the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 20 Oct 2025
Experimental Investigation of the Direct Shear Strength Parameters of Compacted Snow
Abstract. Compacted snow is utilized as a building material in various construction and engineering applications across global high-latitude regions. For the safety assessment of snow and ice structures in cold regions, cohesion and internal friction angle are key shear strength parameters for compacted snow. This study investigates 69 test conditions, considering variations in initial density, sintering time, and sintering temperature. Using direct shear tests, the variation patterns of the shear strength of compacted (machine-made) snow under normal pressures below 100 kPa were analyzed. Results show that under high sintering degree conditions and low normal pressures, the shear stress–displacement curve tends to exhibit strain softening. As initial density increases from 450 to 650 kg·m-3;, both cohesion and internal friction angle increase linearly. With sintering time increasing from 0 to 60 days, cohesion first rises and then falls, while the internal friction angle steadily decreases. As sintering temperature decreases from -5 to -25 °C, cohesion decreases, whereas the internal friction angle increases slightly. A Genetic Algorithm-Back Propagation (GA-BP) neural network was employed to develop a predictive model for shear strength parameters, providing benchmark values for cohesion and internal friction angle under various conditions. These benchmarks can be adaptively adjusted when additional influencing factors require consideration. This study provides essential strength parameters for the design and construction of compacted snow structures and offers a framework for accounting for the influence of other factors on these parameters.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(3469 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(3469 KB) - Metadata XML
- BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-4768', Anonymous Referee #1, 24 Nov 2025
- AC1: 'Reply on RC1', Hui Xu, 08 Apr 2026
-
RC2: 'Comment on egusphere-2025-4768', Anonymous Referee #2, 12 Mar 2026
Summary:
This manuscript presents methods, datasets, and results from a laboratory experiment related to the shear strength of compressed snow cylinders, where snow samples were obtained from an artificially created snowpack. The findings of this study could have significance for cold regions structural engineering applications, where infrastructure is built from compressed snow, and therefore stability of said infrastructure is highly dependent upon shear and normal strength of various snowpack types. Methods from the study include both an empirical approach involving laboratory testing, as well as a modeling approach, using a neural prediction network. Results of the two approaches include an observational dataset, as well as simulated model output, respectively.
The model was calibrated and validated via partitioning of the empirically derived data. Four parameters were varied and tested in different combinations, using an applied horizontal shear force held constant for each variation of the other parameters. The varied parameters included initial compaction density, applied normal force, snow crystal sintering time, and sintering temperature. Internal friction angle and cohesion of the various samples were calculated using a previously established algorithm from the literature.
The approach and results of this study are interesting and a valuable contribution to snow physics and mechanics, although the overall novelty of this experiment is limited. If the methods presented here can be replicated in future experiments, results from multiple studies could eventually provide a statistically significant body of evidence for establishing numerical thresholds for these snow shear strength parameters. Such thresholds would be valuable for practical use in snow engineering.
Overall Comments:
Shear Stress vs Shear Strength: The authors use these words somewhat interchangeably; however, these are not really the same thing. Conceptually related, but distinct, shear stress is the internal force per unit area acting within a material due to external loads, while strength is the maximum stress said material can withstand before failure (plastic deformation). For this study, stress is the demand placed on each snow sample, while strength is the capacity of each sample. A material fails when the applied stress equals or exceeds its strength.
This is commonly a grey area within the field of snow mechanics, because shear stress is a quantifiable variable with established units of measurement, while shear strength is more nebulous. Previous studies offer approaches for defining snow strength, but it is overall still more of a concept than it is a specific quantity. While sometimes perplexing, defining snow strength is still an open field of research with much opportunity for creativity.
Generally, I understand and agree with the authors’ good approach of defining shear strength as the yield stress of the snow samples, i.e. the quantity of applied shear force at which plastic deformation of the snow begins. However, this needs to be stated more clearly near the beginning of the paper. Refer to the line-by-line comments below, where I’ve identified sections of the paper that need this clarification.
Abstract: Add a little more context (a couple sentences here) and briefly describe the conditions of the experiment. It would be easier to understand the summary of results and what the paper is about, with a simple description of the experiment set up, i.e., “experiments were conducted in a cold lab with environmental controls, using an instrument that does….” Which parameters were measured vs. calculated vs. modeled? Specify which parameters are the “essential strength parameters”.
