Single-Particle Polarization Tomography Enables Highly Accurate Phase Identification of Cloud Particles

Chen, Yang; Xu, Xiang; Xiao, Ruishan; Yao, Fangzhen; Shi, Xiaojing; Wang, Jiaxin; Wang, Jun; Liang, Gu; Yue, Zhiguo

doi:10.5194/egusphere-2026-3702

Preprints

https://doi.org/10.5194/egusphere-2026-3702

Preprints

14 Aug 2026

| 14 Aug 2026

Status: this preprint is open for discussion and under review for Atmospheric Measurement Techniques (AMT).

Single-Particle Polarization Tomography Enables Highly Accurate Phase Identification of Cloud Particles

Yang Chen, Xiang Xu, Ruishan Xiao, Fangzhen Yao, Xiaojing Shi, Jiaxin Wang, Jun Wang, Gu Liang, and Zhiguo Yue

Abstract. The phase state of cloud particles and their transformation processes are fundamental to understanding cloud and precipitation formation, variations in radiative energy budgets, and the evolution of severe weather systems. At present, for small-scale cloud particles smaller than 50 μm, conventional identification methods based on morphological features are readily constrained by imaging resolution and feature overlap, making high-accuracy phase discrimination challenging. To address this issue, we propose a polarization tomographic imaging method that integrates digital holography and polarimetric imaging, enabling the simultaneous acquisition of morphological parameters of individual particles within a particle ensemble, including three-dimensional coordinates, particle size, area, perimeter, the major and minor axes of the minimum-area bounding rectangle, and circularity, together with polarization parameters. Through observation experiments in an ice cloud chamber, liquid droplet and ice crystal samples smaller than 50 μm were analyzed. By combining morphological and polarization parameters, accurate identification of liquid droplets and ice crystals in mixed-phase particles was achieved, with an accuracy of 98.33 %, representing an improvement of 4.16 percentage points over the conventional circularity-based identification method. In addition, a continuous 10 s observation window during the late stage of mixed-phase cloud glaciation in the ice cloud chamber experiment was selected to calculate liquid water content, the number concentrations of liquid droplets and ice crystals, and their particle-number fractions. The results show that the droplet number concentration and liquid water content decrease with time, whereas the relative fraction of ice crystals gradually increases, reflecting the microphysical process of liquid water conversion into ice in a mixed-phase environment. This method can not only effectively improve the accuracy of phase identification for small-scale cloud particles but also provide information on the microphysical evolution of cloud particles, offering important support for studies of cloud microphysical processes, improvement of model parameterizations, weather modification, and early warning of severe weather events.

Received: 24 Jun 2026 – Discussion started: 14 Aug 2026

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Yang Chen, Xiang Xu, Ruishan Xiao, Fangzhen Yao, Xiaojing Shi, Jiaxin Wang, Jun Wang, Gu Liang, and Zhiguo Yue

Status: open (until 20 Oct 2026)

Post a comment Subscribe to comment alert

RC1:
'Comment on egusphere-2026-3702', Anonymous Referee #1, 08 Sep 2026 reply
Mixed phase clouds are important constituents of the atmosphere, for instance with respect to radiation effects and precipitation generation. Their microphysical characterization, which is crucial for modelling and forecasting, is nevertheless difficult, mostly due to the different phases of hydrometeors in the cloud. Since the interplay between particles of different phases of water, i.e. water droplets and ice particles, governs the most interesting microphysical processes, distinguishing between liquid and solid when measuring the particles is crucial. In the manuscript, Chen et al. present a digital holography method integrating polarization tomographic imaging that is applied for phase identification of small particles in a mixed phase cloud environment. The method is very promising, and the topic doubtlessly fits within the scope of Atmospheric Measurement Techniques.
The paper is in general well written, easy to follow but contains some repetitions. The figures are of good quality. I list some points that should be considered before publication:
Major points
If I understand correctly, the authors introduce their method and an instrument based on that method. As an example of applications, they show some results of measurements in a cloud chamber. First, the experiments for instrument characterization and testing should be described in more detail and in a distinguishable manner. I recommend moving the description of the cold chamber to the main part of the manuscript, to the Experimental System section. Second, it is mentioned that glass beads were used for size and polarimetric characterization. Do they work for water and ice? If yes, why? If not, which correction should have been applied? Third, concentration, size, LWC, etc. have been measured, but these were not compared to other measurement techniques. It would be important to present such an intercomparison; however, I understand that the authors might only want to show one application for which their method/instrument can be employed. If this latter is indeed the case, it should be emphasized in the paper.
Minor points
I recommend a thorough reread of the paper in order to get rid of repetitions. (I will mention some of them below.)

