Estimating the Snow Density using Collocated Parsivel and MRR Measurements: A Preliminary Study from ICE-POP 2017/2018

Chang, Wei-Yu; Yang, Yung-Chuan; Hung, Chen-Yu; Kim, Kwonil; Lee, Gyuwon

doi:https://doi.org/10.5194/egusphere-2023-3147

Preprints

https://doi.org/10.5194/egusphere-2023-3147

Preprints

16 Jan 2024

| 16 Jan 2024

Estimating the Snow Density using Collocated Parsivel and MRR Measurements: A Preliminary Study from ICE-POP 2017/2018

Wei-Yu Chang, Yung-Chuan Yang, Chen-Yu Hung, Kwonil Kim, and Gyuwon Lee

Abstract. A new method is developed to derive the hydrometer's bulk density and bulk water fraction from collocated measurements from Micro-Rain Radar (MRR) and Particle Size and Velocity disdrometer (Parsivel). Rigorous particle scattering simulation, namely the T-matrix method, is applied to particle size distribution data of Parsivel to calculate the reflectivity (Z_HH). The possible combinations of the particle's ice, air, and water are derived to compare them with the MRR-measured Z_HH. The combination of minimum water fraction subsequently determines the bulk density (ρ_bulk). The proposed method is applied to the data collected from the International Collaborative Experiments for Pyeongchang 2018 Olympic and Paralympic Winter Games (ICE-POP 2018) Projects and its pre-campaign. The estimated was examined by self-evaluation of reflectivity weighted fall velocity (Vz) of MRR and independent comparison of the liquid-equivalent snowfall rate (SR) of collocated Pluvio. The bias values are adequately low (Vz: -0.27~0.14 m s^-1, SR: 0.52~0.74 mm hr^-1). The correlation coefficient of calculated SR from estimated ρ_bulk and observed SR from Pluvio can be up to 0.74. The results indicate the capability to derive reliable ρ_bulk through the proposed method, leveraging the compact and easily deployable designs of MRR and Parsivel. The derived bulk density of the two warm-low cases (28 February and 07 March 2018) shares a similar transition as the systems were decaying. The particles with higher bulk density and bulk water fraction were found in the coastal sites (BKC and GWU: mean ρ_bulk values are 0.12 to 0.25 g cm^-3), typically accompanied by higher liquid-water constituents (mean values of the top 5 % bulk water fraction are 0.33 to 0.50) than the inland sites (YPO and MHS: mean ρ_bulk values are 0.09 to 0.08 and mean values of the top 5 % bulk water fraction are 0.001 to 0.029) during such synoptic conditions.

Received: 27 Dec 2023 – Discussion started: 16 Jan 2024

Download & links

Wei-Yu Chang, Yung-Chuan Yang, Chen-Yu Hung, Kwonil Kim, and Gyuwon Lee

Status: final response (author comments only)

