Continuous Microstructural Characterization of a shallow East Antarctic Ice Core Using Machine Learning Super-Resolution and Micro-CT Imaging

Bagherzadeh, Faramarz; Freitag, Johannes; Frese, Udo; Wilhelms, Frank

doi:10.5194/egusphere-2026-3185

Preprints

https://doi.org/10.5194/egusphere-2026-3185

Preprints

25 Jun 2026

| 25 Jun 2026

Continuous Microstructural Characterization of a shallow East Antarctic Ice Core Using Machine Learning Super-Resolution and Micro-CT Imaging

Faramarz Bagherzadeh, Johannes Freitag, Udo Frese, and Frank Wilhelms

Abstract. This study presents a continuous, high-resolution (60 μm) microstructural analysis of a 130 m long ice core B40 drilled in Austral summer 2012/13 at the German Research station Kohnen on the East Antarctic plateau using artificial intelligence (AI)-enhanced micro-computed tomography (micro-CT) imaging and pore network modeling. Antarctic ice cores serve as valuable archives of past climate and environmental conditions. They consist of a sequence of compacted layers with varying density, ice and pore space structure, which are shaped differently depending on the climatic and environmental conditions during deposition and subsequent compaction. The in-depth development of microstructural features provides insights into the densification and pore closure processes as well as climate trends during the formation of the firn column.

However, traditional micro-CT imaging techniques applied to full core samples, although effective, often lack the necessary resolution to capture the intricate microstructure of ice samples fully. To address this limitation, we applied AI-driven super-resolution enhancement methods to improve the clarity and detail of micro-CT images, enabling a more precise quantification of ice core properties. Following AI enhancement, the study performed a comprehensive microstructure analysis on the full firn column, including geometrical, morphological, topological, and transport-related properties. The obtained metrics, including mean intercept length (MIL), cluster size, porosity (density), tortuosity, permeability, etc., can characterize the microstructural evolution of the ice column from snow to bubbly ice across different depths. The results reveal a systematic evolution in pore geometry, connectivity, and transport efficiency with depth, capturing the snow-to-ice transition with high fidelity. Fine and coarse layers were distinguished using the K-means clustering method, and anisotropy was detected in both the ice matrix and the pore space. Principal component analysis (PCA) showed that densification is influenced by multiple factors; in particular, during the pore close-off stage, variations in coordination number and throat dimensions led to distinct densification behaviors among the samples. These findings could contribute to the improvement of densification and air transport models by providing highly accurate data on the microstructure of ice cores and even enabling the definition of new microstructure-related climate proxy parameters.

Received: 04 Jun 2026 – Discussion started: 25 Jun 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

Faramarz Bagherzadeh, Johannes Freitag, Udo Frese, and Frank Wilhelms

Status: final response (author comments only)

RC1:
'Comment on egusphere-2026-3185', Anonymous Referee #1, 03 Aug 2026
Bagherzadeh et al. use microCT to conduct a microstructural analysis along a continuous 130 m-long ice core that captures the snow to ice transition in a region of the East Antarctica Plateau. Collecting three-dimensional and such large-volume data at high-resolution requires extensive scan time, which the study reduces by collecting low-resolution 3D data (in this case 120 um) and then applying a previously developed deep learning super-resolution method to increase 3D image resolution down to 60 um. In this study, the authors add to their method by applying additional steps in their data processing. The approach taken and the achievement of such a long continuous ice core analysis has potential to advance the documentation and characterization of ice core microstructures, as well as support interpretations of paleoclimate proxies contained within the ice. There is potential for contribution to our understanding of firn densification, pore close-off and gas-transport behavior.
General comments:
There is great potential for the authors’ approach and study to advance capabilities in reconstructing ice core records and for understanding the physical evolution of ice sheets as they transition from snow to ice. The manuscript will require much improvement primarily regarding writing structure, presentation (grammar, formatting), and graphical communication before it could be considered for publication. All grammatical and basic formatting errors should be corrected before a resubmission.
The explanation of the DLSR method could be improved, as many readers may not be well versed in AI methods at this point in time. It is right to reference your previous work, but readers would benefit from additional information that expresses the main ideas around your method. I suggest adding to Figure 1 a graphic that demonstrates physically/visually what is happening during the main steps in your method, particularly the “new” steps you present in this study. It can be difficult to imagine what is occurring in certain steps of the computational workflow. Can you put this into a graphical representation? (i.e. more than the flowchart with text).
As stated by the authors, a number of results from the microCT dataset are consistent with current views of the firn column and transition of snow to ice. However, they fail in many cases to reference the previous studies that have established those views. The advance in this study was not so obvious to me when reading the manuscript. Make clear what the new information gained is in the discussion/conclusion. Perhaps this is why it seems its impact is overstated but also consider whether you have overstated your claims.
There is terminology used that is misleading: “grain clusters” and “solid ice clusters” which are used synonymously. Please see a detailed comment below. If the authors are making a true inference about clusters of ice grains based on absorption contrast tomography (i.e. standard microCT, which cannot differentiate between grains), this would be a critical flaw in the analytical approach. I suggest clarifying your terminology so that it is consistent with what you are truly measuring. Otherwise, such issues propagate in the literature.
Specific comments (see additional specific comments and technical corrections in the marked-up PDF that is attached):
I attach a copy of the manuscript with mark-ups/comments. The marked copy represents the second version submitted by the authors after the editor’s initial review. The authors need to seriously proofread their manuscript to correct all citation and grammatical errors. Many errors seem to be inconsistencies perhaps made by a program. I have made a start at pointing out the errors, but this is not a reviewer’s duty and my suggestions are not exhaustive. In-text citations, for example, need to be formatted correctly with parentheses that separate them from the text. Another example is words that are capitalized when they shouldn’t be, or words that are not in the correct tense, as well as incorrect symbols and a lack of space between words in many instances. The authors should not resubmit this manuscript until it is proofread. In their first rebuttal letter to the editor, the authors state that grammatical corrections were made, which is false.

