Spatial snowpack properties in a snow-avalanche release area: An extreme dry-slab avalanche case on Mt. Nodanishoji, Japan, in 2021

Katsuyama, Yuta; Katsushima, Takafumi; Adachi, Satoru; Takeuchi, Yukari

doi:https://doi.org/10.5194/nhess-2023-5

Preprints

https://doi.org/10.5194/nhess-2023-5

Preprints

09 Feb 2023

| 09 Feb 2023

Status: this discussion paper is a preprint. It has been under review for the journal Natural Hazards and Earth System Sciences (NHESS). The manuscript was not accepted for further review after discussion.

Spatial snowpack properties in a snow-avalanche release area: An extreme dry-slab avalanche case on Mt. Nodanishoji, Japan, in 2021

Yuta Katsuyama, Takafumi Katsushima, Satoru Adachi, and Yukari Takeuchi

Abstract. An extreme dry-slab snow-avalanche occurred on 10 Jan. 2021 at Mt. Nodanishoji, Gifu, Japan, during a heavy snowfall. The avalanche ran down approximately 2,800 m and caused damage to trees and infrastructures. Although this avalanche was estimated to be the second largest in Japan, physical snowpack properties and their vertical structure and spatial distribution, that caused the avalanche, were not addressed in the release area just after the avalanche fall, mainly due to unsafe and lousy weather. Based on a snow depth distribution observed by an unmanned aerial vehicle and a numerical snowpack simulation in the avalanche release area, the spatial distributions of the mechanical snowpack stability and slab mass and their temporal evolutions were estimated in this study. The procedure was validated by comparing the calculation results with the observed snowpit and spatial snow depth data. The results indicated that two heavy snowfall events, ~3 and 10 days before the avalanche onset, generated two different weak layers made of precipitation particles and associated slabs above other weak layers. The older weak layer was only generated on the northward slope due to its low temperature, whereas the newer layer was predominant over the avalanche release area. The fraction of contributions of the slabs associated with the two weak layers to the total slab mass over the calculation domain was found to be 1 : 2.

Received: 21 Jan 2023 – Discussion started: 09 Feb 2023

Download & links

Yuta Katsuyama, Takafumi Katsushima, Satoru Adachi, and Yukari Takeuchi

Status: closed

RC1:
'Comment on nhess-2023-5', Anonymous Referee #1, 28 Feb 2023

General comments
Thank you for the opportunity to review this work investigating snowpack conditions relating to a large slab avalanche event on Mt. Nodanishoji, Japan in January 2021.
In this work, the authors attempt to characterize conditions snowpack conditions in the release area of the avalanche using UAV-derived snow depth data from a field campaign in March 2022 and snowpack information derived from SNOWPACK model results forced with an operational atmospheric model with 5 km spatial resolution. The authors interpret their results as indicating two heavy snowfall events in the two weeks preceding the avalanche resulted in two weak layers of precipitation particles which contributed to the avalanche’s release. I found the paper generally well-structured and understandable, but it may benefit from some language editing.
Case-study documentation of large snow avalanches plays an important role in better understanding the processes behind these events. In particular, work such as the present study which combines a variety of field and model-based methods can help address scientific questions related to the spatial variability of snowpack conditions leading to large avalanches and falls well within the scope of NHESS. The authors’ premise to combine UAV-supported snow depth mapping and snowpack modeling to better characterize a high-impact avalanche event can be of interest to a broad audience of scientists and practitioners reading NHESS.
My main concern with this work stems from the lack of field data presented from the event, or even the season of the event, itself. Although the authors have made an effort to justify their methods by, for example, showing the similarity in wind conditions (Figure 11) between the snow season in which the avalanche occurred and the snow season in which they conducted field work, the assumption that fully modeled snow distribution and snowpack conditions (e.g. Section 4.4) can robustly represent a specific avalanche event in the absence of any field validation data from the event itself seems rather difficult to justify in this case. Without the inclusion of additional data which can help validate the authors’ model outputs, I am unsure if this work is publishable as an event case-study or as a methodological work. I do not have an adequate mental image of this avalanche after reading the paper and cannot even identify on the provided figures (e.g. Figures 4, 8, and 9) where in the study domain the avalanche occurred.
The authors have clearly put a great deal of effort into this paper and present a compelling methodological premise for event documentation. I wonder if photos of the event / event’s aftermath (i.e. such as those in the referenced Japan Avalanche Network’s event report) could serve as first-order, qualitative validation data? Do any snow profiles from the 2020/2021 season exist in the area in which the avalanche occurred? Such data could, along with a much more comprehensive discussion of model and result uncertainties help bring this work to a publishable level.
Specific comments
Line 23 – please specify if you are referring to destructive size or relative size here; the JAN citation has this as a size 4 avalanche.
Line 25 – second largest avalanche in Japan ever recorded? This line needs clarification.
Line 48-50 – this citation and information does not seem to fit with the other specific high-magnitude events included as background.
Figures – please add some sort of scale to the maps in the Figures 3, 4, 6, 8, 9
Section 4.4 – these results represent the core of the paper, based on the title, but also, in my opinion, represent the shakiest results with uncertainty and error transmitted throughout the model chain (e.g modeled weather -> modeled snowpack conditions -> modeled stability index). I think a more comprehensive treatment of uncertainty needs to be included in any future iterations of this work.
With regards to the weak layers identified via SNOWPACK modeling (as an example of where more discussion or validation data are needed) – as I understand the results, two layers of preserved precipitation particles under up to 2 m of storm snow resulted in the avalanche release. However, the conclusion states “the temperature difference between snow layers depending on aspect would cause these two different WLs through the metamorphosis process.” I would think that metamorphosis of any kind in the snowpack would result in the destruction of the precipitation particles, either through rounding or facet development. In any case, although I have never worked in a similar snow climate, it’s a bit difficult for me to imagine preserved precipitation particles persisting as a weak layer under a 2 m slab, especially over a ten-day period in the case of WL-B. Perhaps a figure showing the modeled snowpack stratigraphy prior to avalanche release (ideally supplemented with some observed snowpack data from the 2020/2021 season) could help clear this up?

