the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 11 Jul 2025
Early Identification of Reservoir-Bank Landslides in Deeply Incised Mountain Canyon Areas with Interferometric Baseline Optimization
Abstract. The complex geological conditions in deeply incised mountainous canyon areas make reservoir-bank landslides a frequent hazard. Accurate interferogram selection and baseline network configuration are crucial for SBAS-InSAR-based landslide monitoring, yet are severely challenged by seasonal vegetation decorrelation. To overcome this limitation, this study proposes a novel Vegetation-Adaptive WCTM that integrates time-series vegetation dynamics into interferometric baseline optimization. This approach establishes a vegetation–coherence coupling model to dynamically adjust coherence thresholds based on quantified vegetation coverage levels and synergizes ERA5 meteorological data with tropospheric delay modeling for atmospheric correction. The results demonstrate significant advancements:（1）The deformation rate standard deviation is reduced by 0.520 and 0.192 compared to traditional short-temporal baseline and average coherence threshold methods, respectively, corresponding to a 29.1 % improvement (1.2668 vs. 1.7865).（2）140,146 additional valid phase-unwrapping points were obtained, indicating substantially improved interferometric processing quality.（3）39 landslides were successfully identified, representing a 22 % increase compared to conventional methods (32 landslides), with 7 new high-risk sites discovered even during low-coherence vegetation seasons. Based on field verification with drone surveys, typical landslides were selected to analyze their spatial distribution and temporal evolution patterns, demonstrating the applicability of the method in deeply incised mountainous canyon areas. These findings provide theoretical and technical support for regional disaster prevention and mitigation efforts.
- Preprint
(3536 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- AC1: 'Comment on egusphere-2025-3054-On the Transferability of the Vegetation-Adaptive WCTM Method to Diverse Environments and Sensor Configurations', Hong Wenyu, 23 Aug 2025
-
CC1: 'Comment on egusphere-2025-3054', yinzhu long, 05 Sep 2025
The vegetation-adaptive WCTM method proposed in this study provides an important innovation for the engineering application of SBAS-InSAR technology in areas with complex vegetation cover. By ingeniously combining the characteristics of temporal vegetation variations with interferometric coherence analysis, this method dynamically adjusts the coherence threshold, thereby significantly alleviating seasonal decorrelation problems. It demonstrates strong theoretical significance and practical value. In the case study of the Baihetan Reservoir area, the method not only achieved a 29.1% improvement in deformation accuracy but also successfully identified 32 historical landslides and 7 newly emerging landslide hazards, fully proving its effectiveness and reliability in complex environments.
Notably, this study adopts a multi-source data collaborative analysis approach, integrating Sentinel-2 vegetation indices, ERA5 atmospheric reanalysis data, and UAV validation data to establish a complete technical workflow. In particular, the use of the HyP3 cloud computing platform to process large-scale InSAR data highlights the advantages of efficient processing in modern remote sensing technology and provides a practical engineering reference for similar studies.
For further improvement, the following aspects may be considered: first, providing a more detailed theoretical basis or sensitivity analysis for the classification criteria of vegetation cover levels would help enhance the universality of the method; second, under complex terrain conditions, atmospheric delay correction still has room for improvement, and incorporating more meteorological data sources may further increase accuracy.
Citation: https://doi.org/10.5194/egusphere-2025-3054-CC1 -
AC2: 'Reply on CC1', Hong Wenyu, 06 Sep 2025
Respected Professor Long Yinzhu,
Hello! Thank you very much for taking the time out of your busy schedule to carefully review our paper " Early Identification of Reservoir-Bank Landslides in Deeply Incised Mountain Canyon Areas with Interferometric Baseline Optimization" and for providing such insightful and constructive comments. Your affirmation of the research value and innovation of this paper, as well as your precise understanding of the technical details, has greatly encouraged our entire team.
You specifically pointed out that the vegetation-adaptive WCTM method proposed in this study provides significant innovation for the engineering application of SBAS-InSAR technology in complex vegetation-covered areas, and you fully affirmed the value of the complete technical workflow we established by integrating multi-source data (Sentinel-2 vegetation indices, ERA5 reanalysis data, and UAV validation data). We feel truly honored that you so profoundly understand the core ideas and contributions of our work.
