the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 16 Jul 2024
Optimizing rock glaciers activity classification in South Tyrol (North-East Italy): integrating multisource data with statistical modelling
Abstract. As a consequence of climate warming, high-altitude periglacial and glacial environments exhibit the clearest signs of cryosphere degradation, and the Alps serve as a natural laboratory for studying the primary effects on permafrost-related features. Our research in South Tyrol, North-East Italy, aimed to develop an updated classification system for rock glaciers activity, based on remote sensing data and statistical models, with the aim of categorizing them as active, transitional, or relict according to the recent RGIK guidelines. Since the current regional inventory includes activity attributes based only on morphological observations and differential SAR interferometry (DInSAR) coherence, it lacks a comprehensive definition integrating climatic drivers, displacement rates, and morphometric parameters. To address this, we utilized the Alaska Satellite Facility's InSAR cloud computing, employing small baseline subset (SBAS) approach and MintPy algorithms to extract velocity data for each rock glacier in South Tyrol. Additionally, we analyzed geomorphological and climatic maps derived from in-situ and remote sensing data to obtain descriptive parameters influencing rock glaciers development and activity. From a wide range of potential variables, we selected eight key predictors, representing physical (e.g. temperature), morphological (e.g. roughness), and dynamic (e.g. velocity and coherence indicators) attributes. These predictors were successively integrated in a multiclass generalized additive mixing model (GAM) classifier to categorize the landforms. Applying this model to the entire dataset (achieving an AUC over 0.9) allowed us to address gaps in previous classification methods and provided activity attributes for previously unclassified rock glaciers, along with associated uncertainty values. Our approach improved classification accuracy, leaving only 3.5 % of features unclassified compared to 13 % in morphological classification and 18.5 % in DInSAR-based methods. The results revealed a predominance of relict features (~75 %) and a smaller number of active ones (~10 %). The distribution of active, transitional, and relict classes suggests that the transition from active to relict states is not a direct process. Instead, an intermediate transitional phase is commonly observed. This comprehensive approach refines the categorization of mapped features and improves our understanding of the factors influencing rock glaciers activity in alpine environment.
-
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
(2073 KB)
-
Supplement
(584 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2073 KB) - Metadata XML
-
Supplement
(584 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2024-1511', Anonymous Referee #1, 04 Nov 2024
In this paper, the authors integrate existing rock glacier results from South Tyrol region, including the Autonomous Province of Bolzano/Bozen (PAB) rock glacier inventory data and the DInSAR-derived movement status by Bertone et al., (2019). By combining geomorphological characteristics of the rock glaciers, climatic driving factors, and InSAR products, the statistical model is calibrated and validated. This model is then used to optimize the identification of A (Active), T (Transitional), and R (Relict) states of rock glaciers in the region, while also describing the relationship between rock glacier states and multiple driving variables.
General comments
- The reliability of optimizing rock glacier state. The author combines the PAB (2010) and Bertone (2019) rock glacier inventories in South Tyrol and uses various statistical factors to setup the GAM model, which is then applied to optimize the rock glacier states in the study area. However, it should be noted that if the original rock glacier states are not entirely accurate, the resulting GAM will inevitably carry uncertainties, making it inappropriate to use such a model to further optimize these states. Further assessment of the classification accuracy should be implemented or discussed.
- In Section 3.4, it is unclear on the training data used to support the setup of GAM model. How many rock glacier samples were used and treated to train the model, including the samples for each status type (A, T, R), should be clearly presented.
- The paper uses SBAS-InSAR products, including velocity and coherence, as inputs to implement the statistical model, which is a distinctive aspect of the study. However, when the movement rate map of existing rock glaciers is available, classification can be conducted directly from a kinematic perspective (RGIK, 2023): for example, 10 cm/yr > v > 1 cm/yr for T (Transitional), v > 10 cm/yr for A (Active), and v < 1 cm/yr for R (Relict). This quantitative description of rock glacier movement status would be more straightforward. Moreover, if the prerequisite for identifying rock glacier states is to perform DInSAR or SBAS-InSAR to obtain product input for the statistical model, it further limits the application of rock glacier analysis in mountainous regions. To comprehensively define rock glacier states by integrating climatic driving factors, displacement rates, and morphological parameters, it is advisable to compare the results obtained from the current comprehensive definition with the states identified solely based on movement velocity data, or to use empirical data to demonstrate that the states identified by this comprehensive method are more accurate.
