Construction of a Joint Newmark–Runout Model for Seismic Landslide Risk Identification: A Case Study in the Eastern Tibetan Plateau
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThis article takes the eastern region of the Tibetan Plateau as an example to study seismic landslide risk identification using a joint Newmark-Runout model. Initially, by conducting geological data surveys and collection in the target area, the development characteristics and distribution patterns of seismic landslides in the region are understood. Then, through static and dynamic seismic landslide hazards, the optimized seismic landslide hazard is derived. After optimization, the seismic landslide risk identification is finally obtained. However, in the process of identifying seismic landslide hazards, there is a lack of introduction to the calculation methods and insufficient efforts to reduce the impact of computational errors on the results. I recommend accepting the manuscript for publication after a major revision. The specific comments that need addressing are outlined below:
1. The figures in the article need clearer labeling. For instance, the layout and numbering of Fig.1b and Fig.1c should be revised; the title layout in Table 1 needs adjustment, ensuring the title is aligned with the table and consistent with the formatting of subsequent tables; the unit for effective cohesion in the table should also be corrected. Additionally, the title of Fig.8 should be adjusted to avoid spanning pages. These changes will enhance the clarity of the article and help readers quickly grasp its overall structure.
2. In Section 3.1.1, paragraph 1, lines 14 to 15, the critical acceleration for the study area is determined using the physical parameters and related data of the geological units. A few questions arise: How can the error in the calculated critical acceleration be minimized to fall within an acceptable range? And is the calculated critical acceleration for the entire study area sufficiently accurate?
3. In Section 3.1.1, paragraph 4, lines 5 to 6, during the process of adjusting the equation coefficients based on the statistical relationship between seismic slope displacement and landslide probability in the Luding earthquake zone, the second step— “calculate regional area and landslide area within each seismic slope displacement interval and their ratio” —may involve errors. It is important to account for how these errors might affect the final probability results.
4. In line 12 of paragraph 4 in Section 3.1.1, it states, “The new corrected functional equation was used to calculate the landslide occurrence probability in the Luding earthquake area, then which was divided into 5 levels using natural discontinuity method,” However, the analysis process and results of the natural discontinuity method were not adequately explained. The seismic landslide hazard was simply categorized into five levels without sufficient justification, making the result less convincing. The methodology should be properly explained to strengthen the validity of the findings.
5. In Section 3.3.2, line 11 of the first paragraph, the simulation trajectory line is mentioned, but the process and results are not explained. Providing a clear explanation of the simulation process and presenting the results for the trajectory line would help readers better understand and appreciate this part of the content.
6. In line 12 of paragraph 1 in Section 3.3.2, where “the spatial frequency of moving rock masses was further calculated” is mentioned, there are various methods available for this calculation. It is important to clarify which specific method was used, as without this explanation, the classification of results to represent the dynamic hazard of earthquake-induced landslides lacks a solid theoretical foundation.
7. In Section 3.3.3, paragraph 1, line 3, it is mentioned that “the static and dynamic seismic landslide hazard index were normalized.” Since there are various methods of normalization, it would be helpful to clearly explain the specific approach used in this study to ensure that the results are both accurate and convincing.
8. In Section 3.3.3, paragraph 1, lines 3 to 4, the earthquake landslide hazard index is derived by combining the normalized indicators. However, the method used for combining these indicators and the fusion process itself should be clearly explained to ensure that the final comprehensive earthquake landslide hazard index is both reliable and credible.
9. In the first line of the third paragraph of Section 3.3.3, it is stated that “Taking the Luding earthquake-induced landslides as the validation samples, the relationship between the seismic landslide hazard and the landslide spatial distribution was statistically analyzed.” It would be helpful to clarify whether the sample size is sufficiently large to make the results more generalizable.
10. In the first and second lines of the second paragraph of Part 4, it is mentioned that “Based on the optimized seismic landslide hazard evaluation results, the linear road data in the Luding earthquake area were overlaid to obtain the seismic landslide risks facing roads.” It would be important to clarify how the accuracy of the linear road data samples and the superimposed data are ensured.
11. In Section 3.3.2, from the first to the third sentence of the first paragraph, “The grid elements with very high and high static seismic landslide hazard were extracted as landslide source points, which are most likely to have instability and long runout.” may be modified to: “Grid elements with very high and high static seismic landslide hazards were extracted as potential landslide source points, as they are the most prone to instability and long runout distances.”
