The Precipitation Imaging Package: Assessment of Microphysical and Bulk Characteristics of Snow

Round 1

Reviewer 1 Report

• Precipitation events containing rimed particles should be added in the analysis. Also one site observation at MQT cannot conclude the advantage of PIP.
• Even though the title of this paper is “Introducing the precipitation imaging package: Assessment of microphysical and bulk characteristics of snow”, snow microphysical characteristics from mass retrieval methods from von Lerber and Wood are only presented in section 3.2. There is no results from PIP. The title should be changed or results of the microphysical characteristics from PIP should be presented.
• It would be better to write down the unit of variables/parameters. Also appendix that contains the symbols and their meaning with the unit can increase the readability of this paper.

Specific comments are shown below.

• Line 198 and other lines: “c is 0.15, x is -0.86”à minus sign should be written as “ 

• Table 1: the caption should include “low SLR snow events”.
• The reviewer suggests the authors to discuss how these high-order products from PIP can be utilized to model development and evaluation.
• Line 711: the authors do not have to conclude the temperature dependence of the N0 and lamda, which did not present in this manuscript.
• “ctime/rho_liq” is missing in the equation (4), compare to the equation. What is the unit of R in equations (1) and (4)?

Author Response

Please see attached PDF

Author Response File: Author Response.pdf

Reviewer 2 Report

1. English grammar need to be improved.
2. Last para of Introduction part  is vague and need to be revised. Novelty of the paper shall be clearly presented.
3. Please describe the reason why the large variation is found in High SLR in comparison to low SLR in figure 3.
4. Please describe the reason for large deviation in the results for Figure 6 (d) in case of high SLR snow events for Wood Snow rate LWE
5. PIP SLR vs Wood SLR performance is weak for both low and high SLR as depicted in Figure 7 (b) and (d). Why?

Author Response

Please see the attached PDF

Author Response File: Author Response.pdf

Reviewer 3 Report

1. Repeated use of ‘this work’ in abstract

2. L 39: Suggest modifying colloquial phrase ‘are key’.

3. L 49: precipitation -> hydrometer

4. L 137: Word order - “Alter-type wind shield”

5. L 143: Suggest removal of unnecessary word “successfully”

6. L197: Should rho_e be defined here as a symbol, to help with L196 and 197?

7. The use of rho_e for both equivalent and edensity is confusing.  Please make these distinct symbols in formulae.

8. Could a mathematical symbol be used for edensity instead of the word edensity?  Such as a capital rho?

9. L742: Suggest rewording ‘not able’ - this is missing a preposition I believe.

10. General comment - can the conclusions be made more quantitative by highlighting selected statistical results from the previous sections?

Author Response

Please see the attached PDF

Author Response File: Author Response.pdf

Reviewer 4 Report

Review of “Introducing the Precipitation Imaging Package: Assessment of microphysical and bulk characteristics of snow

Recommendation: Major revisions required.

MAJOR COMMENTS

This paper looks at the processing of data obtained by the Precipitation Imaging Package (PIP) over a multi-year period at the NWS Marquette Office and compares three different methods that are used to derive the higher order products characterizing the precipitation characteristics. Overall the paper is well written and for the most part (exception noted below) the methodology is easy to understand and the results seem scientifically sound. Further, given the PIP will play an important role in the evaluation of GPM retrievals, the subject matter is of interest and is timely. Thus, I think that the paper should be published. However, there was one aspect of the presentation (calculation of effective density) that was not clear. Further, I have a number of other suggestions that could improve the quality of the presentation. Thus, I am recommending major revisions before the paper can be deemed acceptable for publication.

The overall conclusion of the study is that the PIP produces physically consistent bulk snow characteristics through comparison with established mass retrieval methods. However, this conclusion is hard to interpret. First, there is no motivation on what is good enough consistency in these products to meet the goal of the study. Is 1% good enough? Is 10% adequate? Without some sort of overall justification in how good the consistency or products needs to be, it is hard to interpret this conclusion. There should be better uncertainty estimates. The authors say this will be subject of a future paper, but surely a paper assessing the characteristics of snow should include these error characteristics in order to interpret the results. The paper could also be improved if the conclusions were more quantitative (e.g., the snow characteristics were consistent within xx% of those derived with established methods).

I find it curious that the authors refer to the different calculations of the PIP properties as retrieval methods. To me, a retrieval method in atmospheric science typically refers to retrieval of geophysical quantities (e.g., cloud properties such as effective radius, liquid water content, total concentration, or temperature, humidity, aerosol concentration, etc.) rather than derivation from an in-situ probe. Only one of the three “retrieval” methods used in this paper involve remote sensors. Can a term other than retrieval be used to describe this methodology.

I find the use of the word “Introducing” in the paper title strange. I would think that the first paper using the PIP or the first paper deriving physical quantities from the PIP could be justified with the use of the word introducing. But, this is not the first such paper so I would recommend removing the word introducing from the paper title.