RC1:
'Review of egusphere-2023-3147', Anonymous Referee #1, 14 Feb 2024

## Summary
The authors propose a new method to estimate the bulk density and liquid fraction of falling snow and mixed-phase precipitation from collocated measurements of a micro rain radar (MRR) and a particle size and velocity (Parsivel) disdrometer. The method uses T-matrix simulations to retrieve the reflectivity factor (Z) from the Parsivel particle size distributions (PSD) assuming spherical precipitation particles with different combinations of ice an liquid water volume ratios. The simulated Z is compared with the observed Z from the MRR to constrain the possible solutions. The method is applied to two case studies and evaluated by comparing the derived bulk density with the reflectivity-weighted fall velocity (V_Z) from the MRR and the liquid-equivalent snowfall rate (SR) from a collocated precipitation gauge. The results show reasonable agreement between the derived and observed quantities, and are able to capture spatial and temporal variations of the bulk density across different sites and synoptic conditions.
The proposed bulk density retrieval method should be interesting and relevant to the ACP readers. However, I have some major concerns especially on the retrieval of bulk water fraction (v_w). My critical comments are presented below.
## General comments
Do the authors see a possibility that the retrieved high values of bulk density and liquid fraction during periods of low reflectivity and precipitation rate could be, at least partly, an artifact from applying the method on observations of very weak precipitation? The retrieval seems somewhat unstable in these conditions which is not surprising considering factors such as low signal to noise ratio due to weak signal. Yet, the authors draw considerable attention to the retrievals of these periods of very weak precipitation. It is worth considering which disciplines would benefit from the microphysical retrievals of weak mixed phase precipitation (<0.5mm/h)? I would suggest either introducing a threshold for minimum reflectivity or precipitation rate where the method is applied, or otherwise critically reviewing the method's performance in weak precipitation, where the precipitation rate falls below the sensitivity of the Pluvios.
I don't feel that the retrieval of v_w is adequately demonstrated in the case studies since there seem to be no time periods where there would be both a) agreement between derived and measured terminal velocity and b) notable precipitation intensity at the same time. It raises the concern how the method would perform in mixed phase situations with larger particles or higher precipitation rates. In my mind, the best way to address this would be to replace one of the case studies with another one that has significant intensity of wet snow, or discuss the possible limitations of the method's application.
In the presented case studies, it seems like non-zero v_w values only occur when bulk density is nearly saturated at over 0.9 g/cm. I'm concerned whether this is physically reasonable especially given the assumption of spherical particles. This raises the question whether, effectively, the liquid fraction would act as a kind of an overflow buffer in the calculations when density alone cannot explain the reflectivity values. This could rise from the assumption of maximum ice volume fraction (v_i). Or could it be explained with the drizzle-like nature of the precipitation during the v_w signals? This concern could be dispelled with a counter example.
The manuscript lacks discussion on the implications of assuming spherical particles. This might be significant consideration given the wide range of different particle habits and their shapes and preferred falling alignments.
Since the manuscript considers the liquid fraction, it would be worth showing or at least mentioning if melting layer signals were detected in the MRR observations.
The viewpoint in the manuscript is more technical and focuses less on microphysics. As such, the topic is suitable for ACP but might be even better suited for AMT. This is a possible consideration for resubmission after revisions.
The text is well organized and easy to follow. Relevant data and findings are visualized and given in tables. However, I noticed quite a few grammar errors especially with prepositions and articles, and therefore recommend some form of thorough language checking process for the manuscript.
## Specific comments
The title refers to snow density, but retrievals are attempted for snow, mixed phase precipitation and light drizzle. If liquid fraction plays an important role in the revised manuscript, it would be good to be reflected in the title.
The title indicates this is a preliminary study. Are the authors working on a more comprehensive analysis? Worth mentioning in the discussion.