At the same time, the two improvement suggestions you proposed are highly insightful and have pointed the way for the deepening of our follow-up research:
1.) Regarding the theoretical basis and sensitivity analysis of the vegetation coverage classification standards: We completely agree with your views. The classification thresholds in this study were primarily set based on the statistical characteristics of vegetation coverage in the study area and seasonal variation patterns. Your suggestion is very valuable. On this basis, we will further explore the variation mechanisms of interferometric coherence under different vegetation coverage conditions, strengthen the interpretation of the theoretical basis for the classification standards, with the aim of establishing a more universal and robust optimization criterion.
2.) Regarding the integration of more meteorological data sources to improve atmospheric delay correction accuracy: The point you raised is crucial. In deeply incised canyon areas, the complex topography and spatiotemporal heterogeneity of atmospheric effects mean that relying solely on ERA5 reanalysis data, while effective in mitigating most errors, still leaves room for improvement. We plan in our next research phase to collect meteorological station data from around the study area, attempt to assimilate or fuse station observations with ERA-5 and other reanalysis data, to build a more localized and precise atmospheric delay correction model, thereby further reducing noise and enhancing monitoring accuracy in extremely complex terrain.
Your insightful comments have provided extremely valuable guidance for our research work. We have incorporated your suggestions into our future research plans. Once again, we extend our most sincere gratitude to you! Your affirmation is a tremendous encouragement to us, and your suggestions are the driving force for our continued progress.
We sincerely look forward to future opportunities to continue learning from you on related topics and to engage in more in-depth exchanges.
Wishing you success in your work and good health!
Sincerely,
Xi Wenfei
And on behalf of all authors
September 6, 2025
Citation: https://doi.org/10.5194/egusphere-2025-3054-AC2
-
AC2: 'Reply on CC1', Hong Wenyu, 06 Sep 2025
-
RC1: 'Comment on egusphere-2025-3054', Anonymous Referee #1, 18 Nov 2025
This manuscript establishes a vegetation–coherence coupling model to dynamically adjust coherence thresholds based on quantified vegetation coverage levels and synergizes ERA5 meteorological data with tropospheric delay modeling for atmospheric correction. The following issues are present in the manuscript:
1. To further clarify the innovative aspects of the manuscript, it is recommended that the authors explicitly list the contributions in the Introduction section in a bullet-point format.
2. In general, all variables and Greek letters should be in italic format, while constants should not be italicized. Vectors and matrices should be in bold format. Please ensure that the equations throughout the manuscript satisfy these formatting conventions.
3. Is the method proposed in the manuscript exclusively tailored to reservoir-bank landslides in deeply incised mountain canyon areas? Given that studies on "underwater color disparities" have already identified the color differences between underwater and terrestrial environments, it is necessary to clarify whether the proposed method can be extended to other scenarios.
4. In the Conclusion section, authors can add outlooks for future work.Citation: https://doi.org/10.5194/egusphere-2025-3054-RC1 -
AC3: 'Reply on RC1', Hong Wenyu, 20 Nov 2025
Dear Reviewer,
We sincerely appreciate your valuable comments and positive recognition of the innovativeness of our manuscript titled "Early Identification of Reservoir Landslides in Deeply Incised Alpine Gorge Areas Considering Interferometric Baseline Optimization." Your constructive suggestions have been instrumental in improving our work, and we have carefully revised the manuscript accordingly. Below we provide point-by-point responses to your comments:
Comment 1: Further clarify the innovativeness of the manuscript by explicitly listing the contributions in bullet points in the introduction.
Response: We thank you for this important suggestion. We have strengthened the final paragraph of the Introduction section by explicitly listing three main contributions of this study in bullet points:
(1) Proposed a vegetation-adaptive Weighted Coherence Threshold Method (WCTM) that establishes a vegetation-coherence coupling model to dynamically adjust coherence thresholds, effectively mitigating the impact of seasonal vegetation decorrelation on interferogram quality;
(2) Integrated ERA5 high-resolution meteorological reanalysis data with tropospheric delay modeling to significantly reduce atmospheric delay errors in complex terrain, thereby improving deformation inversion accuracy;
(3) Achieved high-precision early identification of reservoir landslides in the deeply incised alpine gorge area of the Baihetan reservoir region, demonstrating a 22% increase in identification rate compared to conventional methods and validating the method's applicability and effectiveness under extreme topographic conditions.