The structure of the abstract requires improvement. It should present a logical flow of background information, problems, objectives, results, implications etc. Make sure to emphasize the significance of your approach and not just the results (which seem overstated based on information contained in the abstract, or perhaps how it is phrased or structured).

Use a full acronym for “deep learning super-resolution”, like DLSR, similar to PNM and micro-CT, and use it from the beginning of the manuscript.

Please use abbreviations consistently throughout the manuscript (e.g. SR or DLSR). Also, put the abbreviation at the very first use of the phrase in the manuscript, and then you don’t need to write out the phrase again.

Paragraph 3 of the Introduction (starting line 35) does not read well. It reads as sentences and ideas that do not flow together. I suggest the authors connect the sentences better, so that the reader can understand the progression of information towards the ice microstructure or an argument being made.

Lines 67-98 of the introduction do not read smoothly or connect information or ideas well. Sentences generally read as independent pieces of information the reader cannot easily link together. Only provide the most relevant background information, but it needs to flow and progress towards an idea that supports reasoning for the author’s work or arguments.

The authors should clarify that it is not just image resolution that is a limitation with standard micro-CT approaches. Perhaps a much greater issue is that ice grains cannot be detected with "standard micro-CT" (i.e. absorption contrast), yet ice grains are a fundamental component of the ice core microstructure and make up the majority of the microstructure by volume at depth. There is a new micro-CT method that can detect ice grains (see Barbee et al 2026). Could a method like this be integrated in the type of work done here in the future? If not, what about EBSD or standard AFA?

The use of phrases like “grain clusters” and “solid ice clusters” need to be clarified or referenced. This terminology implies you can detect grains that are in clusters, which is not the case. Please clarify what you mean by this with a definition early in the manuscript, or please change it to an accurate terminology. I assume the “spherical ice cluster” metric is how the authors deal with the issue of not being able to visualize or measure the ice grains in 3D (defining their presence and shape). If that is the case, please state it so it is clear, and please consider using a different term other than "grain clusters and solid ice clusters". The terms are inaccurate and misleading. When choosing terminology, consider that bubbles can occur between or inside the ice grains. Would “high-density regions” or “regions of ice concentration” or similar work? Something that implies reduced pores.

It would be great to see 3D imagery over 130 m of ice core, or at least representative sections or small areas of imagery over the depth of the core. Given this is a significant advance made by this work, it would be interesting to see the stratigraphy in 3D. This would give readers the ability to connect the quantitative data with visuals.

The figure captions in general could be expanded and details elaborated on to help the reader understand the graphs.

Some data plots do not have legends for the reader to understand what the colored data points represent. Please add or extend your legends.

In some areas of the manuscript, there is a lack of citations after background information or when results are discussed compared with earlier studies or views. Please add the appropriate citations.
Citation: https://doi.org/10.5194/egusphere-2026-3185-RC1
- AC1: 'Reply on RC1', Faramarz Bagherzadeh, 04 Sep 2026
  