Citation: https://doi.org/10.5194/nhess-2023-5-RC1
- AC1: 'Reply on RC1', Yuta Katsuyama, 14 Mar 2023
  
  Thank you very much indeed for your kind review. We reply to the comments as followings. Your comments and our replies are in italics and solids, respectively.
  
  In this work, the authors attempt to characterize conditions snowpack conditions in the release area of the avalanche using UAV-derived snow depth data from a field campaign in March 2022 and snowpack information derived from SNOWPACK model results forced with an operational atmospheric model with 5 km spatial resolution. The authors interpret their results as indicating two heavy snowfall events in the two weeks preceding the avalanche resulted in two weak layers of precipitation particles which contributed to the avalanche’s release. I found the paper generally well-structured and understandable, but it may benefit from some language editing.
  
  Case-study documentation of large snow avalanches plays an important role in better understanding the processes behind these events. In particular, work such as the present study which combines a variety of field and model-based methods can help address scientific questions related to the spatial variability of snowpack conditions leading to large avalanches and falls well within the scope of NHESS. The authors’ premise to combine UAV-supported snow depth mapping and snowpack modeling to better characterize a high-impact avalanche event can be of interest to a broad audience of scientists and practitioners reading NHESS.
  Thank you very much again for your understanding of the values and importance of our study. An English editing service has already checked this manuscript. If pointing out hardly understandable statements as specifically as possible, we might ask the English editing service for further English corrections.
  
  My main concern with this work stems from the lack of field data presented from the event, or even the season of the event, itself. Although the authors have made an effort to justify their methods by, for example, showing the similarity in wind conditions (Figure 11) between the snow season in which the avalanche occurred and the snow season in which they conducted field work, the assumption that fully modeled snow distribution and snowpack conditions (e.g. Section 4.4) can robustly represent a specific avalanche event in the absence of any field validation data from the event itself seems rather difficult to justify in this case. Without the inclusion of additional data which can help validate the authors’ model outputs, I am unsure if this work is publishable as an event case-study or as a methodological work. I do not have an adequate mental image of this avalanche after reading the paper and cannot even identify on the provided figures (e.g. Figures 4, 8, and 9) where in the study domain the avalanche occurred.
  As you pointed out, we could not provide any field observation data in the season of the Mt. Nodanishoji avalanche. We also agree that the field observation of the season is desired. However, as we introduced in Section 1, the field observation just after the avalanche was impossible due to bad weather and the risks. Actually, there are no observation data just before/after the Mt. Nodanishoji avalanche, to our knowledge. Hence, to improve the validity of the method, we suddenly conducted an additional observation of the spatial snow depth map using UAV on 3 Mar. 2023. We will compare the spatial snow depth map observed on 17 Mar. 2022 and the corresponding map calculated from the UAV observation on 3 Mar. 2023 after the editor accepted us to submit the revised manuscript. This comparison enables us to validate the assumption that the spatial snow depth pattern, even estimated based on the following winter season, would be comparable to the observation.
  
  The authors have clearly put a great deal of effort into this paper and present a compelling methodological premise for event documentation. I wonder if photos of the event / event’s aftermath (i.e. such as those in the referenced Japan Avalanche Network’s event report) could serve as first-order, qualitative validation data? Do any snow profiles from the 2020/2021 season exist in the area in which the avalanche occurred? Such data could, along with a much more comprehensive discussion of model and result uncertainties help bring this work to a publishable level.
  The photos of JAN’s report (JAN, 2022) are not permitted to reuse in this study due to the inclination of the copyright holder. Moreover, nobody observed physical snowpack variables such as grain type, strength, and mass of snowpack layers around Mt. Nodanishoji just before/after the avalanche due to terrible weather conditions. The largest uncertainty of our method and results stemmed from the assumption that the spatial snow depth pattern, even estimated based on the following winter season, would be comparable to the observation. As we already responded above, this kind of uncertainty will be addressed by the additional observation of the snow depth map on 3 Mar. 2023.
  
  Specific comments
  Line 23 – please specify if you are referring to destructive size or relative size here; the JAN citation has this as a size 4 avalanche.
  We will introduce the first guess of the avalanche size (size-4) by JAN in the revised manuscript.
  
  Line 25 – second largest avalanche in Japan ever recorded? This line needs clarification.
  Yes. The avalanche is the second largest in Japan ever recorded. We will correct it as per your suggestion in the revised manuscript.
  