I think the writing in the paper should be made more quantitative. Although I don’t provide all specific instances in the paper where this is a problem, I do point out a couple of examples. In the abstract, it is stated that “this work confirms the ability of the PIP derived products to quantify properties of snow rate and equivalent density and demonstrates that the PIP produces physically realistic snow characteristics.” If these statements were made more quantitative this would be much better. What is the accuracy of the derived snow rate/equivalent density? What are the error characteristics on the snow characteristics? Next, in the abstract it is stated that “this work illustrates fundamentally different microphysical and bulk features of low and high snow to liquid ratio events….” What does fundamentally different mean quantitatively? A 10% difference? A factor of 2? If these statements are made more quantitative, the paper will be much more useful: there always will be an error in derived quantities, the question is how much the error is: a 0.01% error is very different than an 80% error! There are multiple places in the manuscript where this lack of quantification should be addressed, and the above is just two examples of these problems.

In addition to being more quantitative in the writing, MDEF

My most major comment relates to the methodology for the calculation of density from the Precipitation Imaging Package. First, the words equivalent density (edensity), effective density, density, bulk density, average density, and bulk equivalent density are used. It should be clearly stated a priori what all of these different densities refer to in order to avoid any confusion, and maybe an Appendix defining the different terms would help. Terms are scattered throughout the manuscript which may be part of the reason I had trouble following the derivation of the density in Section 2.2.1. It is stated that the equivalent density (edensity) is developed using the discrete snowfall events and the snow field observations. The average bulk density rho(t) is determined so that when integrating Eq. (4) one gets agreement with the R measured for the event. But, if this is being averaged over the whole event, there would not seem to be time dependence so how is rho as a function of t determined? It would seem to be constant over the whole event. And, more importantly, since there is an integration over De in equation 4, how does one obtain the edensity as a function of De? It would seem that there is no information to derive rho as a function of De so I don’t see how the resulting edensity can be fit to a power law relationship similar to Equation 3. Perhaps I am missing something here, but I cannot follow what the authors are doing here. Because of not understanding this, I also don’t see how it can be further refined by relating to the relative fall speeds. When I was reading this section, I was also concerned about how the packing of snow on top of each other affects the calculations, but I saw that this was at least acknowledged as a source of error later on. I think it should be explicitly mentioned here as well.

Assuming that the above point can be adequately described, there would then seem to be a question regarding Eq. 7. The parameter R is calculated. But, it appears that rho_e was derived using the measured value of R as a constraint. So, provided all the calculations were done correctly, should this not just be the original value of R used to derived it?

In addition to showing the differences in the exponential fit parameters, it would also be nice to show a density plot of the size distributions as that would be a more straightforward way of seeing the differences in the low and high SLR cases.

Although an interesting read, I’m not convinced that Section 3.4 adds much to the manuscript beyond what was already presented in the statistical analysis. If something needed to be removed to reduce the length of the manuscript, I would remove this section. Note that instead of this section I would have liked to have seen more comments on why the various methods differed in the way that they did. Is a conclusion that the Wood algorithm is not as good as the other 2 algorithms? If so, why? What assumptions are made that leads to problems. Although there is a very good description of how the algorithms differ and under what circumstances they differ, there could be improvements made in describing why the algorithms differ.

It would also be useful to include a measure of bias between the algorithms. Whether biases exist between the quantities is of fundamental importance. Quantities such as mean absolute error, mean absolute percent error, mean bias percentage and root mean squared error should all be given when comparing the different “retrieved” quantities.

MINOR COMMENTS

Line 44. For snow it is not more complicated processes that drive estimation uncertainties, but rather more complicated properties. Properties are measured, processes are inferred from those mesurements.

Line 46: I don’t think particle size distributions are a bulk characteristics: a bulk characteristic is something like a water content, total concentration or effective radius that takes care of integrating over the whole size distributions.

Line 53: “are active areas of research”

Line 56: “and ITS predecessor”

Line 110-111: Why in particular is a 1-min resolution chosen? It would seem that the averaging period should depend on the intensity of the rain. While there could be sufficient counting statistics under conditions of heavy precipitation in a 1-min period, there might not be when the intensity is less. There should be some more discussion on what is needed to get a statistically significant particle size distribution.

Line 163: I cannot understand the reason why a constant C_time is needed in this equation. There may be some conversion of units required to derive units of mm/hr for R, but surely this is true for almost any equation that is written and a constant is typically not included in an equation because of this.

Line 270: I don’t think this constant is needed.

Lines 288-290: I understand that this is not the methodology of the authors, but it seems quite arbitrary to choose a constant and method so that the “best” agreement is found. This does not seem a good way to learn a lot about the processes that are at work here.

Lines 303 and 308: I think some characters were not properly printed out on my version of the manuscript.

Line 313: What parameters for projected area and shape?

Line 413: What do you mean quantitatively by slight?

Line 451: This paragraph has shown the SLR needs to be considered, not snow type and environmental conditions. Although the two are no doubt closely related, the concluding sentence should be reworded (and a call for future investigation on dependence on cloud type and environmental condition would be fine)

Author Response

Please see the attached PDF

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The manuscript has been improved based on the comments. 

Reviewer 4 Report