For instance, lines 114 and 117. And regarding the Mueller matrix in lines 115 and 151.

Again: lines 154 and 156.

The title of section 2.1 is about birefringent crystals; however, at its end, liquid droplets are considered. This section should be organized somewhat better.

Section 2.2: Some symbols, like the area S and the magnification M, have symbols very similar to those used in section 2.1.

Line 190: "This study mainly focuses on…" – I think the focus of the study should be defined earlier. Furthermore, I think the focus lies not really on the concentration of the particles and on the LWC, but rather on the method itself.

20: For the LWC, the projected area equivalent diameter is used. This will definitely introduce an error into the particle's mass, i.e., into the LWC. Please comment on this in the manuscript.

Line 220: Is the imaging lens the same as the microscope lens in Fig. 1?

Line 225: This is not obvious from Fig. 1d or from the description of how the polarization orientations are handled for the acquisition.

Line 229: How did you conduct the experiments? Inside the chamber, or did you use a sample line? Please clarify.

Line 239: Have the images been reconstructed only for particles? Was the 1 mm interval sufficient, or was this interval only used to further constrain the particle position?

Line 269: How did you obtain the nominal median diameter of the droplets?

Line 275: "preset chamber conditions" – this is a very loose formulation. Please be more precise.

Figure 3: I recommend that the authors indicate somehow (for example, with arrows) where the expected particle sizes would lie. Furthermore, are the "standard particles" in the caption the glass microspheres?

Line 287: Circularity is an important parameter used by the authors. Please provide a definition for it.

Figure 4 shows the circularity of droplets and ice crystals. How do you know which particles you are measuring? Was it ensured that exclusively liquid droplets or ice particles were present?

Lines 306 and 315: repetition.

Line 321: In my view, the overlap is large, not only "partial."

Lines 323–325: This has already been mentioned earlier.

Figure 8: What are the green symbols showing?

The text from line 375 to line 387 should be rewritten to get rid of repetitions.

Figure 9: I recommend presenting the measured particle size(s). That would nicely show how the Bergeron-Wegener-Findeisen process works.

Reply
Citation: https://doi.org/10.5194/egusphere-2026-3702-RC1
- AC1: 'Reply on RC1', Xiang Xu, 15 Sep 2026 reply
  
  Please refer to the PDF for the response to the review comments.
  
  Reply
  
  Citation: https://doi.org/10.5194/egusphere-2026-3702-AC1
- AC2: 'Reply on RC1', Xiang Xu, 19 Sep 2026 reply
  
  Dear Referee,
  We apologize that the previously submitted response file contained a few minor editorial errors. We have now carefully checked the file and corrected the minor errors. A corrected version of the response has therefore been resubmitted.
  These corrections are purely editorial and do not affect the scientific content of the manuscript or our responses to the reviewer’s comments. We sincerely apologize for any inconvenience and thank you for your understanding.
  Yours sincerely,
  
  The Authors
  
  Reply
  
  Citation: https://doi.org/10.5194/egusphere-2026-3702-AC2
RC2: 'Comment on egusphere-2026-3702', Anonymous Referee #2, 07 Oct 2026 reply