Specific comments
Line 28: “Our approach improved classification accuracy, leaving only 3.5% of features unclassified compared to 13% in morphological classification and 18.5% in DInSAR-based methods.” If "feature" here refers to the number of rock glaciers, such as the 3.5% representing 63 out of a total of 1,779, then it doesn't represent accuracy. Instead, it should be considered an enhancement over previous work, providing a more complete or comprehensive cataloging of rock glacier states.
Line 60: “Although widely used, this classification brings two relevant limitations both from subjectivity point of view (activity attribution based on geomorphological approach is depended on the operator skills) as well as categorization since the activity of rock glaciers is considered constant over time at the scale of decades to centuries.” The classification into "intact" and "relict" does not inherently introduce subjectivity; rather, it is the geomorphological classification process that carries subjective factors. If we classify Active (A), Relict (R), and Transitional (T) states based on geomorphological characteristics, it would also involve subjectivity. The use of initial identification data, such as the x-axis in Table 1, which is also based on geomorphological characteristics, introduces subjectivity into the GAM as well. Please explain why the assumption of long-term invariability in activity state identification would be considered a limitation of the "relict" and "intact" classification.
Line 125: Bertone et al., (2019)
Lines 193-199: What role does precipitation play in the overall text? Precipitation was not included as an input in the GAM; is it meant to be part of the discussion on precipitation? However, there doesn’t appear to be any statistical information provided to support the author's discussion on precipitation.
Section 3.3: How were the velocity datasets from ascending and descending InSAR results integrated? Why was the calculation of slope velocities for rock glaciers not performed, given that both ascending and descending InSAR maps were derived?
Line 253: Is the 100m value an empirical choice? If a unit within a rock glacier system is entirely occupied by other units within a 100 m buffer zone, how is this situation handled? I agree with the author's idea of calculating the increment by comparing the values inside and outside the buffer zone. This increment can potentially distinguish between the rock glacier's intrinsic movement and movement caused by external factors. It seems that further analysis or application of this increment has not been addressed in the following sections.
Line 285: Cross-validation is generally used because the data is limited, and it helps improve the model's generalization capability. It also allows for better evaluation and enhances the model's ability to fit data outside the training set. Please clarify the rationale for consecutively using 2-fold, 5-fold, and 10-fold cross-validation.
Line 309: From the boxplot (Figure 5c), it appears that VRM (Vector Ruggedness Measure) doesn't show a significant signal, which might suggest that surface roughness is unrelated to the activity status. Therefore, the inclusion of VRM in the GAM model seems unjustified. There are many other potential factors that could serve as surface condition indicators, such as terrain curvature and vegetation cover.
Section 4.2: Were the normalized or raw values of the eight variables used in the GAM model? How many rock glacier samples were used? Is it the number of A+T+R as mentioned in Table 1? Additionally, how did the author deal with the rock glaciers located in the layover/shadow regions of the SAR data?
Line 383: The GAM is trained as a classifier based on the environmental factors of the original rock glacier data and the DInSAR products. However, it seems unreasonable to apply the trained GAM model to all 1,779 rock glaciers in the region, including those initially used as training datasets.
Lines 465-472: The paper lacks information on the statistical relationship between precipitation and rock glacier activity status.
Lines 500-507: The current method for evaluating identification accuracy involves both InSAR movement signals and distinct morphological characteristics, providing a quantitative perspective first and then a subjective morphological identification perspective. However, if the current GAM identification method requires InSAR products, such as movement velocity and coherence as inputs, why not directly use the velocity map along with geomorphological features to evaluate the status?
Lines 541-551: Vlos cannot fully represent the true movement pattern of rock glaciers. Could this limitation affect the uncertainty of the GAM? There might be cases where some rock glaciers have a large Vlos but a much smaller actual movement rate, thus introducing uncertainty.
Comments on Figures
Figure 2: The "calibration" step in the “multiclass GAM model” is vague. Typically, calibration involves adjusting something that was previously incorrect to make it correct. Could you clarify what this step entails in the context of your model?
Figure 3: The units of the unwrapped phase should be “rad” rather than “cm/yr.”
Figure 4: Do the input dataset “look vectors” correspond to Figure 4b (the visibility map)?