  Line 48-50 – this citation and information does not seem to fit with the other specific high-magnitude events included as background.
  We will delete this sentence from the manuscript.
  
  Figures – please add some sort of scale to the maps in the Figures 3, 4, 6, 8, 9
  We will draw scales to the maps in the revised manuscript.
  
  Section 4.4 – these results represent the core of the paper, based on the title, but also, in my opinion, represent the shakiest results with uncertainty and error transmitted throughout the model chain (e.g modeled weather -> modeled snowpack conditions -> modeled stability index). I think a more comprehensive treatment of uncertainty needs to be included in any future iterations of this work.
  The uncertainty caused by the model chain and meteorological input can be assessed from the spatial variability of the calculated snowpack. Since the meteorological input to SNOWPACK model highly depends on topography (Section 3.4.2), the spatial variation of estimates over the avalanche release area implies the sensitivity of the model chain to the model input. However, the estimated stability index was uniformly low over the domain (Fig. 8a). Therefore, the uncertainties from the model chain and meteorological input would not be a significant problem on the snowpack estimations in the case of Mt. Nodanishoji avalanche. We will include this discussion in the revised manuscript.
  
  With regards to the weak layers identified via SNOWPACK modeling (as an example of where more discussion or validation data are needed) – as I understand the results, two layers of preserved precipitation particles under up to 2 m of storm snow resulted in the avalanche release. However, the conclusion states “the temperature difference between snow layers depending on aspect would cause these two different WLs through the metamorphosis process.” I would think that metamorphosis of any kind in the snowpack would result in the destruction of the precipitation particles, either through rounding or facet development. In any case, although I have never worked in a similar snow climate, it’s a bit difficult for me to imagine preserved precipitation particles persisting as a weak layer under a 2 m slab, especially over a ten-day period in the case of WL-B. Perhaps a figure showing the modeled snowpack stratigraphy prior to avalanche release (ideally supplemented with some observed snowpack data from the 2020/2021 season) could help clear this up?
  As you pointed out, precipitation particles (PPs) are not metamorphosed from the initial state (i.e. PPs are the initial state in the metamorphosis process). We will correct the statement “the temperature difference between snow layers depending on aspect would cause these two different WLs through the metamorphosis process” in the revised manuscript. Moreover, we will add a figure showing the temporal evolution of snowpack stratigraphy in the revised manuscript. The SNOWPACK model identifies the metamorphosis from PPs to another type based on dendricity, of which time evolution depends on temperature, not overburdening snow (Legning et al., 2002). Hence, in the case of the Mt. Nodanishoji avalanche, PPs lasted long periods due to low temperatures independent of overburdening snow.
  
  Japan Avalanche Network (JAN): https://snow.nadare.jp/magazines/2021/000031.html, last access: 22 June 2022.
  
  Lehning, M., Bartelt, P., Brown, B., Fierz, C., Satyawali, P.: A physical SNOWPACK model for the Swiss avalanche warning Part II. Snow microstructure, Cold Reg. Sci. Technol., 35, 147–167, doi: 10.1016/S0165-232X(02)00073-3, 2002.
  
  Citation: https://doi.org/10.5194/nhess-2023-5-AC1
RC2:
'Comment on nhess-2023-5', Anonymous Referee #2, 03 Mar 2023
The manuscript uses numerical snowpack simulation and an UAV-flight to reproduce the snowpack properties in the release area of the extreme dry-slab avalanche on Mt. Nodanishoji, Japan, in 2021. The validation of the procedure is based on a snowpit and manual snow depth measurements in the study area.
The approach of simulating the structure of the snowcover and the identification of the different weak layers can be very useful to better understand such extreme events and may help to predict similar events in the future. However, the field acquisitions were conducted one year after the extreme avalanche event, which makes the comparison and drawing of coherent conclusions questionable. The authors argue with generally similar wind conditions and weather situations, but a well-founded validation of the simulated snowpack at the time of the event is not given.
Furthermore, the overlap of the presented study area for the field acquisitions, the numerical simulations and the avalanche event is not clear. How was the extent of the areal acquisition and the numerical simulation defined? Why are those not fully overlapping?
The above concerns need to be addressed to allow for a re-evaluation of the publication of this manuscript.
In addition to the main points you could improve:
Structure of the introduction: I would suggest starting with a more general view on big snow avalanches, state of the art in methods for snow depth acquisitions or the numerical modelling of snowpack and describe the specifications of the event later, or in a separate section.

Snow height calculation from the UAV flight: The 5m DEM from GSI seems to introduce a large error in the observed snow depths. A suggestion would be to conduct another UAV flight in snow-off conditions, to match the spatial resolution. If that was not possible, the question remains, why the 5m DEM was first interpolated to 10cm, but the final result was again resampled to 5m? It is also not clear how the 10m threshold for realistic snowheights was defined. Also, the role of the shrub areas, and its connection to the avalanche event is not well explained.

Overall, the concept of the different dependencies and interactions between the presented data and methods is difficult to follow and understand. A summarizing overview of all data, methods and dependencies would be necessary.