Introduction: I don’t fully agree with the first part of the last paragraph in this section. I’m not sure that most previous studies within the snow engineering field of research are, in fact, focused on low density snow. I’d say the opposite is true. Most engineering related snow research actually does focus on compacted/altered snow, while avalanche research typically explores unaltered/natural, in-situ snowpacks. Either way, citations of studies on low density snow strength for snow engineering purposes should be provided. Or consider removing this section.
The novelty of this paper is not that it explores compacted snow. This study still has value in terms of contributing quantifiable results that could support (or not support) conclusions made by related studies. The results of this study could also be used in the future for establishing thresholds for certain snow parameters for engineering applications. The value is more practical than novel, which is fine.
Neural network and prediction modeling: The agreement between the measured data and the modeled data from the neural network looks a little too good. The level of validation is sky high. I’m wondering if the model is overfit, considering 80% of the empirical dataset was used for calibration. Also, I’m not a neural network expert by any means, but isn’t this a ton of computation power for predicting only four variables? I still think having a prediction model for these parameters is very valuable, but this looks like over-fitting to me.
Discussion Section: This section needs a lot of work and major revision and improvements. It seems like the authors ran out of steam by the time they got to this part of the manuscript. Discussions are often the most difficult section to craft, yet the most important part of most papers. Organizing this section by snow parameter seems like a good way to start, but there are actually zero conclusions drawn from the results of the study. Each parameter listed in this section is accompanied only by some literature citations.
The authors need to clearly communicate conclusions drawn from conducting the experiment, supported by the study outcomes. The overall messages that the authors wish for the readers to take away need to be stated, using specific examples from the results section as evidence for conclusions drawn. Refer to the results presented, including the empirical measurements, results from calculations, figures etc., and the modeled output. The conclusion also needs more work and should be tied back into how this study is valuable for snow structural engineering.
Additional, line-by-line comments are provided to the authors and journal editor in an accompanying document provided by this reviewer.
- AC2: 'Reply on RC2', Hui Xu, 08 Apr 2026
Peer review completion
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-4768', Anonymous Referee #1, 24 Nov 2025
Experimental Investigation of the Direct Shear Strength Parameters of Compacted Snow
By Haifeng Huo, Hui Xu, Jixiu Wu, Tao Li, Jingjin Liu, Enzhao Xiao and Xueyuan Tang
Summary
This paper presents an experimental setup to investigate the shear strength parameters of compacted snow. It also proposes a methodology for preparing compacted snow samples with distinct parameters, including initial density, sintering time, and sintering temperature. Using their apparatus, the authors obtained shear stress–displacement curves for samples under different normal loads. From these curves, they derived the shear strength based on the peak shear stress when present, or at a fixed displacement of 4 mm when no peak was observed. They also calculated the internal friction angle and cohesion using a Mohr–Coulomb relationship. Finally, the authors proposed a neural network to predict the shear parameters based on the four tested variables.
The paper is generally well written, and the methodology appears appropriate for the research question. While the novelty of the work is limited, the measurements and dataset produced in this study are valuable and warrant publication. My main concern is the choice of a neural network to predict a relatively simple relationship using only four variables. This choice seems driven more by popularity than by scientific necessity, especially since it is neither justified nor discussed. In addition, the discussion section would benefit from deeper analysis, as it currently reads like a list of bullet points. It may also be useful to add a dedicated section addressing the limitations and biases of the study, and how these may influence the results.
Major Comment:
- The choice of the neural network needs to be explained and discussed. The choice seems to be rather “overkill” for such a simple relation (almost linear) with only 4 input variables. Usually, neural networks are chosen for non-linear behavior with a significant number of input variables. This chosen neural network is then prone to overfitting and migth boost the performance metrics. Also, neural network is better with a significant dataset, more around 200-300 data, rather than only 50. At least address it in the methodology and in the discussion should be necessary.
- Discussion: This section currently reads like a bullet-point list. It should be rewritten to improve the flow and strengthen the connections between paragraphs. There is also a need for an additional paragraph addressing the potential biases of the study. The paragraph on shear rate appears abruptly and feels underdeveloped, with only a few sentences. It should be better introduced and expanded, particularly regarding how shear rate affects the validity of the results, as this is closely related to Figure 4.
Specific comments (line number)
63: Which studies? A few references are required here as it is the base of the novelty of the study.
Specific comments (line numbers)
74–75: Is there any reference for this process?
89: What is the resulting sow grain type, I’m guessing rounding grains or facets but was this observed? What exactly is meant by a natural sintering environment? Does this refer to isolating the sample from the surrounding air in the cold room? If possible, add a photo to Figure 1.