Figure 5: How were the outliers identified, and how was the lower limit chosen, especially for the coherence, such as in Figure 5d? The median and quartile changes with R, T, A are quite reasonable, but I've noticed that there are many outliers close to your lower limit. Please explain how the lower limit was determined and why there are so many outliers, not just in Figure 5d. Also, why not calculate and present the “mean velocity,” “variance of velocity,” and “velocity outside delta (∆)” plots like the coherence panel?
Figure 5h: Is a velocity threshold (0.02 m/yr) being applied?
Figure 6f: Although the relationship may not be immediately apparent, I have observed that many rock glaciers are frequently located in convergent areas. This pattern is intriguing, and any insights or explanation regarding this observation would be valuable.
Figure 11b: Would replacing the LOS velocity values with slope velocity result in a smoother transition from red to blue?
Citation: https://doi.org/10.5194/egusphere-2024-1511-RC1 - AC1: 'Reply on RC1', Chiara Crippa, 02 Dec 2024
-
RC2: 'Comment on egusphere-2024-1511', Anonymous Referee #2, 24 Feb 2025
Title: Optimizing rock glaciers activity classification in South Tyrol (North-East Italy): integrating multisource data with statistical modelling
1) Overall quality and general comments
Rock glaciers are key indicators of permafrost in alpine regions, formed by a seasonally frozen detrital layer overlying supersaturated debris of ice or pure ice, and characterized by gravity flow. Their distribution is influenced by topographic and climatic factors at different scales, and they play a crucial role in high-altitude hydrology by storing ice and water. Traditionally, rock glaciers are classified as active, inactive, or relict based on ice content and movement. However, rising permafrost temperatures have led to an accelerating trend, encouraging an updated classification that considers sediment transport efficiency. In the regional territory of South Tyrol, two rock glaciers activity classifications coexist (Autonomous Province of Bolzano/Bozen and Bertone et al., 2019). By combining geomorphological characteristics, climatic driving factors, and InSAR products, the authors develop a statistical model to refine the classification of rock glaciers.
This study represents an innovative contribution since it integrates multiple variables into a multiclass generalised additive mixing (GAM) model to predict rock glacier activity. Using remote sensing, ground-based data, and digital terrain models, the workflow involves extracting velocity and environmental attributes at a regional scale, calibrating and validating a multiclass predictive GAM, and applying it to classify landforms based on their activity status.
The integration of remote sensing data and statistical modelling significantly advances current methods for assessing rock glacier dynamics. The study is well-structured, with a clear research objective and methodology. The statistical approach, particularly the use of a multiclass GAM model, is effective for the research aims. The discussion is robust, highlighting both its contributions and its limitations. The figures and tables are clear, informative, and support the understanding of the concepts. Finally, this work advances the understanding of rock glacier dynamics by refining their classification system and linking their activity states to a range of predictor variables.
2) Individual scientific questions
3.4.1. Statistical modelling
How did the authors ensure the robustness of the GAM model in terms of the selection and evaluation of predictor variables?
4.1. Exploratory Data Analysis
How did you decide which variables to retain for further analysis, and why were some variables, such as elevation, excluded to avoid redundancy despite their high discriminatory power? Could you clarify the rationale behind this choice?
3) Specific comments on the manuscript
Line 119: How many rock glaciers are present in the analysed dataset?
Line 127: The classification 'n.d.' is unclear. Could you please clarify its meaning and usage in this context?
Line 148: Could you explain in more detail how the variables were extracted and assigned to each individual rock glaciers?
Lines 207-210: "Using this SCD parameter, a potential correlation between the rock glaciers’ activity at a regional level was made[...]" Could you explain this statement more clearly? How was the correlation assessed, and what were the main findings regarding the SCD in relation to the rock glaciers' activity?
Figure 4: Does the term "look vector" refer to the Line of Sight (LOS) of the satellites? Could you also better explain if the shadowing and layover effects part is the C index analysis?
Figure 4: Is the vLOS referring to vertical velocity? Additionally, could you adjust the colour scale bar to range between -8 and 8 cm/year to improve the clarity of the data representation?
Lines 244-248: “For each rock glacier polygon, mean values for environmental and climatic variables were assigned based on the values within the polygon boundary. Furthermore, for DInSAR-related variables (i.e., velocity and coherence), additional statistical descriptors […]”. Can you explain how the uncertainty was computed for each rock glacier, based on the SAR data coverage? How did you assess the spatial uncertainty within each polygon?
Line 243: "Starting from the distribution map of the rock glaciers and considering their displacement range, we made two distinctions [...]". Could you clarify the rationale behind the choice of a 100-meter buffer around each mapped landform? How was this distance determined, and how does it affect the classification?