L42: This sentence seems to suggest that riming and melting are the only processes affecting snow density. While these are important, one should not forget that, e.g., the primary particle habit and aggregation have great impact, too.
L59: As one of the selling points of the new density retrieval method is the use of robust intsruments that require little maintenance, perhaps it would be good to mention studies that use, e.g., PIP or Parsivel for density retrievals. Such studies could be found, for example, by doing citation analysis on Brandes et al. (2007) and Huang et al. (2010), as referred to in the manuscript.
Section 2: Details about the ambient temperature measurements are missing, in particular, the types of the instruments or sensors used. The temperature measurements, while not part of the main retrieval methods presented, should be of great interest for the reader as an indication for melting and other microphysical processes.
Section 2: I would like to see basic information about the Pluvios used such as make, orifice size and shielding.
L130: What is meant by a canting angle when referring to spherical particles?
L133: Why was this temperature range chosen for the simulations? Is it physically reasonable to simulate mixed-phase precipitation in -10 degrees Celsius, for example?
L147: I failed to find references to liquid water fraction from Huang et al. (2010). They seem to just assume particles to consist of ice and air. Either I missed it, or this reference could be more accurate.
L263: "The particle size was" to 'Maximum particle size ranged from'
L279: Since there are many sites involved in this study, it could be useful to have a small map showing their locations.
L315: "number density function" to 'number concentration'
L331: How is mean bulk density calculated here? Is it integrated over the total volume?
L334: Since rain was not excluded from the retrievals, we are now talking about the mean bulk density of mixed precipitation instead of snow. As the densities of rain and snow are quite different, the mean value is easily driven by the fraction of rain, masking the possible signal from snow properties.
L345: This sentence seems to suggest that the density of snow has an impact on the weather. It's unclear to me what was meant here. Please rephrase to clarify.
L526: "The ZHH variation with vw is much less than that with vi". I think, the opposite is true.
L560: What does the "shaded area" refer to in Fig. 5c?
Figures 5d, 9d-11d: There seem to be flat parts in the temperature measurements, where the measured value does not seem to change even for a fraction of a degree. These look like a measurement errors, perhaps gaps in the measurements. This raises concerns about the quality of the temperature measurements. The measurements should be checked and erroneous values omitted from the analysis. If there is a cause for concern about the quality of the measurements, it should be discussed in the manuscript.
Figure 6: The integration times in these figures are quite long. For example, in Fig. 6b, there seem to be multiple modes in the (D, v) distribution. Because of the long integration time, it is unclear if these modes are co-existing or if the dominating particle type is evolving over time. Have the authors considered analyzing the particle properties such as (D, v) distribution in shorter time intervals?
## Technical comments
The manuscript uses a mixed notation style for the symbols of quantities. Some are in italics, some are not, some use subscripts, some don't. I recommend more consistency in the style of these symbols. One could see the style guides of the publisher, if available, or other sources, for reference.
L10: Authors should choose between the spellings "disdrometer" and "distrometer". Currently, mixed spelling is used for this word in the manuscript.
L27: Since riming refers to a process and not a hydrometeor type, change "riming" to 'rimed particles'.
L73-74: The readability of this sentence could be improved by rephrasing.
L121: The authors probably meant 'modified FROM Huang et al. (2010)'.
L142: "as the increasing" to 'with increasing'
L178: "examined with" to 'compared with'
L186: "four with Pluvio" to 'four of them equipped with a Pluvio'
L196: "consistency of" to 'consistent'
L232: Full stop after "shown in Fig. 5"
L267: "relatively weaker" to 'weaker'. There are also other instances of using "relatively" with a comparative in the manuscript. I advice against this.
L274: "precipitation system" to 'precipitation'
L303: "has a more consistent relation" to 'is more consistent with the relation'
L345: "microphysics processes" to 'microphysical processes'
L353: "Consistency" to 'Inconsistency'
L361: The manuscript could be more consistent with how dates are written: with or without ordinal indicators.
L501: "blue lines" to 'blue', as there are also other symbols than lines.
L524: "DSD" to 'PSD'
L569: "derived" to 'shown'
Figure 5d: "MRM" to 'MRR' in the y-axis label.
Figure 6c: label (c) missing from figure