Table 1: Please define σ as the other variables.
Figure 6: Please add the symbol definitions so the figure can be understood on its own.
157: This result is interesting, as lower densities closer to natural compacted snow often exhibit peak strain-softening. Why was 650 kg/m³ used for sintering time tests and 550 kg/m³ for sintering temperature tests?
177: “…and increases again after 250 kg/m³.”
Figure 11: Why is the density plotted with a dashed line while the other curves are solid? You should consider using dashed lines for all curves, since no observations exist outside the measured points and no analytical fit is provided.
Section 4: Why use a neural network? It seems excessive for a four-variable input and a relatively simple interaction between variables.
284: Rephrase the sentence so it is clear that you compared the shear strength with all the other input variables.
337-339: Can you compare with other studies for your “low density”? It is surprising to me that you have no softening peak with lower densities that are closer to natural snow. Maybe discuss the influence of the other parameters on that matter (sintering time chose = 5 days).
Citation: https://doi.org/10.5194/egusphere-2025-4768-RC1 - AC1: 'Reply on RC1', Hui Xu, 08 Apr 2026
-
RC2: 'Comment on egusphere-2025-4768', Anonymous Referee #2, 12 Mar 2026
Summary:
This manuscript presents methods, datasets, and results from a laboratory experiment related to the shear strength of compressed snow cylinders, where snow samples were obtained from an artificially created snowpack. The findings of this study could have significance for cold regions structural engineering applications, where infrastructure is built from compressed snow, and therefore stability of said infrastructure is highly dependent upon shear and normal strength of various snowpack types. Methods from the study include both an empirical approach involving laboratory testing, as well as a modeling approach, using a neural prediction network. Results of the two approaches include an observational dataset, as well as simulated model output, respectively.
The model was calibrated and validated via partitioning of the empirically derived data. Four parameters were varied and tested in different combinations, using an applied horizontal shear force held constant for each variation of the other parameters. The varied parameters included initial compaction density, applied normal force, snow crystal sintering time, and sintering temperature. Internal friction angle and cohesion of the various samples were calculated using a previously established algorithm from the literature.
The approach and results of this study are interesting and a valuable contribution to snow physics and mechanics, although the overall novelty of this experiment is limited. If the methods presented here can be replicated in future experiments, results from multiple studies could eventually provide a statistically significant body of evidence for establishing numerical thresholds for these snow shear strength parameters. Such thresholds would be valuable for practical use in snow engineering.
Overall Comments:
Shear Stress vs Shear Strength: The authors use these words somewhat interchangeably; however, these are not really the same thing. Conceptually related, but distinct, shear stress is the internal force per unit area acting within a material due to external loads, while strength is the maximum stress said material can withstand before failure (plastic deformation). For this study, stress is the demand placed on each snow sample, while strength is the capacity of each sample. A material fails when the applied stress equals or exceeds its strength.
This is commonly a grey area within the field of snow mechanics, because shear stress is a quantifiable variable with established units of measurement, while shear strength is more nebulous. Previous studies offer approaches for defining snow strength, but it is overall still more of a concept than it is a specific quantity. While sometimes perplexing, defining snow strength is still an open field of research with much opportunity for creativity.
Generally, I understand and agree with the authors’ good approach of defining shear strength as the yield stress of the snow samples, i.e. the quantity of applied shear force at which plastic deformation of the snow begins. However, this needs to be stated more clearly near the beginning of the paper. Refer to the line-by-line comments below, where I’ve identified sections of the paper that need this clarification.
Abstract: Add a little more context (a couple sentences here) and briefly describe the conditions of the experiment. It would be easier to understand the summary of results and what the paper is about, with a simple description of the experiment set up, i.e., “experiments were conducted in a cold lab with environmental controls, using an instrument that does….” Which parameters were measured vs. calculated vs. modeled? Specify which parameters are the “essential strength parameters”.
Introduction: I don’t fully agree with the first part of the last paragraph in this section. I’m not sure that most previous studies within the snow engineering field of research are, in fact, focused on low density snow. I’d say the opposite is true. Most engineering related snow research actually does focus on compacted/altered snow, while avalanche research typically explores unaltered/natural, in-situ snowpacks. Either way, citations of studies on low density snow strength for snow engineering purposes should be provided. Or consider removing this section.
The novelty of this paper is not that it explores compacted snow. This study still has value in terms of contributing quantifiable results that could support (or not support) conclusions made by related studies. The results of this study could also be used in the future for establishing thresholds for certain snow parameters for engineering applications. The value is more practical than novel, which is fine.