Lines 264-266: “To discern the key factors influencing the distinction between A, R, and T rock glacier classes, we performed an initial Exploratory Data Analysis. This exploration served […]”. Could you provide more details on how this exploratory analysis was performed, and how it helped with the model?
Lines 167-272: “GAM was employed to investigate the associations between the chosen predictor variables derived from both environmental and DInSAR datasets and the response variables. GAM provides […]”. Could you provide more explanation on the use of GAM in this context? A brief discussion of the relevant literature and how GAM has been applied in other studies would strengthen this section.
-
AC2: 'Reply on RC2', Chiara Crippa, 03 Mar 2025
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-1511/egusphere-2024-1511-AC2-supplement.pdf
-
AC2: 'Reply on RC2', Chiara Crippa, 03 Mar 2025
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2024-1511', Anonymous Referee #1, 04 Nov 2024
In this paper, the authors integrate existing rock glacier results from South Tyrol region, including the Autonomous Province of Bolzano/Bozen (PAB) rock glacier inventory data and the DInSAR-derived movement status by Bertone et al., (2019). By combining geomorphological characteristics of the rock glaciers, climatic driving factors, and InSAR products, the statistical model is calibrated and validated. This model is then used to optimize the identification of A (Active), T (Transitional), and R (Relict) states of rock glaciers in the region, while also describing the relationship between rock glacier states and multiple driving variables.
General comments
- The reliability of optimizing rock glacier state. The author combines the PAB (2010) and Bertone (2019) rock glacier inventories in South Tyrol and uses various statistical factors to setup the GAM model, which is then applied to optimize the rock glacier states in the study area. However, it should be noted that if the original rock glacier states are not entirely accurate, the resulting GAM will inevitably carry uncertainties, making it inappropriate to use such a model to further optimize these states. Further assessment of the classification accuracy should be implemented or discussed.
- In Section 3.4, it is unclear on the training data used to support the setup of GAM model. How many rock glacier samples were used and treated to train the model, including the samples for each status type (A, T, R), should be clearly presented.
- The paper uses SBAS-InSAR products, including velocity and coherence, as inputs to implement the statistical model, which is a distinctive aspect of the study. However, when the movement rate map of existing rock glaciers is available, classification can be conducted directly from a kinematic perspective (RGIK, 2023): for example, 10 cm/yr > v > 1 cm/yr for T (Transitional), v > 10 cm/yr for A (Active), and v < 1 cm/yr for R (Relict). This quantitative description of rock glacier movement status would be more straightforward. Moreover, if the prerequisite for identifying rock glacier states is to perform DInSAR or SBAS-InSAR to obtain product input for the statistical model, it further limits the application of rock glacier analysis in mountainous regions. To comprehensively define rock glacier states by integrating climatic driving factors, displacement rates, and morphological parameters, it is advisable to compare the results obtained from the current comprehensive definition with the states identified solely based on movement velocity data, or to use empirical data to demonstrate that the states identified by this comprehensive method are more accurate.
Specific comments
Line 28: “Our approach improved classification accuracy, leaving only 3.5% of features unclassified compared to 13% in morphological classification and 18.5% in DInSAR-based methods.” If "feature" here refers to the number of rock glaciers, such as the 3.5% representing 63 out of a total of 1,779, then it doesn't represent accuracy. Instead, it should be considered an enhancement over previous work, providing a more complete or comprehensive cataloging of rock glacier states.
Line 60: “Although widely used, this classification brings two relevant limitations both from subjectivity point of view (activity attribution based on geomorphological approach is depended on the operator skills) as well as categorization since the activity of rock glaciers is considered constant over time at the scale of decades to centuries.” The classification into "intact" and "relict" does not inherently introduce subjectivity; rather, it is the geomorphological classification process that carries subjective factors. If we classify Active (A), Relict (R), and Transitional (T) states based on geomorphological characteristics, it would also involve subjectivity. The use of initial identification data, such as the x-axis in Table 1, which is also based on geomorphological characteristics, introduces subjectivity into the GAM as well. Please explain why the assumption of long-term invariability in activity state identification would be considered a limitation of the "relict" and "intact" classification.
Line 125: Bertone et al., (2019)
Lines 193-199: What role does precipitation play in the overall text? Precipitation was not included as an input in the GAM; is it meant to be part of the discussion on precipitation? However, there doesn’t appear to be any statistical information provided to support the author's discussion on precipitation.