Review of 'Single-Particle Polarization Tomography Enables Highly Accurate Phase Identification of Cloud Particles' by Yang Chen et al.
The manuscript introduces a polarization tomographic imaging method to identify the phase of cloud chamber generated clouds. The method combines digital holography with polarimetric imaging. For cloud particles with sizes below 50 µm the cloud particle circularity and degree of linear polarization are used for phase identification. A linear support vector machine is used to distinguish liquid from frozen particles based on the two parameters. As an example of the methods a 10 s time window of a glaciation evolution of a cloud chamber generated mixed-phase cloud is presented where the microphysical properties of the cloud particles are derived with the tomographic imaging method.
While I request a few adaptations to the manuscript (see comments below), the method seems promising for the characterization of mixed-phase clouds and is in the scope of AMT. Therefore, I recommend the manuscript to be published after major revisions.
General comments:
-Please elaborate what the reference is for validating liquid and frozen particles. Maybe I missed this in the manuscript. But how did you know the real phase of the particles where the calculation of accuracy, precision, recall, etc. is based on.
-The discussion chapter contains the description of the machine learning linear support vector machine as well as the results. While accuracy, precision, recall should be discussed in the discussion section, I recommend to move the machine learning method in the methods section and the results in the results section.
-Furthermore, the results of the glaciation experiment in the cloud chamber are also presented in the discussion section. Why are they not in the results section? The discussion section should be about the interpretation of the results not presenting additional results.
-Please discuss the measurement uncertainty for the particle size. You performed measurements with glass beads, but no final conclusion is taken here.
-Introduction: There are studies on in-situ single particle phase discrimination based on the light scattering properties e.g. Waitz et al. (2021), Kaye et al. (2008) or Ulanowski et al. (2014). Please also refer to these.
-Please provide more information on the cloud chamber characteristics: Volume, cleaning procedures before experiments, experiments preparation, how is the supersaturation controlled? Which instruments are used to monitor temperature, humidity, aerosol population?
-Is there a limitation in maximum particle concentration of the instrument?
-Please add missing references for the used definitions. E.g. for accuracy, precision, recall, dolp the references are missing.
Specific comments:
L56-59: please refer to the original literature in distinguishing liquid from ice phase using polarization lidar (e.g. Liou and Lahore, 1974 etc.). This is not a recent finding from 5 years ago.
L89: Please state measurement uncertainty for mean particle sizes here. How is the size of the ice particles calculated? Is it the maximum dimension?
L95 and L375-L378: What is the reference for the validation? How do you know the real particle phase? Is it before and after the addition of INPs to the cloud chamber? How do you know all particle nucleated?
Also 98.33% suggest more robust statistics than 60/61 samples. I would suggest to say 98% instead keeping the number of significant digits. Similar for accuracy, precision, recall.
L96: You need to cite the accuracy of conventional methods here when stating that your results are better.
L105-107: Sphericity is needed as additional requirement. Non-spherical, e.g. larger droplets do change the polarization of the incident light.
L223-225: Please provide details on the DoFP polarization camera (polarization measurement uncertainty, resolution, name, manufacturer, etc.).
L322: Why 'largely' preserving and not 'preserving'. What prevents the droplets from fully preserving the polarization of the incident light?
L409: What stands t = 0s for. Was this the injection of AgI? Why is a time frame of 10 s chosen?
References:
Kaye, P. H., Hirst, E., Greenaway, R. S., Ulanowski, Z., Hesse, E., DeMott, P. J., ... & Connolly, P. (2008). Classifying atmospheric ice crystals by spatial light scattering. Optics letters, 33(13), 1545-1547.
Ulanowski, Z., Kaye, P. H., Hirst, E., Greenaway, R. S., Cotton, R. J., Hesse, E., & Collier, C. T. (2014). Incidence of rough and irregular atmospheric ice particles from Small Ice Detector 3 measurements. Atmospheric Chemistry and Physics, 14(3), 1649-1662.
Waitz, F., Schnaiter, M., Leisner, T., & Järvinen, E. (2021). PHIPS-HALO: the airborne Particle Habit Imaging and Polar Scattering probe–Part 3: Single-particle phase discrimination and particle size distribution based on the angular-scattering function. Atmospheric Measurement Techniques, 14(4), 3049-3070.

Reply

Citation: https://doi.org/10.5194/egusphere-2026-3702-RC2

Yang Chen, Xiang Xu, Ruishan Xiao, Fangzhen Yao, Xiaojing Shi, Jiaxin Wang, Jun Wang, Gu Liang, and Zhiguo Yue

Viewed

Total article views: 294 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
173	82	39	294	32	37

HTML: 173
PDF: 82
XML: 39
Total: 294
BibTeX: 32
EndNote: 37

Views and downloads (calculated since 14 Aug 2026)

Month	HTML	PDF	XML	Total
Aug 2026	39	8	6	53
Sep 2026	86	55	27	168
Oct 2026	48	19	6	73

Cumulative views and downloads (calculated since 14 Aug 2026)

Month	HTML	PDF	XML	Total
Aug 2026	39	8	6	53
Sep 2026	86	55	27	168
Oct 2026	48	19	6	73

Viewed (geographical distribution)

Total article views: 276 (including HTML, PDF, and XML) Thereof 276 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 17 Oct 2026

Short summary

Accurate phase identification of small cloud particles is essential for understanding mixed-phase cloud processes, yet it remains particularly challenging for particles smaller than 50 µm. To address this challenge, we developed a single-particle polarization tomography method that combines morphological and polarization information. The method achieved 98.33 % accuracy and provided information on the microphysical evolution of cloud particles.


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