Comments on the Quality of English LanguageMinor editing of English language required.
Author Response
Comment 1: The figures in the article need clearer labeling. For instance, the layout and numbering of Fig.1b and Fig.1c should be revised; the title layout in Table 1 needs adjustment, ensuring the title is aligned with the table and consistent with the formatting of subsequent tables; the unit for effective cohesion in the table should also be corrected. Additionally, the title of Fig.8 should be adjusted to avoid spanning pages. These changes will enhance the clarity of the article and help readers quickly grasp its overall structure.
Response 1: Thank you for pointing this out. We have revised some figures to make them clearer. The layout and numbering of Fig.1b and Fig.1c in Figure 2 are revised as Figure 2b and Figure 2c in Figure 2, respectively. The title layout in Table 1 is correct, with a 4.6 cm indent on the left side, so no modification is needed. The unit for effective internal cohesion in table 1 is revised as kPa, and the unit for weight of rock masses is revised as kN/m3. The title of Figure 8 is adjusted to avoid spanning pages. The numbering of Fig. 15 in Figure 14 is revised as Figure 16 in Figure 15. The numbering of Fig. 4e, Fig. 4f and Fig. 16 are revised as Figure 4e, Figure 4f, and Figure 17 in Figure 16, respectively.
Comment 2: In Section 3.1.1, paragraph 1, lines 14 to 15, the critical acceleration for the study area is determined using the physical parameters and related data of the geological units. A few questions arise: How can the error in the calculated critical acceleration be minimized to fall within an acceptable range? And is the calculated critical acceleration for the entire study area sufficiently accurate?
Response 2: Agree. The slope critical acceleration can be derived from the slope static safety factor, and its error is mainly determined by the slope static safety factor. So, we increase the calculation process of slope static safety factor and the adjustment process of the initial parameters of engineering geological units. Compared with previous research results, the slope critical acceleration is ensured within a reasonable range. This change can be found in page 8, lines 261-270.
Updated text in the manuscript:
Firstly, the calculation of the slope static safety factor in the Luding earthquake area was conducted. During the iterative calculation, the initial parameters of engineering geological units were adjusted to ensure that the slope static safety factor exceeds 1 without seismic dynamic loading [36], while maintaining a minimum parameter value of 1.01 [26]. The final assigned parameters are shown in Table 1. Subsequently, the slope critical acceleration in the Luding earthquake area can be calculated successively (Figure 7). When compared with previous research findings, the resulting value range is deemed reasonable [26,36,40]. Regions characterized by lower slope critical acceleration exhibit a higher susceptibility to landslide occurrences.
Comment 3: In Section 3.1.1, paragraph 4, lines 5 to 6, during the process of adjusting the equation coefficients based on the statistical relationship between seismic slope displacement and landslide probability in the Luding earthquake zone, the second step—“calculate regional area and landslide area within each seismic slope displacement interval and their ratio”—may involve errors. It is important to account for how these errors might affect the final probability results.
Response 3: Agree. We have revised the second step of the process of adjusting the equation coefficients to emphasize this point. This change can be found in page 11, lines 300 to 306.
Updated text in the manuscript:
(2) calculate the regional area and landslide area within each seismic slope displacement interval, along with their ratio, which serves as an indicator of landslide occurrence probability on the vertical axis (Figure 9). Here, the calculation of the coverage area is based on the 30 m raster data, which cannot express the fine details of the land-slide distribution, so the calculated landslide area may have a reduced error, which leads to the conservative result of the final landslide occurrence probability
Comment 4: In line 12 of paragraph 4 in Section 3.1.1, it states, “The new corrected functional equation was used to calculate the landslide occurrence probability in the Luding earthquake area, then which was divided into 5 levels using natural breakpoint method,” However, the analysis process and results of the natural breakpoint method were not adequately explained. The seismic landslide hazard was simply categorized into five levels without sufficient justification, making the result less convincing. The methodology should be properly explained to strengthen the validity of the findings.