  We thank the reviewer for the careful and constructive comments and for recognizing the potential of our approach for continuous three-dimensional characterization of firn and ice-core microstructure. We have substantially revised the manuscript to improve its structure, clarity, terminology, graphical presentation, and language. We have also carefully proofread the entire manuscript and corrected grammatical, typographical, citation, formatting, abbreviation, symbol, and spacing errors.
  We agree that the DLSR method required a clearer explanation for readers who may not be familiar with deep-learning approaches. We have expanded the description of the method and revised Fig. 1 to provide a more intuitive visual representation of the main processing steps, including the additional steps introduced in this study. We now introduce and consistently use the abbreviation “deep learning super-resolution (DLSR)” at its first occurrence and have standardized the use of abbreviations throughout the manuscript.
  We appreciate the reviewer’s comment regarding the distinction between established knowledge and the new information provided by this study. We have revised the Introduction to better identify the specific contribution of our work and to add appropriate references to previous studies. In particular, we clarify that the main advance is the continuous, three-dimensional, multi-scale quantification of microstructural properties over approximately 130 m across.
  We agree that our original use of the terms “grain clusters” and “solid ice clusters” could incorrectly imply that individual ice grains or grain boundaries are resolved by absorption-contrast micro-CT. This was not our intention.
  We thank the reviewer for this important comment and agree that the terms “grain clusters” and “solid ice clusters” may be misleading, as the present measurements do not allow us to directly identify or characterize individual ice grains or their three-dimensional clustering. To avoid ambiguity and to more accurately describe the measured quantities, we will replace these terms throughout the manuscript with “local ice thickness” and “local pore thickness,” respectively. These terms more directly reflect the spatial variations in ice and pore regions that can be quantified from our measurements, without implying that individual ice grains or their 3D morphology can be resolved. We will also revise the terminology and relevant descriptions in the manuscript to make this distinction clear.
  We have substantially rewritten and reorganized the Introduction, particularly the paragraphs identified by the reviewer, to improve the logical progression from firn densification and microstructural evolution to the motivation and objectives of this study. Background information has been condensed where appropriate, and additional citations have been added where established findings or comparisons with previous studies are discussed.
  We agree that representative three-dimensional imagery would help readers connect the quantitative results with the underlying microstructure. We have therefore added a new figure showing representative 3D volumes from different depths across the firn-to-ice transition. We have also expanded the figure captions and legends throughout the manuscript to explain the plotted variables, colors, symbols, and other graphical elements more clearly.
  Finally, we have carefully checked the entire manuscript again for the issues identified by the reviewer, including citation formatting, grammar, capitalization, tense, symbols, spacing, units, and consistency of terminology. We apologize that some of these issues remained in the previous version despite our earlier corrections and appreciate the reviewer’s detailed assessment, which helped us identify and address them more comprehensively.
  
  Citation: https://doi.org/10.5194/egusphere-2026-3185-AC1
RC2:
'Comment on egusphere-2026-3185', Martin Schneebeli, 24 Aug 2026

This manuscript presents a significant advance in firn-core imaging and interpretation. The authors show on a single core that their multi-scale analysis of a CT-scanned firn core leads to a much improved understanding of snow and firn densification. The manuscript demonstrates the power of the construction of a digital twin and machine-learning in ice core analysis.
An aspect that is not considered and could be included is the observed spatial variability between firn cores. Such local (at the scale of 10s of meters) could explain the observed differences in lock-in depths.
Technical correction:
The citations use wrong or incomplete parentheses. Units are typeset in italic, but should be typeset in roman font (and a space between number and unit).

Citation: https://doi.org/10.5194/egusphere-2026-3185-RC2
- AC2: 'Reply on RC2', Faramarz Bagherzadeh, 04 Sep 2026
  
  We thank the reviewer for the positive assessment of our manuscript and for highlighting the potential of the multi-scale analysis, digital twin, and machine-learning approach for improving the understanding of snow and firn densification.
  
  We agree with the reviewer that spatial variability at the scale of tens of meters is an important aspect that may contribute to differences in firn structure and lock-in depth. Our analysis is based on a single firn core and therefore cannot directly quantify the spatial variability between cores. However, the high-resolution, multi-scale approach presented here provides a framework that could be used to investigate such variability in future studies by applying the same analysis to multiple nearby firn cores.
  We have therefore added a discussion of this point to the revised manuscript. In particular, we now acknowledge that local variations in snow accumulation, wind redistribution, surface conditions, and subsequent metamorphism may lead to differences in the densification history and pore-space structure over relatively short spatial distances. These variations could potentially contribute to the observed differences in lock-in depth. We emphasize that distinguishing these spatial effects from temporal variability and other controls on firn densification will require comparable high-resolution analyses of multiple cores from the same region.
  
  About citation issues.Thank you for pointing this out. We have carefully checked the manuscript and corrected the citation parentheses throughout the text. We have also corrected the formatting of units so that they are typeset in roman font with an appropriate space between numerical values and units.
  
  Citation: https://doi.org/10.5194/egusphere-2026-3185-AC2

Faramarz Bagherzadeh, Johannes Freitag, Udo Frese, and Frank Wilhelms

Viewed

Total article views: 195 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
109	68	18	195	12	5

HTML: 109
PDF: 68
XML: 18
Total: 195
BibTeX: 12
EndNote: 5

Views and downloads (calculated since 25 Jun 2026)

Month	HTML	PDF	XML	Total
Jul 2026	41	21	7	69
Aug 2026	41	23	3	67
Sep 2026	27	24	8	59

Cumulative views and downloads (calculated since 25 Jun 2026)

Month	HTML	PDF	XML	Total
Jul 2026	41	21	7	69
Aug 2026	41	23	3	67
Sep 2026	27	24	8	59

Viewed (geographical distribution)

Total article views: 161 (including HTML, PDF, and XML) Thereof 161 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 17 Sep 2026

Short summary

We studied a 130 m ice core from East Antarctica to better understand how snow slowly turns into ice and traps air over time. Using advanced image enhancement and 3D imaging, we examined the ice in unprecedented detail along the entire core. We found clear changes in the size, shape, and connections of air spaces with depth. These results provide new information that can improve interpretations of past climate and help refine models of ice formation and air movement in polar ice.


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