REVIEW for GMD # gmd-2024-179:

 “Diagnosis of winter precipitation types using Spectral Bin Model (SBM): Comparison of five methods using ICE-POP 2018 field experiment data” by Bang et al.

Overview:

The authors evaluated five diagnosis schemes of identifying winter precipitation types using data from the ICE-POP 2018 field experiment. They found that the scheme using one-dimensional spectral bin model (SBM) with the climatological snow density-diameter relationship for the Pyeongchang region demonstrates superior performance. The manuscript is well written, clear, and easy to follow. I have only some minor comments regarding clarification or justification for consideration.

Specific Comments:

1. Section 2.2: How large uncertainties of these observations? The authors should discuss them to enhance the manuscript’s robustness.
2. Lines 156-162: Was there only one sounding available for each precipitation event? Should the earlier soundings be used as environmental profiles to diagnose precipitation types?
3. Lines 221-222: How to determine critical values for different sites? Please clarify.
4. Lines 293-294: Please justify “initialized as unrimed low-density snow aggregates”.
5. Lines 298-303: The authors argued that “the assumption of mass conservation” may be valid. However, how about PSDs? Given the same mass, PSDs at the surface and in the upper atmosphere could differ significantly. Please justify it.
6. Figures 10-12: Which SBM method, original or optimized one is shown in these figures?
7. Line 500: Why were not all examples within each group included, especially given the limited number of examples? A justification for this selection would be helpful.
8. Lines 546-547: Should the authors also consider updating the Vt-D relationship for ice particles?

Overview:

This paper is an evaluation of precipitation type diagnosis algorithms in a region of complex terrain and coastal influences in South Korea.  In general, the paper is well written, and the results are clearly explained.  I think that the paper is ready for publication after the authors address some minor issues.  

Specific comments:

Line 34: Does vaporization = evaporation?  I would recommend using evaporation here (as already used elsewhere in the paper), as it is more commonly used in meteorology and will be more familiar to readers. 

Lines 41-77: I think it would be worthwhile to mention precipitation type diagnosis algorithms that work in conjunction with microphysical parameterizations within numerical weather prediction models.  For example, the algorithm described in this paper:

https://doi.org/10.1175/WAF-D-15-0136.1

If you briefly described those algorithms, you could distinguish them from the types of algorithms you are evaluating herein (which are based purely on observations). 

Lines 309-321: Can you explicitly describe h.  Is it the overall hit rate?  Whereas h’ is averaged across three p types?  So if one p-type does particularly badly, but only has a few cases, h’ will be much lower than h.  Is that right?  I think the distinction between h and h’ could be more clearly described, which would help the reader interpret results. 

Lines 525-527: It seems like collision-coalescence is an important factor to include in SBM.  Can you provide some more detail on its effects and reasons for exclusion? 

The manuscript compared the performance of five different schemes for identifying winter precipitation types (limited to rain, snow, and mixed rain-snow) using data from the ICE-POP 2018 field experiments. The study demonstrated that an enhanced spectral bin model (SBM) provided relatively better results. While the manuscript contains valuable analyses and experimental results, the current presentation and English writing do not meet the publication quality. A major revision is recommended. Thorough proofreading will improve the manuscript greatly. Many places require further editing and some editing examples have been provided.

Major Comments:

1. The "hit rate" alone does not reveal the full picture. Suggest adding the False Alarm Rate (FAR) to get a better picture of the performance of different schemes
2. There are too many uncommon acronyms in the manuscript. Suggest reverting some of them back to their original forms as much as possible (acronyms in equations may be kept) to help readers understand easily.

WL -> warm layers, IL -> warm layer, CL -> cold layer, R-> rainfall rate, SR -> snowfall rate, etc

Also, it will be good to add a "List of Acronyms"

          3.  The hydrometeor falling velocity is usually called the "terminal velocity" and referred to as "Vt" instead of "Vf".

          4.  Lines92-93: "Tw-Γlow nomogram (Sims and Liu, 2015) methods" -> Need to be consistent in the manuscript on how to refer to this method, either "the Tw-Γlow method" or "the Sims and Liu method". Avoid using different names in different locations (such as lines 355, 566 etc) for the same method. Revise similarly for other methods throughout the manuscript.

Comments:

Line 24: "The results show that the SBM has the highest overall skill score for winter precipitation":  SMB did not outperform all other 4 schemes in all situations as demonstrated in the manuscript. Please revise accordingly.

Line 29: "which uses climatological relationships for Colorado region" -> "which uses a snow density-diameter relationship for the Colorado region".

Line 73: "The addition of sublimation and evaporation is predicated on the idea that these processes may…"  –> "The addition of sublimation and evaporation is because these processes may…"

Lines 87-88: "the intensive observation data density…" –>  "The high-density observational data provided by the ICE-POP network enables a comprehensive evaluation and refinement of previously proposed WPT diagnostic methods." 