English language: Some passages and expressions are difficult to understand in terms of language. I would suggest considering using an English editing service.
Citation: https://doi.org/10.5194/nhess-2023-5-RC2
- AC2: 'Reply on RC2', Yuta Katsuyama, 14 Mar 2023
  
  Thank you very much indeed for your kind review. We reply to the comments as followings. Your comments and our replies are in italics and solids, respectively.
  
  The manuscript uses numerical snowpack simulation and an UAV-flight to reproduce the snowpack properties in the release area of the extreme dry-slab avalanche on Mt. Nodanishoji, Japan, in 2021. The validation of the procedure is based on a snowpit and manual snow depth measurements in the study area.
  
  The approach of simulating the structure of the snowcover and the identification of the different weak layers can be very useful to better understand such extreme events and may help to predict similar events in the future. However, the field acquisitions were conducted one year after the extreme avalanche event, which makes the comparison and drawing of coherent conclusions questionable. The authors argue with generally similar wind conditions and weather situations, but a well-founded validation of the simulated snowpack at the time of the event is not given.
  Thank you very much again for your understanding of the values of this study. As you pointed out, we could not provide any field observation data in the season when the Mt. Nodanishoji avalanche occurred. However, as we introduced in Section 1, the field observation just after the avalanche was impossible due to bad weather and the risks. Hence, to improve the validity of the method, we suddenly conducted an additional observation of the spatial snow depth map using UAV on 3 Mar. 2023. We will compare the spatial snow depth map observed on 17 Mar. 2022 and the corresponding map calculated from the UAV observation on 3 Mar. 2023 after the editor accepted us to submit the revised manuscript. This comparison enables us to validate the assumption that the spatial snow depth pattern, even estimated based on the following winter season, would be comparable to the observation.
  
  Furthermore, the overlap of the presented study area for the field acquisitions, the numerical simulations and the avalanche event is not clear. How was the extent of the areal acquisition and the numerical simulation defined? Why are those not fully overlapping?
  The spatial snow depth map was observed in a part of the release area of the Mt. Nodanishoji avalanche and where the UAV flight was safely conducted (Fig. 2a). The numerical simulation was performed in the area where the spatial snow depth was successfully observed (Fig. 4). It was difficult to make figures explicitly depicting the boundary of the avalanche release area since the eyewitness (JAN, 2022) was only direct information about the release area of the Mt. Nodanishoji avalanche. However, we could roughly understand the extent of the release area from pictures just after the avalanche (JAN, 2022).
  
  The above concerns need to be addressed to allow for a re-evaluation of the publication of this manuscript.
  In addition to the main points you could improve:
  
  Structure of the introduction: I would suggest starting with a more general view on big snow avalanches, state of the art in methods for snow depth acquisitions or the numerical modelling of snowpack and describe the specifications of the event later, or in a separate section.
  We will reconstruct Section 1 to start with a general view of the extreme avalanches in the revised manuscript. We will then describe the specifications of the Mt. Nodanishoji avalanche.
  
  Snow height calculation from the UAV flight: The 5m DEM from GSI seems to introduce a large error in the observed snow depths. A suggestion would be to conduct another UAV flight in snow-off conditions, to match the spatial resolution. If that was not possible, the question remains, why the 5m DEM was first interpolated to 10cm, but the final result was again resampled to 5m? It is also not clear how the 10m threshold for realistic snowheights was defined. Also, the role of the shrub areas, and its connection to the avalanche event is not well explained.
  The snow depth observed by UAV had large errors only around the sharp and gapped topography due to the lack of resolution of the DEM but did show a comparable result with the snow-probe observation in other topographies (Fig. 4a). The total area where the errors were large was significantly smaller than the other area, so our conclusion would not be primarily affected by the problem of the DEM resolution. Moreover, a stereo imagery measurement cannot apply to the field in summer covered by shrubs like Mt. Nodanishoji. We processed the multiple resolutions (5m -> 10 cm -> 5m) due to a technical reason. To calculate the snow depth (i.e., subtracting the DEM from the DSM), we needed to arrange these in the spatial resolution. Then, because the 10m resolution was enough for the numerical simulation and because the spatial resolution was preferred to be adjusted with the lowest resolution of the datasets, we reverted the resolution to 5m from 10cm. The threshold of 10 m was simply the general height of trees in the study area (Section 3.1).
  
  Overall, the concept of the different dependencies and interactions between the presented data and methods is difficult to follow and understand. A summarizing overview of all data, methods and dependencies would be necessary.
  We hope the above responses can improve your understanding of the concept. Moreover, in the revised manuscript, we will append a schematic diagram of how we obtain the snowpack estimations from several datasets.
  
  English language: Some passages and expressions are difficult to understand in terms of language. I would suggest considering using an English editing service.
  We are very sorry that our manuscript contains English, which is difficult to understand. However, the English editing service has already checked this manuscript before the initial submission. If pointing out hardly understandable statements as specifically as possible, we might ask the English editing service for further English corrections.
  
  Japan Avalanche Network (JAN): https://snow.nadare.jp/magazines/2021/000031.html, last access: 22 June 2022.
  