## References
References in this review follow the same notation as the reviewed manuscript.

Citation: https://doi.org/10.5194/egusphere-2023-3147-RC1
- AC1: 'Reply on RC1', Wei-Yu Chang, 05 Apr 2024
  
  Dear Reviewer,
  The authors sincerely appreciate your valuable comments and suggestions to help improve the manuscript. We have revised the manuscript titled “Estimating the Snow Density using Collocated Parsivel and MRR Measurements: A Preliminary Study from ICE-POP 2017/2018 ”. that was submitted to ACP (Atmospheric Chemistry and Physics) on 3 January, 2024. Based on your suggestions, we have put substantial effort into additional analysis. The manuscript has been thoughtfully revised regarding the comments from all reviewers.
  One of the major concerns of the proposed density retrieval algorithm using collocated MRR and Parsivel is lacking the uncertainty analysis. As per the reviewer’s suggestion, we have performed substantial investigations of the retrieval uncertainty. The impacts of the measurement uncertainty of the Parsivel and the MRR on the bulk density retrieval are analyzed quantitatively. The measurement issue of Parsivel is also investigated to understand its impact on bulk density retrieval. The results are summarized in the revised manuscript as a Discussion section.
  The MRR data quality issue has been examined per the reviewer’s suggestion. The post-processed data have replaced entire MRR raw data by applying the algorithm from Maahn and Kollias (2012). All the bulk density, bulk water fraction, and reflectivity-weighted velocity retrievals have been recalculated. The figures have been revised as well.
  The original purpose of utilizing reflectivity-weighted velocity to filter adequate retrieval is no longer needed and has been removed in the revised manuscript. The quality of the retrieval results have been greatly improved by applying the post-processed MRR data per the reviewer’s suggestion. The low SNR MRR measurement has been removed. The comparison of reflectivity-weighted velocity is mainly used to identify the inadequate retrieval due to the attenuation effect on MRR reflectivity.
  The performance of the retrieved bulk density has been validated by the snowfall rate (SR) from collocated Pluvio measurements and reflectivity-weighted fall velocity (Vz) from MRR. In addition to SR and Vz, the performance of the retrieved bulk density has been compared with the precipitation imaging package (PIP), a video disdrometer (Newman et al., 2009; Pettersen et al., 2020). The PIP was also deployed at the MHS site during ICE-POP 2018 (Tokay et al., 2023). The comparison of retrieved bulk density between the proposed algorithm in this study and PIP has shown good agreement with each other. The high consistency further confirms the performance of the retrieved bulk density. Since there is no direct bulk water fraction measurement for validation, the authors consider the validation of bulk density retrieval to PIP and Pluvio as “indirect” evidence to support the bulk water fraction retrieval.
  The SR and Vz validation analysis shows that the algorithm can adequately retrieve the bulk density and bulk water fraction. The consistency of the retrieved bulk density to collocated PIP confirms the performance of the proposed algorithm in this study. The advantage of the proposed algorithm is that it utilizes collocated Parsivel and MRR, which are commercially available, commonly used, and robust instruments. The Parsivel and MRR can operate unattentively and need little maintenance. Further application of the proposed algorithm helps derive long-term observation data on snow properties. The authors believe the proposed algorithm can provide an alternative choice if sophisticated instruments (e.g., 2DVD, PIP, SVI, MASC) are unavailable.
  The manuscript has also been revised carefully following the reviewer’s suggestions on English wording. The authors would like to express our sincere appreciation for the comments. The point-to-point replies to every comment have been prepared. Please see the following replies. The added or modified sentences in the revised manual are in red for your convenience. We would appreciate any feedback on the revisions.
  
  Citation: https://doi.org/10.5194/egusphere-2023-3147-AC1
RC2:
'Comment on egusphere-2023-3147', Anonymous Referee #2, 22 Feb 2024

Please see the attached review document.

Citation: https://doi.org/10.5194/egusphere-2023-3147-RC2
- AC2: 'Reply on RC2', Wei-Yu Chang, 05 Apr 2024
  