Response 4: Agree. We have increased the analysis results of the natural breakpoint method. The analysis process of the natural breakpoint method can be established in the ArcGIS platform, and the relevant operation process is mature and is detailed in the software operation manual. This change can be found in page 11, lines 308 to 313. The seismic landslide hazard can be divided into 3, 4 and 5 levels, but the 5-level classification scheme is the most used. So, we also adopt the 5-level classification scheme.
Updated text in the manuscript:
The newly corrected functional equation (Equation 6) was utilized to calculate the landslide occurrence probability within the Luding earthquake area, which was sub-sequently classified into five categories (≥20%, 15%-20%, 5%-15%, 1%-5%, <1%) using the natural breakpoint method on the ArcGIS platform, thereby representing static seismic landslide hazard levels as very high, high, moderate, low, and very low (Figure 10).
Comment 5: In Section 3.3.2, line 11 of the first paragraph, the simulation trajectory line is mentioned, but the process and results are not explained. Providing a clear explanation of the simulation process and presenting the results for the trajectory line would help readers better understand and appreciate this part of the content.
Response 5: Agree. We have revised and increased the simulation process of trajectory lines of rock mass movement after seismic landslide occurrence using Rockfall model. And, the results for the trajectory lines have been shown in Figure 11. These changes can be found in page 12, lines 332 to 335, page 13, lines 346-348.
Updated text in the manuscript:
Based on the above parameters and Rockfall model, the trajectory lines of rock mass movement after seismic landslide occurrence were calculated [Figure 11]. Detailed instructions for Rockfall model can be found in software manuals and related references [28,29].
Figure 11. The spatial distribution of the trajectory lines of rock mass movement after seismic landslide occurrence within the Luding earthquake area.
Comment 6: In line 12 of paragraph 1 in Section 3.3.2, where “the spatial frequency of moving rock masses was further calculated” is mentioned, there are various methods available for this calculation. It is important to clarify which specific method was used, as without this explanation, the classification of results to represent the dynamic hazard of earthquake-induced landslides lacks a solid theoretical foundation.
Response 6: Agree. We have revised and increased the line density method in the ArcGIS platform to calculate the spatial frequency of moving rock masses, and explain their meaning. These changes can be found in page 12, lines 336 to 340.
Updated text in the manuscript:
Based on the trajectory lines of rock mass movement, the spatial frequency of moving rock masses can be calculated using the line density method on the ArcGIS platform. The more the trajectory lines that pass through a certain location, the higher the spatial frequency of moving rock masses at this location, indicating that the land-slide hazard at this location is greater.
Comment 7: In Section 3.3.3, paragraph 1, line 3, it is mentioned that “the static and dynamic seismic landslide hazard index were normalized.” Since there are various methods of normalization, it would be helpful to clearly explain the specific approach used in this study to ensure that the results are both accurate and convincing.
Response 7: Agree. We have revised and increased the Min-Max normalization method to normalize the static and dynamic seismic landslide hazard index. This changes can be found in page 14, lines 354 to 355.
Updated text in the manuscript:
Firstly, the static and dynamic seismic landslide hazard index were normalized using the Min-Max normalization method (Equation 7).
Comment 8: In Section 3.3.3, paragraph 1, lines 3 to 4, the earthquake landslide hazard index is derived by combining the normalized indicators. However, the method used for combining these indicators and the fusion process itself should be clearly explained to ensure that the final comprehensive earthquake landslide hazard index is both reliable and credible.
Response 8: Agree. We have revised and increased the factor overlay method via the ArcGIS raster calculator to fuse the two normalized indexes. These changes can be found in page 14, lines 355 to 358.
Updated text in the manuscript:
Subsequently, the two normalized indexes are fused together to derive a comprehensive seismic landslide hazard index using the factor overlay method via the ArcGIS raster calculator. The weights of both layers are set to 0.5.
Comment 9: In the first line of the third paragraph of Section 3.3.3, it is stated that “Taking the Luding earthquake-induced landslides as the validation samples, the relationship between the seismic landslide hazard and the landslide spatial distribution was statistically analyzed.” It would be helpful to clarify whether the sample size is sufficiently large to make the results more generalizable.
Response 9: Agree. We have increased the clarification of seismic landslide samples. This change can be found in page 14-15, lines 374 to 379.