Section 3.3: How were the velocity datasets from ascending and descending InSAR results integrated? Why was the calculation of slope velocities for rock glaciers not performed, given that both ascending and descending InSAR maps were derived?
Line 253: Is the 100m value an empirical choice? If a unit within a rock glacier system is entirely occupied by other units within a 100 m buffer zone, how is this situation handled? I agree with the author's idea of calculating the increment by comparing the values inside and outside the buffer zone. This increment can potentially distinguish between the rock glacier's intrinsic movement and movement caused by external factors. It seems that further analysis or application of this increment has not been addressed in the following sections.
Line 285: Cross-validation is generally used because the data is limited, and it helps improve the model's generalization capability. It also allows for better evaluation and enhances the model's ability to fit data outside the training set. Please clarify the rationale for consecutively using 2-fold, 5-fold, and 10-fold cross-validation.
Line 309: From the boxplot (Figure 5c), it appears that VRM (Vector Ruggedness Measure) doesn't show a significant signal, which might suggest that surface roughness is unrelated to the activity status. Therefore, the inclusion of VRM in the GAM model seems unjustified. There are many other potential factors that could serve as surface condition indicators, such as terrain curvature and vegetation cover.
Section 4.2: Were the normalized or raw values of the eight variables used in the GAM model? How many rock glacier samples were used? Is it the number of A+T+R as mentioned in Table 1? Additionally, how did the author deal with the rock glaciers located in the layover/shadow regions of the SAR data?
Line 383: The GAM is trained as a classifier based on the environmental factors of the original rock glacier data and the DInSAR products. However, it seems unreasonable to apply the trained GAM model to all 1,779 rock glaciers in the region, including those initially used as training datasets.
Lines 465-472: The paper lacks information on the statistical relationship between precipitation and rock glacier activity status.
Lines 500-507: The current method for evaluating identification accuracy involves both InSAR movement signals and distinct morphological characteristics, providing a quantitative perspective first and then a subjective morphological identification perspective. However, if the current GAM identification method requires InSAR products, such as movement velocity and coherence as inputs, why not directly use the velocity map along with geomorphological features to evaluate the status?
Lines 541-551: Vlos cannot fully represent the true movement pattern of rock glaciers. Could this limitation affect the uncertainty of the GAM? There might be cases where some rock glaciers have a large Vlos but a much smaller actual movement rate, thus introducing uncertainty.
Comments on Figures
Figure 2: The "calibration" step in the “multiclass GAM model” is vague. Typically, calibration involves adjusting something that was previously incorrect to make it correct. Could you clarify what this step entails in the context of your model?
Figure 3: The units of the unwrapped phase should be “rad” rather than “cm/yr.”
Figure 4: Do the input dataset “look vectors” correspond to Figure 4b (the visibility map)?
Figure 5: How were the outliers identified, and how was the lower limit chosen, especially for the coherence, such as in Figure 5d? The median and quartile changes with R, T, A are quite reasonable, but I've noticed that there are many outliers close to your lower limit. Please explain how the lower limit was determined and why there are so many outliers, not just in Figure 5d. Also, why not calculate and present the “mean velocity,” “variance of velocity,” and “velocity outside delta (∆)” plots like the coherence panel?
Figure 5h: Is a velocity threshold (0.02 m/yr) being applied?
Figure 6f: Although the relationship may not be immediately apparent, I have observed that many rock glaciers are frequently located in convergent areas. This pattern is intriguing, and any insights or explanation regarding this observation would be valuable.
Figure 11b: Would replacing the LOS velocity values with slope velocity result in a smoother transition from red to blue?
Citation: https://doi.org/10.5194/egusphere-2024-1511-RC1 - AC1: 'Reply on RC1', Chiara Crippa, 02 Dec 2024
-
RC2: 'Comment on egusphere-2024-1511', Anonymous Referee #2, 24 Feb 2025
Title: Optimizing rock glaciers activity classification in South Tyrol (North-East Italy): integrating multisource data with statistical modelling
1) Overall quality and general comments
Rock glaciers are key indicators of permafrost in alpine regions, formed by a seasonally frozen detrital layer overlying supersaturated debris of ice or pure ice, and characterized by gravity flow. Their distribution is influenced by topographic and climatic factors at different scales, and they play a crucial role in high-altitude hydrology by storing ice and water. Traditionally, rock glaciers are classified as active, inactive, or relict based on ice content and movement. However, rising permafrost temperatures have led to an accelerating trend, encouraging an updated classification that considers sediment transport efficiency. In the regional territory of South Tyrol, two rock glaciers activity classifications coexist (Autonomous Province of Bolzano/Bozen and Bertone et al., 2019). By combining geomorphological characteristics, climatic driving factors, and InSAR products, the authors develop a statistical model to refine the classification of rock glaciers.