  Citation: https://doi.org/10.5194/nhess-2023-5-AC2

Status: closed

RC1:
'Comment on nhess-2023-5', Anonymous Referee #1, 28 Feb 2023

General comments
Thank you for the opportunity to review this work investigating snowpack conditions relating to a large slab avalanche event on Mt. Nodanishoji, Japan in January 2021.
In this work, the authors attempt to characterize conditions snowpack conditions in the release area of the avalanche using UAV-derived snow depth data from a field campaign in March 2022 and snowpack information derived from SNOWPACK model results forced with an operational atmospheric model with 5 km spatial resolution. The authors interpret their results as indicating two heavy snowfall events in the two weeks preceding the avalanche resulted in two weak layers of precipitation particles which contributed to the avalanche’s release. I found the paper generally well-structured and understandable, but it may benefit from some language editing.
Case-study documentation of large snow avalanches plays an important role in better understanding the processes behind these events. In particular, work such as the present study which combines a variety of field and model-based methods can help address scientific questions related to the spatial variability of snowpack conditions leading to large avalanches and falls well within the scope of NHESS. The authors’ premise to combine UAV-supported snow depth mapping and snowpack modeling to better characterize a high-impact avalanche event can be of interest to a broad audience of scientists and practitioners reading NHESS.
My main concern with this work stems from the lack of field data presented from the event, or even the season of the event, itself. Although the authors have made an effort to justify their methods by, for example, showing the similarity in wind conditions (Figure 11) between the snow season in which the avalanche occurred and the snow season in which they conducted field work, the assumption that fully modeled snow distribution and snowpack conditions (e.g. Section 4.4) can robustly represent a specific avalanche event in the absence of any field validation data from the event itself seems rather difficult to justify in this case. Without the inclusion of additional data which can help validate the authors’ model outputs, I am unsure if this work is publishable as an event case-study or as a methodological work. I do not have an adequate mental image of this avalanche after reading the paper and cannot even identify on the provided figures (e.g. Figures 4, 8, and 9) where in the study domain the avalanche occurred.
The authors have clearly put a great deal of effort into this paper and present a compelling methodological premise for event documentation. I wonder if photos of the event / event’s aftermath (i.e. such as those in the referenced Japan Avalanche Network’s event report) could serve as first-order, qualitative validation data? Do any snow profiles from the 2020/2021 season exist in the area in which the avalanche occurred? Such data could, along with a much more comprehensive discussion of model and result uncertainties help bring this work to a publishable level.
Specific comments
Line 23 – please specify if you are referring to destructive size or relative size here; the JAN citation has this as a size 4 avalanche.
Line 25 – second largest avalanche in Japan ever recorded? This line needs clarification.
Line 48-50 – this citation and information does not seem to fit with the other specific high-magnitude events included as background.
Figures – please add some sort of scale to the maps in the Figures 3, 4, 6, 8, 9
Section 4.4 – these results represent the core of the paper, based on the title, but also, in my opinion, represent the shakiest results with uncertainty and error transmitted throughout the model chain (e.g modeled weather -> modeled snowpack conditions -> modeled stability index). I think a more comprehensive treatment of uncertainty needs to be included in any future iterations of this work.
With regards to the weak layers identified via SNOWPACK modeling (as an example of where more discussion or validation data are needed) – as I understand the results, two layers of preserved precipitation particles under up to 2 m of storm snow resulted in the avalanche release. However, the conclusion states “the temperature difference between snow layers depending on aspect would cause these two different WLs through the metamorphosis process.” I would think that metamorphosis of any kind in the snowpack would result in the destruction of the precipitation particles, either through rounding or facet development. In any case, although I have never worked in a similar snow climate, it’s a bit difficult for me to imagine preserved precipitation particles persisting as a weak layer under a 2 m slab, especially over a ten-day period in the case of WL-B. Perhaps a figure showing the modeled snowpack stratigraphy prior to avalanche release (ideally supplemented with some observed snowpack data from the 2020/2021 season) could help clear this up?

Citation: https://doi.org/10.5194/nhess-2023-5-RC1
- AC1: 'Reply on RC1', Yuta Katsuyama, 14 Mar 2023
  
  Thank you very much indeed for your kind review. We reply to the comments as followings. Your comments and our replies are in italics and solids, respectively.
  
  In this work, the authors attempt to characterize conditions snowpack conditions in the release area of the avalanche using UAV-derived snow depth data from a field campaign in March 2022 and snowpack information derived from SNOWPACK model results forced with an operational atmospheric model with 5 km spatial resolution. The authors interpret their results as indicating two heavy snowfall events in the two weeks preceding the avalanche resulted in two weak layers of precipitation particles which contributed to the avalanche’s release. I found the paper generally well-structured and understandable, but it may benefit from some language editing.
  
  Case-study documentation of large snow avalanches plays an important role in better understanding the processes behind these events. In particular, work such as the present study which combines a variety of field and model-based methods can help address scientific questions related to the spatial variability of snowpack conditions leading to large avalanches and falls well within the scope of NHESS. The authors’ premise to combine UAV-supported snow depth mapping and snowpack modeling to better characterize a high-impact avalanche event can be of interest to a broad audience of scientists and practitioners reading NHESS.
  Thank you very much again for your understanding of the values and importance of our study. An English editing service has already checked this manuscript. If pointing out hardly understandable statements as specifically as possible, we might ask the English editing service for further English corrections.
  