Updated text in the manuscript:
No matter how detailed the landslide interpretation and investigation, it cannot fully represent the actual landslide distribution. So far, the landslide inventory we have adopted for the Luding earthquake is the most comprehensive, with 8,685 landslides. Statistical analysis shows that the landslide distribution characteristics conform to the general laws of seismic landslides [21], so it is sufficient as a validation sample for landslide hazard.
Comment 10: In the first and second lines of the second paragraph of Part 4, it is mentioned that “Based on the optimized seismic landslide hazard evaluation results, the linear road data in the Luding earthquake area were overlaid to obtain the seismic landslide risks facing roads.” It would be important to clarify how the accuracy of the linear road data samples and the superimposed data are ensured.
Response 10: Agree. We have increased the clarification of the road lines and building locations in the Luding earthquake area. These changes can be found in page 15, lines 401 to 405.
Updated text in the manuscript:
Using remote sensing interpretation and remote sensing image data from Google Earth, the road lines and building locations in the Luding earthquake area are identified. The linear road data includes both surface roads and tunnels. Tunnel interiors are not threatened by seismic landslides, so they are not included in the seismic landslide risk calculation.
Comment 11: In Section 3.3.2, from the first to the third sentence of the first paragraph, “The grid elements with very high and high static seismic landslide hazard were extracted as landslide source points, which are most likely to have instability and long runout.” may be modified to: “Grid elements with very high and high static seismic landslide hazards were extracted as potential landslide source points, as they are the most prone to instability and long runout distances.”
Response 11: Thanks. We have made changes to the relevant content to emphasize this point. These changes can be found in page 12, lines 322 to 324.
Updated text in the manuscript:
Grid elements with very high and high static seismic landslide hazard were extracted as potential landslide source points, as they are the most prone to instability and long runout distance.
Reviewer 2 Report
Comments and Suggestions for AuthorsIn this study, a joint Newmark-Runout model based for seismic landslide dynamics is proposed, fromt he eastern Tibetan Plateau. The paper is written very carefully and it is a well written paper. The reviewer recommend this paper for acceptance with the following change.
1. Figure 6 and Table1:
Please explain the basis for the geological classification in detail. Also, how did you obtain the physical properties in Table 1?
2. Please clearly explain the basis for setting the five-level risk classification presented in this paper. Are the risks shown in Figures 10, 11, 12, 14, and 15 comparable?
3. This is a case study, is there universality in this research?
Show that this study is universal.
Author Response
Comment 1. Figure 6 and Table1: Please explain the basis for the geological classification in detail. Also, how did you obtain the physical properties in Table 1?
Response 1: Thanks. We have revised and increased the clarification of the basis for the geological classification and the method of parameter assignment of engineering geological units. These changes can be found in page 7, lines 242 to 246, in page 8, lines 254 to 256, in page 8, lines 262 to 266.
Updated text in the manuscript:
The engineering geological units within the study area were classified into 12 categories (Figure 6, Table 1), each exhibiting varying susceptibility to landslide occurrence. The geological strata with similar geological ages, lithology, geological structures, and soil and rock properties are grouped into an engineering geological unit.
Based on the Engineering Geology Handbook of China [39], Standard for Engineering Classification of Rock Mass of China (GB/T 50218-2014) and relevant literatures [40], the attribute parameters of engineering geological units can be preliminarily assigned.
During the iterative calculation, the initial parameters of engineering geological units were adjusted to ensure that the slope static safety factor exceeds 1 without seismic dynamic loading [36], while maintaining a minimum parameter value of 1.01 [26]. The final assigned parameters are shown in Table 1.
Comment 2. Please clearly explain the basis for setting the five-level risk classification presented in this paper. Are the risks shown in Figures 10, 11, 12, 14, and 15 comparable?
Response 2: Thanks. The seismic landslide hazard or seismic landslide risk can be divided into 3, 4 and 5 levels, but the 5-level classification scheme is the most used. So, we also adopt the 5-level classification scheme. The classification method adopted is the natural breakpoint method on the ArcGIS platform. The landslide hazard shown in new Figures 10, 12, and 13 is comparable. Figure 13 is the result of the fusion of Figures 10 and 12, which improves the accuracy of landslide hazard evaluation. The landslide risk shown in new Figures 15 and 16 is not comparable. Figure 16 is a part of Figure 15 and shows the details of a specific area.