This study represents an innovative contribution since it integrates multiple variables into a multiclass generalised additive mixing (GAM) model to predict rock glacier activity. Using remote sensing, ground-based data, and digital terrain models, the workflow involves extracting velocity and environmental attributes at a regional scale, calibrating and validating a multiclass predictive GAM, and applying it to classify landforms based on their activity status.
The integration of remote sensing data and statistical modelling significantly advances current methods for assessing rock glacier dynamics. The study is well-structured, with a clear research objective and methodology. The statistical approach, particularly the use of a multiclass GAM model, is effective for the research aims. The discussion is robust, highlighting both its contributions and its limitations. The figures and tables are clear, informative, and support the understanding of the concepts. Finally, this work advances the understanding of rock glacier dynamics by refining their classification system and linking their activity states to a range of predictor variables.
2) Individual scientific questions
3.4.1. Statistical modelling
How did the authors ensure the robustness of the GAM model in terms of the selection and evaluation of predictor variables?
4.1. Exploratory Data Analysis
How did you decide which variables to retain for further analysis, and why were some variables, such as elevation, excluded to avoid redundancy despite their high discriminatory power? Could you clarify the rationale behind this choice?
3) Specific comments on the manuscript
Line 119: How many rock glaciers are present in the analysed dataset?
Line 127: The classification 'n.d.' is unclear. Could you please clarify its meaning and usage in this context?
Line 148: Could you explain in more detail how the variables were extracted and assigned to each individual rock glaciers?
Lines 207-210: "Using this SCD parameter, a potential correlation between the rock glaciers’ activity at a regional level was made[...]" Could you explain this statement more clearly? How was the correlation assessed, and what were the main findings regarding the SCD in relation to the rock glaciers' activity?
Figure 4: Does the term "look vector" refer to the Line of Sight (LOS) of the satellites? Could you also better explain if the shadowing and layover effects part is the C index analysis?
Figure 4: Is the vLOS referring to vertical velocity? Additionally, could you adjust the colour scale bar to range between -8 and 8 cm/year to improve the clarity of the data representation?
Lines 244-248: “For each rock glacier polygon, mean values for environmental and climatic variables were assigned based on the values within the polygon boundary. Furthermore, for DInSAR-related variables (i.e., velocity and coherence), additional statistical descriptors […]”. Can you explain how the uncertainty was computed for each rock glacier, based on the SAR data coverage? How did you assess the spatial uncertainty within each polygon?
Line 243: "Starting from the distribution map of the rock glaciers and considering their displacement range, we made two distinctions [...]". Could you clarify the rationale behind the choice of a 100-meter buffer around each mapped landform? How was this distance determined, and how does it affect the classification?
Lines 264-266: “To discern the key factors influencing the distinction between A, R, and T rock glacier classes, we performed an initial Exploratory Data Analysis. This exploration served […]”. Could you provide more details on how this exploratory analysis was performed, and how it helped with the model?
Lines 167-272: “GAM was employed to investigate the associations between the chosen predictor variables derived from both environmental and DInSAR datasets and the response variables. GAM provides […]”. Could you provide more explanation on the use of GAM in this context? A brief discussion of the relevant literature and how GAM has been applied in other studies would strengthen this section.
-
AC2: 'Reply on RC2', Chiara Crippa, 03 Mar 2025
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-1511/egusphere-2024-1511-AC2-supplement.pdf
-
AC2: 'Reply on RC2', Chiara Crippa, 03 Mar 2025
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 558 | 165 | 131 | 854 | 55 | 25 | 43 |
- HTML: 558
- PDF: 165
- XML: 131
- Total: 854
- Supplement: 55
- BibTeX: 25
- EndNote: 43
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Cited
Stefan Steger
Giovanni Cuozzo
Francesca Bearzot
Volkmar Mair
Claudia Notarnicola
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2073 KB) - Metadata XML
-
Supplement
(584 KB) - BibTeX
- EndNote
- Final revised paper