  My main concern with this work stems from the lack of field data presented from the event, or even the season of the event, itself. Although the authors have made an effort to justify their methods by, for example, showing the similarity in wind conditions (Figure 11) between the snow season in which the avalanche occurred and the snow season in which they conducted field work, the assumption that fully modeled snow distribution and snowpack conditions (e.g. Section 4.4) can robustly represent a specific avalanche event in the absence of any field validation data from the event itself seems rather difficult to justify in this case. Without the inclusion of additional data which can help validate the authors’ model outputs, I am unsure if this work is publishable as an event case-study or as a methodological work. I do not have an adequate mental image of this avalanche after reading the paper and cannot even identify on the provided figures (e.g. Figures 4, 8, and 9) where in the study domain the avalanche occurred.
  As you pointed out, we could not provide any field observation data in the season of the Mt. Nodanishoji avalanche. We also agree that the field observation of the season is desired. However, as we introduced in Section 1, the field observation just after the avalanche was impossible due to bad weather and the risks. Actually, there are no observation data just before/after the Mt. Nodanishoji avalanche, to our knowledge. Hence, to improve the validity of the method, we suddenly conducted an additional observation of the spatial snow depth map using UAV on 3 Mar. 2023. We will compare the spatial snow depth map observed on 17 Mar. 2022 and the corresponding map calculated from the UAV observation on 3 Mar. 2023 after the editor accepted us to submit the revised manuscript. This comparison enables us to validate the assumption that the spatial snow depth pattern, even estimated based on the following winter season, would be comparable to the observation.
  
  The authors have clearly put a great deal of effort into this paper and present a compelling methodological premise for event documentation. I wonder if photos of the event / event’s aftermath (i.e. such as those in the referenced Japan Avalanche Network’s event report) could serve as first-order, qualitative validation data? Do any snow profiles from the 2020/2021 season exist in the area in which the avalanche occurred? Such data could, along with a much more comprehensive discussion of model and result uncertainties help bring this work to a publishable level.
  The photos of JAN’s report (JAN, 2022) are not permitted to reuse in this study due to the inclination of the copyright holder. Moreover, nobody observed physical snowpack variables such as grain type, strength, and mass of snowpack layers around Mt. Nodanishoji just before/after the avalanche due to terrible weather conditions. The largest uncertainty of our method and results stemmed from the assumption that the spatial snow depth pattern, even estimated based on the following winter season, would be comparable to the observation. As we already responded above, this kind of uncertainty will be addressed by the additional observation of the snow depth map on 3 Mar. 2023.
  
  Specific comments
  Line 23 – please specify if you are referring to destructive size or relative size here; the JAN citation has this as a size 4 avalanche.
  We will introduce the first guess of the avalanche size (size-4) by JAN in the revised manuscript.
  
  Line 25 – second largest avalanche in Japan ever recorded? This line needs clarification.
  Yes. The avalanche is the second largest in Japan ever recorded. We will correct it as per your suggestion in the revised manuscript.
  
  Line 48-50 – this citation and information does not seem to fit with the other specific high-magnitude events included as background.
  We will delete this sentence from the manuscript.
  
  Figures – please add some sort of scale to the maps in the Figures 3, 4, 6, 8, 9
  We will draw scales to the maps in the revised manuscript.
  
  Section 4.4 – these results represent the core of the paper, based on the title, but also, in my opinion, represent the shakiest results with uncertainty and error transmitted throughout the model chain (e.g modeled weather -> modeled snowpack conditions -> modeled stability index). I think a more comprehensive treatment of uncertainty needs to be included in any future iterations of this work.
  The uncertainty caused by the model chain and meteorological input can be assessed from the spatial variability of the calculated snowpack. Since the meteorological input to SNOWPACK model highly depends on topography (Section 3.4.2), the spatial variation of estimates over the avalanche release area implies the sensitivity of the model chain to the model input. However, the estimated stability index was uniformly low over the domain (Fig. 8a). Therefore, the uncertainties from the model chain and meteorological input would not be a significant problem on the snowpack estimations in the case of Mt. Nodanishoji avalanche. We will include this discussion in the revised manuscript.
  
  With regards to the weak layers identified via SNOWPACK modeling (as an example of where more discussion or validation data are needed) – as I understand the results, two layers of preserved precipitation particles under up to 2 m of storm snow resulted in the avalanche release. However, the conclusion states “the temperature difference between snow layers depending on aspect would cause these two different WLs through the metamorphosis process.” I would think that metamorphosis of any kind in the snowpack would result in the destruction of the precipitation particles, either through rounding or facet development. In any case, although I have never worked in a similar snow climate, it’s a bit difficult for me to imagine preserved precipitation particles persisting as a weak layer under a 2 m slab, especially over a ten-day period in the case of WL-B. Perhaps a figure showing the modeled snowpack stratigraphy prior to avalanche release (ideally supplemented with some observed snowpack data from the 2020/2021 season) could help clear this up?
  As you pointed out, precipitation particles (PPs) are not metamorphosed from the initial state (i.e. PPs are the initial state in the metamorphosis process). We will correct the statement “the temperature difference between snow layers depending on aspect would cause these two different WLs through the metamorphosis process” in the revised manuscript. Moreover, we will add a figure showing the temporal evolution of snowpack stratigraphy in the revised manuscript. The SNOWPACK model identifies the metamorphosis from PPs to another type based on dendricity, of which time evolution depends on temperature, not overburdening snow (Legning et al., 2002). Hence, in the case of the Mt. Nodanishoji avalanche, PPs lasted long periods due to low temperatures independent of overburdening snow.
  