The modified text has been highlighted in red in the Introduction.Comment 2: Generally, all variables and Greek letters should be italicized, while constants should not. Vectors and matrices should be boldfaced. Please ensure all equations in the manuscript comply with these formatting standards.
Response: We sincerely apologize for this oversight and thank you for bringing it to our attention. In the revised manuscript, we have systematically reviewed and corrected all mathematical equations, variables, and Greek letters throughout the text. We have ensured that all variables and Greek letters are italicized, constants remain upright, and all vector and matrix symbols are boldfaced, bringing them into full compliance with standard academic publishing format requirements.
Comment 3: Is the method proposed in the manuscript specifically targeted at reservoir bank landslides in deeply incised valleys? Given that "underwater color difference" studies have identified color variations between underwater and terrestrial environments, it is necessary to clarify whether this method can be generalized to other scenarios.
Response: We thank you for this insightful comment. While this study indeed uses reservoir bank landslides in deeply incised valleys as both the application context and validation scenario due to its representative and challenging nature, the core innovation of our method addresses the universal challenge of "radar interferometric decorrelation caused by seasonal vegetation changes." Therefore, its application potential extends beyond reservoir landslides. As you astutely observed, the technical framework (vegetation dynamics-aware interferometric optimization + atmospheric correction) possesses good generalizability and can be extended to other vegetation-affected scenarios such as monitoring of geological hazards in mountainous areas (e.g., landslides, collapses) and stability monitoring of engineering structures in areas with seasonal vegetation variations. We have added corresponding clarification in the Discussion section (end of Section 5.1.3), with the additions highlighted in red.
Comment 4: The authors may add future work prospects in the Conclusion section.
Response: We thank you for this suggestion. We fully agree that this enhances the completeness and forward-looking nature of our research. We have added a separate paragraph at the end of the Conclusion section (Section 5.2) outlining prospects for future work, primarily including: further optimization of the vegetation-coherence model, application of the method to more diverse geographical environments for validation, and exploration of its integration with emerging technologies such as machine learning. These additions have been highlighted in red in the manuscript.
Once again, we deeply appreciate the time and insightful guidance you have provided, which has substantially improved the quality of our manuscript. We hope the revised version meets with your approval and look forward to your further review.
Sincerely,
Xiwenfei (on behalf of all authors)
Corresponding Author Email: wenfeixi@ynnu.edu.cn
Date: November 20 2025Citation: https://doi.org/10.5194/egusphere-2025-3054-AC3
-
AC3: 'Reply on RC1', Hong Wenyu, 20 Nov 2025
-
RC2: 'Comment on egusphere-2025-3054', Anonymous Referee #2, 18 Nov 2025
In response to the problem of SBAS-InSAR accuracy limitation caused by vegetation seasonal decorrelation in the early identification of deep canyon area bank slope, the authors proposed a weighted coherence threshold method （WCTM） integrating the time series vegetation coverage, which provided a new technical reference for geological disaster monitoring in mountainous areas with vegetation activity. With good innovation, this research is very interesting. I recommend acceptance after minor modification. The review comments are as follows:
- Although the limitations of existing methods are mentioned in the introduction, it is necessary to further refine the points of innovation and reflect the importance of research question.
- The issue of the basis for vegetation coverage classification thresholds: The paper directly gives the three key vegetation coverage classification thresholds of 45%, 6%, and 75%, but does not elaborate on the scientific basis for their determination. The authors are requested to supplement the explanation: What standards (such as industry standards previous studies) or data-driven analyses (such as natural breakpoint method, cluster analysis) are these thresholds based on?
- It is suggested to add a clear and overall technical route map at the beginning of Chapter III to help readers quickly understand the process.
- Some of the result description paragraphs are slightly lengthy, repeating the information that has clearly presented in the figure. It is suggested to summarize and generalize the accuracy of the results.
Citation: https://doi.org/10.5194/egusphere-2025-3054-RC2 -
AC4: 'Reply on RC2', Hong Wenyu, 20 Nov 2025
Dear Reviewer,
We sincerely thank you for your valuable comments and positive feedback on our manuscript entitled "Early Identification of Reservoir Landslides in Deeply Incised Alpine Gorge Areas Considering Interferometric Baseline Optimization". Your constructive suggestions have been crucial in enhancing the quality of this paper. We have carefully addressed each of your comments and revised the manuscript accordingly. The specific modifications are detailed below:
Comment 1: Although the introduction mentions the limitations of existing methods, it is necessary to further elaborate on the innovations and reflect the importance of the research problem.