Comment 3. This is a case study, is there universality in this research? Show that this study is universal.
Response 3: Thanks. Yes, this research has certain universal significance in the seismic landslide risk identification. These changes can be found in page 18, lines 474 to 480.
Updated text in the manuscript:
In mountainous and gorge complex geomorphological areas, landslides triggered by earthquakes not only cause damage in the source zone, but also often cause greater damage in the sliding zone and debris accumulation zone. Therefore, the movement process of landslides should be considered as an important factor in the seismic landslide hazard evaluation and seismic landslide risk identification. The proposed joint Newmark-Runout model and research approach have certain universal significance in the seismic landslide risk identification in similar areas.
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsThis article is logically well-structured. It begins with a detailed investigation and collection of address data in the study area, followed by an enumeration and classification of the developmental characteristics and distribution patterns of seismic landslides in the region. Through the analysis of both static and dynamic seismic landslide hazards, the seismic landslide risk is optimized, ultimately leading to the identification of seismic landslide risks. The content is thoroughly explained, with detailed clarification provided for the expression of formulas and the application of methods. The conclusion holds significant practical value, offering a more comprehensive understanding of the identification of landslide risks triggered by earthquakes. However, there are some shortcomings in the article, such as the layout of the images and their captions. The specific methods and the accuracy of certain processed data lack sufficient explanation. After minor revisions, if these issues can be resolved, the quality of the article will be further enhanced, making it more likely to be accepted by the journal.
1. The images in the article need to be clearly labeled. For the progressive relationship shown in Figure 2, it is recommended to use an alternative method of expression, as the current format may confuse the readers. In Figure 17, the images are missing (a) and (b), although (a) appears in the figure caption. This needs to be corrected to ensure consistency.
2. In Section 3.1.1, paragraph 5, line 12, “The newly corrected functional equation (Equation 6) was utilized to calculate the landslide occurrence probability within the Luding earthquake area, which was subsequently classified into five categories (≥20%, 15%-20%, 5%-15%, 1%-5%,” It only introduces the use of the natural breaks method on the ArcGIS platform. Further elaboration on the analysis process of the natural breaks method (either through text or images) is needed to make this section more comprehensive.
3. In Part 4, paragraph 1, lines 2 to 6, “Using remote sensing interpretation and remote sensing image data from Google Earth, the road lines and building locations in the Luding earthquake area are identified. The linear road data includes both surface roads and tunnels. Tunnel interiors are not threatened by seismic landslides, so they are not included in the seismic landslide risk calculation.” Although the accuracy of linear road data is explained, there is a lack of discussion regarding the accuracy after overlaying the linear road data. This part requires further elaboration.
4. In Part 4, paragraph 2, lines 9 to 11, “Conversely, the buildings located at the slope foot along both banks of the Duda River face a heightened seismic landslide risk, which warrants careful consideration.” may be modified to: “Conversely, the buildings situated at the foot of the slopes along both banks of the Duda River are exposed to an elevated risk of seismic-induced landslides, necessitating thorough and careful consideration.”
Comments on the Quality of English LanguageIn Part 4, paragraph 2, lines 9 to 11, “Conversely, the buildings located at the slope foot along both banks of the Duda River face a heightened seismic landslide risk, which warrants careful consideration.” may be modified to: “Conversely, the buildings situated at the foot of the slopes along both banks of the Duda River are exposed to an elevated risk of seismic-induced landslides, necessitating thorough and careful consideration.”
Author Response
Comment 1: This article is logically well-structured. It begins with a detailed investigation and collection of address data in the study area, followed by an enumeration and classification of the developmental characteristics and distribution patterns of seismic landslides in the region. Through the analysis of both static and dynamic seismic landslide hazards, the seismic landslide risk is optimized, ultimately leading to the identification of seismic landslide risks. The content is thoroughly explained, with detailed clarification provided for the expression of formulas and the application of methods. The conclusion holds significant practical value, offering a more comprehensive understanding of the identification of landslide risks triggered by earthquakes. However, there are some shortcomings in the article, such as the layout of the images and their captions. The specific methods and the accuracy of certain processed data lack sufficient explanation. After minor revisions, if these issues can be resolved, the quality of the article will be further enhanced, making it more likely to be accepted by the journal.