  Japan Avalanche Network (JAN): https://snow.nadare.jp/magazines/2021/000031.html, last access: 22 June 2022.
  
  Lehning, M., Bartelt, P., Brown, B., Fierz, C., Satyawali, P.: A physical SNOWPACK model for the Swiss avalanche warning Part II. Snow microstructure, Cold Reg. Sci. Technol., 35, 147–167, doi: 10.1016/S0165-232X(02)00073-3, 2002.
  
  Citation: https://doi.org/10.5194/nhess-2023-5-AC1
RC2:
'Comment on nhess-2023-5', Anonymous Referee #2, 03 Mar 2023
The manuscript uses numerical snowpack simulation and an UAV-flight to reproduce the snowpack properties in the release area of the extreme dry-slab avalanche on Mt. Nodanishoji, Japan, in 2021. The validation of the procedure is based on a snowpit and manual snow depth measurements in the study area.
The approach of simulating the structure of the snowcover and the identification of the different weak layers can be very useful to better understand such extreme events and may help to predict similar events in the future. However, the field acquisitions were conducted one year after the extreme avalanche event, which makes the comparison and drawing of coherent conclusions questionable. The authors argue with generally similar wind conditions and weather situations, but a well-founded validation of the simulated snowpack at the time of the event is not given.
Furthermore, the overlap of the presented study area for the field acquisitions, the numerical simulations and the avalanche event is not clear. How was the extent of the areal acquisition and the numerical simulation defined? Why are those not fully overlapping?
The above concerns need to be addressed to allow for a re-evaluation of the publication of this manuscript.
In addition to the main points you could improve:
Structure of the introduction: I would suggest starting with a more general view on big snow avalanches, state of the art in methods for snow depth acquisitions or the numerical modelling of snowpack and describe the specifications of the event later, or in a separate section.

Snow height calculation from the UAV flight: The 5m DEM from GSI seems to introduce a large error in the observed snow depths. A suggestion would be to conduct another UAV flight in snow-off conditions, to match the spatial resolution. If that was not possible, the question remains, why the 5m DEM was first interpolated to 10cm, but the final result was again resampled to 5m? It is also not clear how the 10m threshold for realistic snowheights was defined. Also, the role of the shrub areas, and its connection to the avalanche event is not well explained.

Overall, the concept of the different dependencies and interactions between the presented data and methods is difficult to follow and understand. A summarizing overview of all data, methods and dependencies would be necessary.

English language: Some passages and expressions are difficult to understand in terms of language. I would suggest considering using an English editing service.
Citation: https://doi.org/10.5194/nhess-2023-5-RC2
- AC2: 'Reply on RC2', Yuta Katsuyama, 14 Mar 2023
  
  Thank you very much indeed for your kind review. We reply to the comments as followings. Your comments and our replies are in italics and solids, respectively.
  
  The manuscript uses numerical snowpack simulation and an UAV-flight to reproduce the snowpack properties in the release area of the extreme dry-slab avalanche on Mt. Nodanishoji, Japan, in 2021. The validation of the procedure is based on a snowpit and manual snow depth measurements in the study area.
  
  The approach of simulating the structure of the snowcover and the identification of the different weak layers can be very useful to better understand such extreme events and may help to predict similar events in the future. However, the field acquisitions were conducted one year after the extreme avalanche event, which makes the comparison and drawing of coherent conclusions questionable. The authors argue with generally similar wind conditions and weather situations, but a well-founded validation of the simulated snowpack at the time of the event is not given.
  Thank you very much again for your understanding of the values of this study. As you pointed out, we could not provide any field observation data in the season when the Mt. Nodanishoji avalanche occurred. However, as we introduced in Section 1, the field observation just after the avalanche was impossible due to bad weather and the risks. Hence, to improve the validity of the method, we suddenly conducted an additional observation of the spatial snow depth map using UAV on 3 Mar. 2023. We will compare the spatial snow depth map observed on 17 Mar. 2022 and the corresponding map calculated from the UAV observation on 3 Mar. 2023 after the editor accepted us to submit the revised manuscript. This comparison enables us to validate the assumption that the spatial snow depth pattern, even estimated based on the following winter season, would be comparable to the observation.
  
  Furthermore, the overlap of the presented study area for the field acquisitions, the numerical simulations and the avalanche event is not clear. How was the extent of the areal acquisition and the numerical simulation defined? Why are those not fully overlapping?
  The spatial snow depth map was observed in a part of the release area of the Mt. Nodanishoji avalanche and where the UAV flight was safely conducted (Fig. 2a). The numerical simulation was performed in the area where the spatial snow depth was successfully observed (Fig. 4). It was difficult to make figures explicitly depicting the boundary of the avalanche release area since the eyewitness (JAN, 2022) was only direct information about the release area of the Mt. Nodanishoji avalanche. However, we could roughly understand the extent of the release area from pictures just after the avalanche (JAN, 2022).
  