Response: We greatly appreciate this suggestion. We fully agree with your perspective and have strengthened the final paragraph of the introduction. The revised text not only outlines the research methodology but also explicitly highlights the three core innovations and their significance in bullet points: 1) proposing a vegetation-adaptive WCTM method to address seasonal decorrelation challenges; 2) developing a multi-source data synergistic error correction scheme; and 3) validating the method's high-precision identification capability in extremely complex terrain. This better emphasizes the breakthrough nature and practical application value of our study compared to existing techniques. The modified content is highlighted in red in the introduction.
Comment 2: Issue regarding the basis for vegetation coverage classification thresholds: The paper directly presents three key vegetation coverage classification thresholds: 45%, 60%, and 75%, without detailing their scientific basis. The authors are requested to supplement the explanation: On what standards (e.g., industry standards, previous research) or data-driven analyses (e.g., natural breaks method, cluster analysis) are these thresholds based?
Response: We thank you for pointing out this important omission. You are correct that the basis for determining these thresholds needs to be clearly stated. We adopted an approach that combines data-driven analysis with references to existing literature. In the revised manuscript, we have added corresponding explanations in Section 4.1 "WCTM Optimization of Interferometric Baseline Results", with the modifications highlighted in red.
Comment 3: It is recommended to add a clear and comprehensive technical flowchart at the beginning of Chapter 3 to help readers quickly understand the process.
Response: We greatly appreciate this valuable suggestion. We have added a clear technical flowchart (which can be newly numbered as Figure 3) at the beginning of Chapter 3 "Research Methods and Data Processing", accompanied by a brief textual description. This visually illustrates the complete workflow from data preparation and core methods (WCTM optimization and atmospheric correction) to result output (deformation acquisition and landslide identification).
Comment 4: Some result description paragraphs are slightly verbose and repeat information clearly presented in the figures. It is recommended to summarize and generalize the accuracy of the results.
Response: We thank you for this correction. We have streamlined and optimized the relevant descriptions in Chapter 4 "Results and Analysis", particularly regarding the content of Figures 8 and 9. In the revisions, we reduced direct restatements of information readily available from the figures. Instead, we strengthened the summarization of deformation patterns, the analysis of landslide dynamic evolutionary characteristics, and emphasized the exploration of relationships between deformation results and driving factors such as reservoir water level fluctuations and precipitation. This shifts the focus of the discussion from describing observations to interpreting patterns, enhancing the depth and generalizability of the analysis. Specific modifications can be found in the highlighted Section 4.3 of the manuscript.
Sincerely,
Wenyu Hong (on behalf of all authors)
Corresponding Author Email: wenfeixi@ynnu.edu.cn
Date: November 20 2025
Citation: https://doi.org/10.5194/egusphere-2025-3054-AC4
-
RC3: 'Comment on egusphere-2025-3054', Anonymous Referee #3, 03 Dec 2025
The manuscript addresses the challenge of landslide identification in complex geological conditions exacerbated by seasonal vegetation decorrelation. It proposes a vegetation-adaptive weighted coherence threshold method to optimize the interferometric baseline. The authors applied this approach to landslide identification and monitoring in the deep-incised canyon areas of southwestern China, demonstrating favorable application results. However, the manuscript presents the following issues, for which the authors are requested to provide clarifications and implement corresponding revisions. Therefore, a major revision of this manuscript is recommended.
- Overall, the manuscript's narrative is overly verbose. Taking the research methodology section as an example, the manuscript dedicates substantial space to listing conventional technical methods, while only briefly covering the approach proposed in this paper. This fails to highlight the advantages of the proposed method. It is recommended that the authors refine the manuscript to enhance its scientific rigor and readability.
- The authors have used 16 mm/yr as the threshold for identifying potential landslide hazards. Please discuss the rationale for this choice.
- There are many established methods for atmospheric correction in InSAR data processing. Please compare the effectiveness of these methods with the approach used in this study, which combines ERA5 meteorological data and tropospheric delay modeling for atmospheric correction.
