe.

Line 336: There is no aircraft vector in 15a like there is in 13a & 14a.

Line 338: Can you comment on features/scales resolved by some of these radars & not by others of note here, and how you would or would not expect that to influence the interpretation?

Line 342: It didn’t become obvious to me until this line that the 3:1 aspect ratio is not image aspect ratio but distance aspect ratio (I think because I am used to seeing aspect ratio used with respect to image dimensions rather than physical dimensions). Perhaps it is not just me and this could be made clear the first time it is mentioned.

Line 344: The streamers appear to be streaming in direction opposite that of wind flow (i.e., the same direction that the aircraft is heading in). Or perhaps I am misunderstanding the wind flow direction or the orientation of the streamers?  

Line 429: Is there any physical/dynamical reasoning to support the observation that the highest snow rates tend to be in the northwest and northeast quadrants?

Figure 5: I think it is worth it to explicitly mention here that blue horizontal lines correspond to left-hand side (or coloring the left-hand vertical axis blue?)

Figure 6: I'm not sure if a) is adding much value. Is it more useful to show the intensity of the low centers than the tracks? Perhaps color-code beginning to end of tracks rather than pressure. Should we be seeing any kind of pattern in the intensity of these low centers? If not then consider omitting or coloring based on beginning to end (e.g., lighter at beginning to dark at end of path).

Figure 9 (and possibly others): You don’t mention what the gray colors are here. These are where you removed melting/mixed pixels?

Figure 13: CFADs of (e) wind speed and (f) wind direction; this wind direction in (f) is the direction from which the wind is blowing, correct?

Figure 14: Perhaps just say “Same as Figure 13 except for…”

Table 1: How is change to mass per unit volume (IWC/LWC) determined? Is this IWC/LWC a ratio or are you indicating an increase to one or the other or both?

Technical/typing corrections:

Abstract (line 7): ‘echo’ should be plural ‘echoes’

Line 60: I believe it should be hyphenated ‘surface-snow-producing’

Line 110: NWS acronym should be defined upon first use in Line 43

Line 199: ‘in terms joint’ should be ‘in terms of joint’?

Line 299: ground-based-scanning-radar-observed (I think would be the proper hyphenation here)

Line 306: snow form should be ‘snow to form’?

Line 337: I think this is supposed to ‘Fig. 15’ not ‘Fig. 15e’

Line 379: ‘is’ should be ‘are’

Line 380: ‘how a individual features’ should be ‘how a storm’s individual features’?

This study investigates an important topic regarding how to improve the quantification of surface snowfall rate from radar reflectivity. The authors applied a large amount of objectively analyzed data from NWS operation radars and airborne and ground-based radars from a field campaign, as well as ASOS data. I find that the analysis is unique and logic, and the conclusions are mostly reasonable. However, I would like to see more clarifications, such as the interpretation between the defined feature and the continuous radar reflectivity values. Some other comments are given below as well.

Specific comments:

Line 31:  The presence of frontolysis in snowbands was analyzed in Han et al. (2007) through solving the Sawyer Eliassen equation. It is relevant to the “frontolysis” discussion here. Please refer to the study and its frontolysis anlysis.

Table 1: It is a good idea to consider the microphysical processes and the difference between the mass and radar reflectivity. But please put in justification. A thorough justification may have to consider radiative transfer calculation, which is not necessary if the authors do not already have those experience or knowledge. But some degree of justification is necessary.

Line 55: “Unlike convective cells …” this sentence is not clear.  I think you are probably saying that in convective cells, the snow/graupel particles aloft melt while they fall to warmer temperature at low levels, which is the melting level. So, the enhanced reflectivity column from the frozen particles does not usually extends to the surface …… However, in snowstorms, it is different. Please make this sentence clear.

Line 63: last sentence of this paragraph. Can you please quantify the range of the factor as well?

Line 67: the key finding is not clear. I think it needs to be clarified that snow bands are instantaneous features in the radar reflectivity at a certain level above the ground. Even if it can be related to the surface snow rate, like you would argue for rain, it is still just instantaneous snowfall rate. Or you may need to say frontogenesis-related primary band are more likely to be associated with hourly accumulated surface snowfall, but multi-bands may not.

Line 130: Please provide information how the reflectivity is rescaled to snow rate. Also, please clarify that this snow rate is not liquid-equivalent as ASOS’s is.

Line 146 and 147: You define two metrics called ‘area x time fraction’. Both are based on the definition of feature, the first one includes strong and faint features, and the second one adds in the background. The features and background are sort of masks from the radar reflectivity, which does not reflect the continuous value from the radar reflectivity. So, why not use radar reflectivity to define a metric to account for more continuous change of the radar reflectivity?

Line 150: I think you mean “Fig. 5d”.

Line 182: Do you mean “Fig. 6a”?

Line 250: “equating snow bands with heavy snow will … over prediction …”. I understand the goal and the method of this study. However, in Numerical Weather Prediction (NWP) practice, I don’t think the snowfall prediction is based on radar reflectivity bands, neither the observed nor the simulated radar reflectivity. I think this statement needs to be modified. Please also give reference of how National Weather Service use NWP models for snowfall prediction.

Line 250 to 253: The problem of using feature mask, instead of the value of reflectivity, to quantify the persistence of the intensity of the snowband over a specific ground site is that the feature mask actually includes a large range of reflectivity values and reflectivity-derived snow rate. Like what you have shown in Figs. 5a and 5b, how do you quantify the changes in the magnitudes within the strong feature?

Also, is snow rate in Figure 5a non-liquid-equivalent? It needs to be clarified.

Section 3.1: It is good to have your figures and analysis supplemented by the video. I would like to view that scenarios in the lower left (89.1%) and upper right (1.5%) quadrants are consistent in supporting that long-lasting elongated high-reflectivity regions observed by operational ground radars correspond to higher surface snowfall rate measured by ASOS. The the low-area x time faction just occurs much more often than high-area x time fraction.

Differences between Fig. 9 (upper right, 1.5% quadrant) and Fig 10 (lower right 4.6% quadrant), is it possible to provide the comparison of the histograms of the reflectivity values that were included in your feature area x time fraction analysis? As the feature mask may have masked out the changes of reflectivity within the feature. Of course, Fig. 9 has slightly larger feature area x time fraction than Fig. 10, 0.62 vs. 0.58, which may have contribution to higher surface snowfall rate too.

As you pointed out in the last scenario (Fig. 12), the radar beam height may be above the locally enhanced features. It is very likely that the height of the radar beam, i.e., the distance between the radar site and the ground station also plays a role.