  The above concerns need to be addressed to allow for a re-evaluation of the publication of this manuscript.
  In addition to the main points you could improve:
  
  Structure of the introduction: I would suggest starting with a more general view on big snow avalanches, state of the art in methods for snow depth acquisitions or the numerical modelling of snowpack and describe the specifications of the event later, or in a separate section.
  We will reconstruct Section 1 to start with a general view of the extreme avalanches in the revised manuscript. We will then describe the specifications of the Mt. Nodanishoji avalanche.
  
  Snow height calculation from the UAV flight: The 5m DEM from GSI seems to introduce a large error in the observed snow depths. A suggestion would be to conduct another UAV flight in snow-off conditions, to match the spatial resolution. If that was not possible, the question remains, why the 5m DEM was first interpolated to 10cm, but the final result was again resampled to 5m? It is also not clear how the 10m threshold for realistic snowheights was defined. Also, the role of the shrub areas, and its connection to the avalanche event is not well explained.
  The snow depth observed by UAV had large errors only around the sharp and gapped topography due to the lack of resolution of the DEM but did show a comparable result with the snow-probe observation in other topographies (Fig. 4a). The total area where the errors were large was significantly smaller than the other area, so our conclusion would not be primarily affected by the problem of the DEM resolution. Moreover, a stereo imagery measurement cannot apply to the field in summer covered by shrubs like Mt. Nodanishoji. We processed the multiple resolutions (5m -> 10 cm -> 5m) due to a technical reason. To calculate the snow depth (i.e., subtracting the DEM from the DSM), we needed to arrange these in the spatial resolution. Then, because the 10m resolution was enough for the numerical simulation and because the spatial resolution was preferred to be adjusted with the lowest resolution of the datasets, we reverted the resolution to 5m from 10cm. The threshold of 10 m was simply the general height of trees in the study area (Section 3.1).
  
  Overall, the concept of the different dependencies and interactions between the presented data and methods is difficult to follow and understand. A summarizing overview of all data, methods and dependencies would be necessary.
  We hope the above responses can improve your understanding of the concept. Moreover, in the revised manuscript, we will append a schematic diagram of how we obtain the snowpack estimations from several datasets.
  
  English language: Some passages and expressions are difficult to understand in terms of language. I would suggest considering using an English editing service.
  We are very sorry that our manuscript contains English, which is difficult to understand. However, the English editing service has already checked this manuscript before the initial submission. If pointing out hardly understandable statements as specifically as possible, we might ask the English editing service for further English corrections.
  
  Japan Avalanche Network (JAN): https://snow.nadare.jp/magazines/2021/000031.html, last access: 22 June 2022.
  
  Citation: https://doi.org/10.5194/nhess-2023-5-AC2

Yuta Katsuyama, Takafumi Katsushima, Satoru Adachi, and Yukari Takeuchi

Viewed

Total article views: 692 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
520	145	27	692	15	17

HTML: 520
PDF: 145
XML: 27
Total: 692
BibTeX: 15
EndNote: 17

Views and downloads (calculated since 09 Feb 2023)

Month	HTML	PDF	XML	Total
Feb 2023	212	40	6	258
Mar 2023	108	32	7	147
Apr 2023	23	14	1	38
May 2023	24	7	1	32
Jun 2023	11	2	0	13
Jul 2023	29	10	2	41
Aug 2023	17	7	1	25
Sep 2023	25	4	0	29
Oct 2023	12	7	3	22
Nov 2023	9	4	1	14
Dec 2023	12	2	0	14
Jan 2024	13	2	0	15
Feb 2024	9	3	4	16
Mar 2024	13	11	1	25
Apr 2024	3	0	3

Cumulative views and downloads (calculated since 09 Feb 2023)

Month	HTML	PDF	XML	Total
Feb 2023	212	40	6	258
Mar 2023	108	32	7	147
Apr 2023	23	14	1	38
May 2023	24	7	1	32
Jun 2023	11	2	0	13
Jul 2023	29	10	2	41
Aug 2023	17	7	1	25
Sep 2023	25	4	0	29
Oct 2023	12	7	3	22
Nov 2023	9	4	1	14
Dec 2023	12	2	0	14
Jan 2024	13	2	0	15
Feb 2024	9	3	4	16
Mar 2024	13	11	1	25
Apr 2024	3	0	3

Viewed (geographical distribution)

Total article views: 646 (including HTML, PDF, and XML) Thereof 646 with geography defined and 0 with unknown origin.

Country	#	Views	%

Cited

Latest update: 07 Apr 2024

Short summary

An extreme dry-slab snow-avalanche, the second largest in Japan, occurred on 10 Jan. 2021 at Mt. Nodanishoji, Gifu, Japan. Based on a snow depth distribution observed by an unmanned aerial vehicle and a numerical snowpack simulation in the avalanche release area, the avalanche was caused by two heavy snowfall events, ~3 and ~10 days before the avalanche onset. These two snowfall events caused two different weak layers made of precipitation particles and associated slabs above other weak layers.


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

Spatial snowpack properties in a snow-avalanche release area: An extreme dry-slab avalanche case on Mt. Nodanishoji, Japan, in 2021

Viewed

Viewed (geographical distribution)

Cited

1 citations as recorded by crossref.