Response 1: Thank you for pointing this out. We agree with this comment. We have revised the layout of the images and their captions, including Figure 2 and Figure 17. We have added some explanation for the specific methods and the accuracy of certain processed data. Detailed revision instructions will be provided in subsequent responses.
Comment 2: The images in the article need to be clearly labeled. For the progressive relationship shown in Figure 2, it is recommended to use an alternative method of expression, as the current format may confuse the readers. In Figure 17, the images are missing (a) and (b), although (a) appears in the figure caption. This needs to be corrected to ensure consistency.
Response 2: We agree with this comment. The progressive relationship shown in Figure 2 have been revised using an alternative method of expression, and Figure 2a has been replaced with a world map. This change can be found in page 5. In Figure 17, we have added (a) and (b) and revised the caption. This change can be found in page 18, lines 462-464.
Updated text in the manuscript:
Figure 17. The representative Luding earthquake-induced destructive landslides. (a) Landslide block road. (b) Landslide damage road protection shed.
Comment 3: In Section 3.1.1, paragraph 5, line 12, “The newly corrected functional equation (Equation 6) was utilized to calculate the landslide occurrence probability within the Luding earthquake area, which was subsequently classified into five categories (≥20%, 15%-20%, 5%-15%, 1%-5%,” It only introduces the use of the natural breaks method on the ArcGIS platform. Further elaboration on the analysis process of the natural breaks method (either through text or images) is needed to make this section more comprehensive.
Response 3: We agree with this comment. We have further elaborated the analysis process of landslide occurrence probability classification using natural breakpoint method. This change can be found in page 11-12, lines 314-324.
Updated text in the manuscript:
Equation 6 indicates that the landslide occurrence probability ranges from 0 to 0.2698. There are no clear standards or regulations for landslide hazard classification indica-tors. Generally, a quantitative and qualitative approach is used to classify landslide hazard levels. The natural breakpoint method is a classification technique based on numerical distribution law, aimed at maximizing the differentiation between classes. Firstly, the landslide occurrence probability value was classified into five classes using the natural breakpoint method on the ArcGIS platform as 0.1937-0.2680, 0.1426-0.1937, 0.0573-0.1426, 0.0135-0.0573, and 0-0.0135. Subsequently, based on existing experience and regional landslide development characteristics, the landslide occurrence probability classes have been adjusted to ≥0.20, 0.15-0.20, 0.05-0.15, 0.01-0.05, and <0.01, there-by representing static seismic landslide hazard levels as very high, high, moderate, low, and very low (Figure 10).
Comment 4: In Part 4, paragraph 1, lines 2 to 6, “Using remote sensing interpretation and remote sensing image data from Google Earth, the road lines and building locations in the Luding earthquake area are identified. The linear road data includes both surface roads and tunnels. Tunnel interiors are not threatened by seismic landslides, so they are not included in the seismic landslide risk calculation.” Although the accuracy of linear road data is explained, there is a lack of discussion regarding the accuracy after overlaying the linear road data. This part requires further elaboration.
Response 4: We agree with this comment. We have added some discussion regarding the accuracy after overlaying the linear road data. This change can be found in page 15, lines 416-420.
Updated text in the manuscript:
The optimized seismic landslide hazard data is in grid format with a resolution of 30 m, and the road data is in linear vector format. Therefore, the seismic landslide risk posed to roads obtained by overlaying the grid and vector data has a precision of 30 m along the road direction.
Comment 5: In Part 4, paragraph 2, lines 9 to 11, “Conversely, the buildings located at the slope foot along both banks of the Duda River face a heightened seismic landslide risk, which warrants careful consideration.” may be modified to: “Conversely, the buildings situated at the foot of the slopes along both banks of the Duda River are exposed to an elevated risk of seismic-induced landslides, necessitating thorough and careful consideration.”
Response 5: We agree with this comment. We have modified these sentences. This change can be found in page 17, lines 444-446, page 19, lines 512-513.
Updated text in the manuscript:
Conversely, the buildings situated at the foot of the slopes along both banks of the Du-da River are exposed to an elevated risk of seismic landslides, necessitating thorough and careful consideration.
while those situated at the foot of the slopes along both banks of the Duda River are exposed to an elevated seismic landslide risk.