- In Figure 6, the authors have marked landslide boundaries with red polygons, but there are no distinct InSAR deformation signals in these areas. Please explain this discrepancy.
- The reliability of the InSAR results in Figures 8(b) and 8(d) is questionable. It is difficult to determine whether the scattered pixels represent deformation, noise, or errors. Additionally, please explain the rationale for selecting the time-series characteristic points P1 and P2. Also, it is suggested that Figures 2 and 3 be moved to the appendix.
- The analysis of the research results in the manuscript does not correspond well with the figures, which may confuse readers. For example, the text states: "Additionally the deformation trends of Point 1 and Point 2 did not synchronize with the water level changes during the two water storage cycles, suggesting a time-lag effect of the landslide deformation in response to water level variations." It is recommended that the authors quantify this time-lag effect using signal processing methods.
Citation: https://doi.org/10.5194/egusphere-2025-3054-RC3 -
AC5: 'Reply on RC3', Hong Wenyu, 10 Jan 2026
Response to Reviewers
Dear Reviewer,
Thank you for taking the time to review our manuscript and for providing valuable feedback. Your comments are highly insightful and have provided essential guidance for improving the quality of the paper. We have carefully considered each point and will implement substantive revisions in the revised manuscript.
Comment 1: Overall, the narrative of the manuscript is too verbose. For instance, in the methodology section, significant space is devoted to listing traditional techniques, while the method proposed in this study is only briefly introduced. This fails to highlight the advantages of the proposed method. The authors are advised to refine the manuscript to enhance its scientific rigor and readability.
Response: Thank you for this suggestion. We have restructured and streamlined the methodology section accordingly. In the revised manuscript, we have condensed the background introduction to traditional InSAR interferogram selection methods, retaining only parts directly relevant to our study. Simultaneously, we have provided a more detailed explanation of the technical workflow, parameter settings, and the rationale behind the proposed Vegetation-Adaptive Weighted Coherence Threshold method. This revision aims to clarify its applicability under conditions of seasonal vegetation decorrelation and complex terrain.
Comment 2: The authors use a threshold of 16 mm/year to identify potential landslide hazards. Please discuss the rationale for this choice.
Response: The selection of this threshold is primarily based on regional characteristics: (1) Referring to existing landslide investigation data in the area, deformation rates during the initial creep stage typically concentrate within the 10-20 mm/year range. (2) Sensitivity analysis, considering the Sentinel-1 data wavelength (C-band) and the region's annual average coherence level (approximately 0.3-0.4) in the deep valley terrain of southwest China, indicates that this threshold effectively distinguishes true deformation from atmospheric phase noise. Therefore, 16 mm/a was chosen as a relatively conservative criterion for preliminary screening of potential deformation zones.
Comment 3: There are many established atmospheric correction methods in InSAR data processing. Please compare these with the method used in this study, which combines ERA5 meteorological data and a tropospheric delay model for atmospheric correction.
Response: We have supplemented the revised manuscript with a comparative discussion of atmospheric correction methods, focusing on their applicability, data requirements, and computational efficiency. In mountainous environments lacking dense ground-based meteorological networks, the method utilizing global reanalysis data (ERA5) combined with a tropospheric delay model offers good applicability, as it effectively mitigates atmospheric delay without relying on high-density in-situ stations. In comparison, correction methods based on empirical models are prone to residual errors in areas with significant topographic relief. Methods relying on GNSS-derived water vapor interpolation are limited by the sparse distribution of meteorological stations in our study area. While atmospheric correction methods based on numerical simulations (e.g., WRF) offer higher accuracy, they come with greater computational costs and are more suitable for local or event-scale studies. GACOS (Generic Atmospheric Correction Online Service) is a mature scheme validated in numerous studies; however, for our study area's complex topography and the requirements of a long time-series analysis, the tropospheric delay correction method directly based on ERA5 data offered better data consistency and processing control, aligning with the needs of regional-scale landslide system identification. Considering both the study area conditions and research objectives, the adopted method achieves a reasonable balance between correction effectiveness and computational efficiency.
Comment 4: In Figure 6, the authors delineated the landslide boundary with a red polygon, but no obvious InSAR deformation signal is observed in these areas. Please explain this discrepancy.