  Dear Reviewer,
  The authors sincerely appreciate your valuable comments and suggestions to help improve the manuscript. We have revised the manuscript titled “Estimating the Snow Density using Collocated Parsivel and MRR Measurements: A Preliminary Study from ICE-POP 2017/2018 ”. that was submitted to ACP (Atmospheric Chemistry and Physics) on 3 January, 2024. Based on your suggestions, we have put substantial effort into additional analysis. The manuscript has been thoughtfully revised regarding the comments from all reviewers.
  One of the major concerns of the proposed density retrieval algorithm using collocated MRR and Parsivel is lacking the uncertainty analysis. As per the reviewer’s suggestion, we have performed substantial investigations of the retrieval uncertainty. The impacts of the measurement uncertainty of the Parsivel and the MRR on the bulk density retrieval are analyzed quantitatively. The measurement issue of Parsivel is also investigated to understand its impact on bulk density retrieval. The results are summarized in the revised manuscript as a Discussion section.
  The MRR data quality issue has been examined per the reviewer’s suggestion. The post-processed data have replaced entire MRR raw data by applying the algorithm from Maahn and Kollias (2012). All the bulk density, bulk water fraction, and reflectivity-weighted velocity retrievals have been recalculated. The figures have been revised as well.
  The original purpose of utilizing reflectivity-weighted velocity to filter adequate retrieval is no longer needed and has been removed in the revised manuscript. The quality of the retrieval results have been greatly improved by applying the post-processed MRR data per the reviewer’s suggestion. The low SNR MRR measurement has been removed. The comparison of reflectivity-weighted velocity is mainly used to identify the inadequate retrieval due to the attenuation effect on MRR reflectivity.
  The performance of the retrieved bulk density has been validated by the snowfall rate (SR) from collocated Pluvio measurements and reflectivity-weighted fall velocity (Vz) from MRR. In addition to SR and Vz, the performance of the retrieved bulk density has been compared with the precipitation imaging package (PIP), a video disdrometer (Newman et al., 2009; Pettersen et al., 2020). The PIP was also deployed at the MHS site during ICE-POP 2018 (Tokay et al., 2023). The comparison of retrieved bulk density between the proposed algorithm in this study and PIP has shown good agreement with each other. The high consistency further confirms the performance of the retrieved bulk density. Since there is no direct bulk water fraction measurement for validation, the authors consider the validation of bulk density retrieval to PIP and Pluvio as “indirect” evidence to support the bulk water fraction retrieval.
  The SR and Vz validation analysis shows that the algorithm can adequately retrieve the bulk density and bulk water fraction. The consistency of the retrieved bulk density to collocated PIP confirms the performance of the proposed algorithm in this study. The advantage of the proposed algorithm is that it utilizes collocated Parsivel and MRR, which are commercially available, commonly used, and robust instruments. The Parsivel and MRR can operate unattentively and need little maintenance. Further application of the proposed algorithm helps derive long-term observation data on snow properties. The authors believe the proposed algorithm can provide an alternative choice if sophisticated instruments (e.g., 2DVD, PIP, SVI, MASC) are unavailable.
  The manuscript has also been revised carefully following the reviewer’s suggestions on English wording. The authors would like to express our sincere appreciation for the comments. The point-to-point replies to every comment have been prepared. Please see the following replies. The added or modified sentences in the revised manual are in red for your convenience. We would appreciate any feedback on the revisions.
  
  Citation: https://doi.org/10.5194/egusphere-2023-3147-AC2

Wei-Yu Chang, Yung-Chuan Yang, Chen-Yu Hung, Kwonil Kim, and Gyuwon Lee

Viewed

Total article views: 259 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
201	38	20	259	8	10

HTML: 201
PDF: 38
XML: 20
Total: 259
BibTeX: 8
EndNote: 10

Views and downloads (calculated since 16 Jan 2024)

Month	HTML	PDF	XML	Total
Jan 2024	72	16	4	92
Feb 2024	62	9	3	74
Mar 2024	27	7	3	37
Apr 2024	40	6	10	56

Cumulative views and downloads (calculated since 16 Jan 2024)

Month	HTML	PDF	XML	Total
Jan 2024	72	16	4	92
Feb 2024	62	9	3	74
Mar 2024	27	7	3	37
Apr 2024	40	6	10	56

Viewed (geographical distribution)

Total article views: 254 (including HTML, PDF, and XML) Thereof 254 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 26 Apr 2024

Short summary

The snow density is derived by collocated MRR and Parsivel from the ICE-POP 2017/2018. The PSD from the Parsivel is applied to the T-matrix backscattering simulation and compared with the Z_HH from MRR. The bulk density and bulk water fraction are derived from comparing simulated and calculated Z_HH. The retrieved bulk density is validated by comparing the snowfall rate measurements from Pluvio. The consistency of snowfall rate confirms the proposed algorithm.


Total:	0
HTML:	0
PDF:	0
XML:	0