Neural network and prediction modeling: The agreement between the measured data and the modeled data from the neural network looks a little too good. The level of validation is sky high. I’m wondering if the model is overfit, considering 80% of the empirical dataset was used for calibration. Also, I’m not a neural network expert by any means, but isn’t this a ton of computation power for predicting only four variables? I still think having a prediction model for these parameters is very valuable, but this looks like over-fitting to me.
Discussion Section: This section needs a lot of work and major revision and improvements. It seems like the authors ran out of steam by the time they got to this part of the manuscript. Discussions are often the most difficult section to craft, yet the most important part of most papers. Organizing this section by snow parameter seems like a good way to start, but there are actually zero conclusions drawn from the results of the study. Each parameter listed in this section is accompanied only by some literature citations.
The authors need to clearly communicate conclusions drawn from conducting the experiment, supported by the study outcomes. The overall messages that the authors wish for the readers to take away need to be stated, using specific examples from the results section as evidence for conclusions drawn. Refer to the results presented, including the empirical measurements, results from calculations, figures etc., and the modeled output. The conclusion also needs more work and should be tied back into how this study is valuable for snow structural engineering.
Additional, line-by-line comments are provided to the authors and journal editor in an accompanying document provided by this reviewer.
- AC2: 'Reply on RC2', Hui Xu, 08 Apr 2026
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 1,342 | 2,569 | 185 | 4,096 | 133 | 142 |
- HTML: 1,342
- PDF: 2,569
- XML: 185
- Total: 4,096
- BibTeX: 133
- EndNote: 142
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Haifeng Huo
Hui Xu
Jixiu Wu
Tao Li
Jingjin Liu
Enzhao Xiao
Xueyuan Tang
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(3469 KB) - Metadata XML
Experimental Investigation of the Direct Shear Strength Parameters of Compacted Snow
By Haifeng Huo, Hui Xu, Jixiu Wu, Tao Li, Jingjin Liu, Enzhao Xiao and Xueyuan Tang
Summary
This paper presents an experimental setup to investigate the shear strength parameters of compacted snow. It also proposes a methodology for preparing compacted snow samples with distinct parameters, including initial density, sintering time, and sintering temperature. Using their apparatus, the authors obtained shear stress–displacement curves for samples under different normal loads. From these curves, they derived the shear strength based on the peak shear stress when present, or at a fixed displacement of 4 mm when no peak was observed. They also calculated the internal friction angle and cohesion using a Mohr–Coulomb relationship. Finally, the authors proposed a neural network to predict the shear parameters based on the four tested variables.
The paper is generally well written, and the methodology appears appropriate for the research question. While the novelty of the work is limited, the measurements and dataset produced in this study are valuable and warrant publication. My main concern is the choice of a neural network to predict a relatively simple relationship using only four variables. This choice seems driven more by popularity than by scientific necessity, especially since it is neither justified nor discussed. In addition, the discussion section would benefit from deeper analysis, as it currently reads like a list of bullet points. It may also be useful to add a dedicated section addressing the limitations and biases of the study, and how these may influence the results.
Major Comment:
Specific comments (line number)
63: Which studies? A few references are required here as it is the base of the novelty of the study.
Specific comments (line numbers)
74–75: Is there any reference for this process?
89: What is the resulting sow grain type, I’m guessing rounding grains or facets but was this observed? What exactly is meant by a natural sintering environment? Does this refer to isolating the sample from the surrounding air in the cold room? If possible, add a photo to Figure 1.
Table 1: Please define σ as the other variables.
Figure 6: Please add the symbol definitions so the figure can be understood on its own.
157: This result is interesting, as lower densities closer to natural compacted snow often exhibit peak strain-softening. Why was 650 kg/m³ used for sintering time tests and 550 kg/m³ for sintering temperature tests?
177: “…and increases again after 250 kg/m³.”
Figure 11: Why is the density plotted with a dashed line while the other curves are solid? You should consider using dashed lines for all curves, since no observations exist outside the measured points and no analytical fit is provided.
Section 4: Why use a neural network? It seems excessive for a four-variable input and a relatively simple interaction between variables.
284: Rephrase the sentence so it is clear that you compared the shear strength with all the other input variables.
337-339: Can you compare with other studies for your “low density”? It is surprising to me that you have no softening peak with lower densities that are closer to natural snow. Maybe discuss the influence of the other parameters on that matter (sintering time chose = 5 days).