Response: Thank you for raising this important point. The discrepancy between the InSAR deformation signals and the geomorphologically defined landslide boundary in Figure 6 is indeed a key observation, and we have reflected on it deeply. We believe this phenomenon aptly illustrates the intrinsic difference between long-term geomorphological features and short-term surface deformation monitoring, contributing to a more comprehensive understanding of the landslide's activity state and evolutionary history.
Specifically, the red polygon in the figure represents the geological boundary of the landslide, delineated based on high-resolution imagery and field surveys, using clear geomorphological features such as the main scarp, lateral tension cracks, and toe bulge. It signifies the overall potential extent of the landslide formed through its long-term geological history. In contrast, the InSAR technique captures minute surface deformation occurring during the specific monitoring period (e.g., 2022-2023), reflecting current activity. The spatial mismatch between the two can be attributed to several factors: the landslide may be in a state of overall, uniform slow creep, where the deformation rate or its spatial gradient is below the detection sensitivity threshold of the InSAR analysis. We fully agree on the importance of explicitly explaining the geological and monitoring significance of this discrepancy. Rather than being contradictory, it suggests that the landslide did not exhibit significant localized acceleration at its boundaries during the monitoring period, indicating a state of relatively stable creep. In response to your suggestion, we will clearly explain the definitions and differences of these two boundaries in the caption of Figure 6 and consider adding a brief discussion in the "Discussion" section on the complementary roles of geomorphological features and InSAR monitoring in landslide analysis to aid reader comprehension.
Comment 5: The reliability of the InSAR results in Figures 8(b) and 8(d) is questionable. It is difficult to discern whether these scattered pixels represent deformation, noise, or error. Furthermore, please explain the rationale for selecting time-series feature points P1 and P2. Additionally, it is suggested that Figures 2 and 3 be moved to the appendix.
Response: Regarding the reliability of scatter points in Figures 8(b) and 8(d), the revised manuscript now includes explanations concerning coherence levels, temporal continuity of time series, and spatial consistency to help distinguish potential deformation signals from noise. The selection of time-series feature points P1 and P2 was based on their relatively high coherence, stable temporal behavior, and representative spatial locations within the potential landslide body. This rationale has been added to the main text. Furthermore, following your advice, the original Figures 2 and 3 have been moved to the appendix.
Comment 6: The analysis of the research results in the manuscript does not align well with the data, which may confuse readers. For example, the text states: "Furthermore, the deformation trends of points 1 and 2 were not synchronized with reservoir water level changes during the two water impoundment cycles, indicating a time-lag effect of landslide deformation in response to water level fluctuations." The authors are advised to use signal processing methods to quantify this time-lag effect.
Response: Thank you for highlighting the issue of potential misalignment between analysis and data. We have addressed this by implementing the suggested signal processing approach. Specifically, we have employed cross-correlation analysis to quantify the temporal relationship between landslide deformation and reservoir water level changes. The revisions include: (1) quantifying the lag time (e.g., -40 time steps for H28 P1) through cross-correlation analysis; (2) explicitly explaining in the text the physical meaning of a "negative lag" (indicating deformation leading water level change); and (3) incorporating discussion on statistical significance with reference to significance thresholds in the cross-correlation plots. These modifications provide clear, data-supported evidence for the conclusion that "landslide deformation exhibits a time-lag effect in response to water level changes," thereby eliminating potential reader confusion.
The above are our point-by-point responses and explanations of the revisions made based on your valuable comments. We believe the manuscript's quality has been significantly enhanced by incorporating your suggestions. Once again, we express our deepest gratitude for the time and expert guidance you have devoted. We kindly ask you to review the revised content.
Sincerely,
Wenfei Xi (on behalf of all authors)
Corresponding Author Email: wenfeixi@ynnu.edu.cn
Date: January 5, 2026
Citation: https://doi.org/10.5194/egusphere-2025-3054-AC5
-
RC4: 'Comment on egusphere-2025-3054', Anonymous Referee #4, 28 Jan 2026
This study utilizes interferometric baseline optimization to identify the reservoir-bank landslides in deeply incised mountain canyon areas. They established a vegetation-coherence coupling model to dynamically adjust coherence thresholds based on quantified vegetation coverage levels and synergized ERA5 meteorological data with tropospheric delay modeling for atmospheric correction. It’s a good work to improve the applicability of the InSAR in deeply incised mountainous canyon areas. However, there are areas that require clarification and strengthening, such as the abstract, figures, and field investigations. Revisions in these areas are needed to improve the quality and persuasiveness of the manuscript.