Lines 142-143: "a low-speed mode … of graupel or hail":  Why hail has a low fall velocity (terminal velocity)?

Fig. 2:  "Fsite" is not defined in Fig. 2b

Line 172: How is the "normalized frequency" computed?

Line 177: why is a threshold value of "0.05" critical? How do we get this threshold?

Lines 249-250: "For example, …": (1) Based on Fig.4d, Zero Tw and a lapse rate of 6 °C/km correspond to about 80% probability. Where did the "probability of 0.86" come from? "0.45" is not accurate either, (2) Suggest adding a zero Tw line to help users examine Fig4d.

Lines 259-261: This sentence is redundant or it should appear where the Tw0-Γlow method was first introduced.

Line 286: How were the rainfall rate and the snowfall rate data obtained?

Lines 408-413: "For RASN, …": The last sentence should be moved to the location before "For RASN…" and then revise the remain part accordingly.

Line 314: What's the purpose of using h'? It looks like h' is not needed since the hit rates were calculated individually for SN, RASN and RA and plotted in figures (such as Fig. 6) which are enough for the discussions.

Line 340: "We evaluate the accuracy of the H850 method": "Accuracy" has a special meaning and formula in statistics. Suggest changing the "accuracy" to "performance" or similar words throughout the manuscript.

Lines 346-351: What is the conclusion from this discussion ( about different thresholds for the Tw0-Γlow method)? If there are no conclusions from this discussion, it may be removed.

Lines 364-365: "The Tw0 method exhibits a relatively large difference between h (86.3 %) and ℎ′ (68.4 %), with the inclusion of Tw0 improving the diagnosis of SN (Figs. 6c and 6d).": (1) it discussed two conclusions in one sentence and does not read well; (2) this sentence can be removed without any evident impact.

Lines 364-367: The low performance in diagnosing RASN in Fig. 6c (the Tw0 method) should be highlighted here and h' is not needed.

Lines 367-368: "This is supported by…(Sims and Liu, 2015)." This logic here is not clear. Consider revising.

Fig. 6: "0(0.0%)" labels can be removed from the figure.

Lines 391-395: "The dependence of the skill scores on the terrain for all five methods is also explored…:  The paragraph before this one already discussed the impact of terrain on the skill scores. Consider revising to make the logic smoother or merging this paragraph to the previous one.

Lines 395-435: These paragraphs did not discuss the "dependence on wet-bulb temperature profiles" and hence should be placed before section 4.2. Either put them in section 4.1 or add a new section.

Line 407: "In contrast": I don't think there is an evident "contrast" here. It is safe to remove it.

Lines 408-412: "For RASN…": move the last sentence and put it before "For RASN…"

Lines 417-420: Two sentences here, and the second one repeats the first one.

Line 425: "The Γlow of RA varies widely, though it tends toward negative values": Fig. 9d showed that most Γlow values are positive. So why it is stated "tends toward negative values"?

Fig. 10: remove the "RA" legend as there are no RA cases in this figure.

Lines 428 and 474: What is "complex atmospheric profiles" or "complex atmospheric vertical structure"? Need clarification on this.

Table 3: Need a separating line between Group 1 and Group 2

Fig. 13a,b,c: No discussion on these three figures in the manuscript. If not needed, they can be removed.

Fig. 14b, f: NO discussion on these two figures. If not needed, they can be removed.

Fig 14e,h: No description of these two figures in the figure caption.

Line 540 Why define "Doppler velocity" as -Vr instead of Vr? 

Line 542: Figs. 15c and 15d  -> Figs. 15a and b

Lines 582-583: Do you mean "These results suggested that SBM simulations tend to produce less melting compared to the observed precipitation"? 

Lines 590-591: "The performance of the original SBM was superior to some existing optimized methods (the H850 and RH0-T0 nomogram methods)"  ->  add "some" before "existing" since SBM did not outperform all methods in all situations

Lines: 592-593: (1) "should" -> "will";  (2) what does "other reanalysis filed data" refer to? (3) how would the 3D WPT algorithm differ from the 2D WPT algorithm?

Edits:

Line2 145-146: "very strong inversion strength"  -> "very strong inversion"

Line 269: "the next largest size" -> "the next larger size"

Line 340: "an RH0-T0 nomogram" -> "the RH0-T0 method"

Line 341: "a Tw0-Γlow nomogram" -> "the Tw0-Γlow method", revise similarly throughout the manuscript

Line 343: "The lowest h is achieved by the RH0-T0 nomogram" -> "The lowest h is from the RH0-T0 method"

Line 400: "radiational cooling" -> "radiative cooling"

Line 420: "The Tw0 for cases" -> "The Tw0 for RASN cases"

Line 428: "A cold RASN case" -> "a RASN case"