- In abstract. I suggested that don’t highlight your main conclusion via compare the variation of number. A valuable conclusion should be focus on what we know about the new thing or interesting thing. For example, (2) 140,146 additional valid phase-unwrapping points were obtained, indicating substantially improved interferometric processing quality. The reader may don’t care what exactly number of valid phase-unwrapping point have been found, but care about what exactly it means. Is that means that the new method can found more landslides in vegetation covered areas?
- Introduction. Comment on the first paragraph: short the first paragraph, please just focus on the big research background introduce in the first paragraph.
- Figure 1. It should be Legend but not Legebd. By the way, the map of China in the top-right should be (a). In the bottom-left, Why only Yunnan has been highlight using pink color?
- Figure 7. I keep my query about the polygons in Figure 7. I think some of the polygons are not the boundaries of landslides.
- Again, Figure 8. Especially, the polygon has showing in the Figure 8a. I believe that it’s just a deep valley but not a landslide. Because I did not see the typical landslide scarp in this figure.
Citation: https://doi.org/10.5194/egusphere-2025-3054-RC4 -
AC6: 'Reply on RC4', Hong Wenyu, 01 Feb 2026
Response to Reviewers
Dear Reviewer,
Thank you for taking the time to review our manuscript and for providing valuable feedback. Your comments are highly insightful and have provided essential guidance for improving the quality of the paper. We have carefully considered each point and will implement substantive revisions in the revised manuscript.
Comment 1: In abstract. I suggested that don’t highlight your main conclusion via compare the variation of number. A valuable conclusion should be focus on what we know about the new thing or interesting thing. For example, (2) 140,146 additional valid phase-unwrapping points were obtained, indicating substantially improved interferometric processing quality. The reader may don’t care what exactly number of valid phase-unwrapping point have been found, but care about what exactly it means. Is that means that the new method can found more landslides in vegetation covered areas?
Response: Thank you for the reviewer’s valuable suggestion. We have revised the Results statements in the abstract to avoid emphasizing our main conclusions through numerical comparisons. Instead, we highlight the key new capability enabled by our method: by explicitly incorporating vegetation-cover variations to adaptively adjust the coherence threshold, the proposed approach improves phase-unwrapping reliability and the spatial continuity of the deformation field under low-coherence conditions during high-vegetation seasons, thereby enhancing landslide detectability in vegetated, deeply incised canyon environments. Quantitative metrics (e.g., the increase in quality-controlled unwrapped pixels and the reduction in deformation noise) are retained only as supporting evidence to objectively document the magnitude of improvement. The abstract has been rewritten accordingly, and the corresponding changes have been marked in the revised manuscript.
Comment 2: Introduction. Comment on the first paragraph: short the first paragraph, please just focus on the big research background introduce in the first paragraph.
Response: Thank you for the reviewer’s valuable suggestions. We understand your recommendation for a more focused opening in the Introduction section. Accordingly, in the revised manuscript, we have adjusted the structure: the first paragraph now only briefly outlines the research background of reservoir landslides, the representativeness of the study area, and the necessity of early identification. The technical details originally included in the first paragraph—such as interferometric pair selection and baseline network construction—have been moved to a separate later paragraph, allowing for a clearer and more organized review of existing methods and their limitations. These adjustments have improved the overall logic and flow of the Introduction. We appreciate your thoughtful comments.
Comment 3: Figure 1. It should be Legend but not Legebd. By the way, the map of China in the top-right should be (a). In the bottom-left, Why only Yunnan has been highlight using pink color?
Response: Thank you for pointing out the cartographic issues in Figure 1. We have revised the figure accordingly to improve its consistency and readability: (1) the misspelling “Legebd” in the legend has been corrected to “Legend”; (2) following common cartographic conventions, the China locator map in the upper-right corner has been designated as panel (a), and the remaining study-area and thematic maps have been labeled sequentially as (b), (c), etc.; and (3) regarding the previous use of pink shading to highlight only Yunnan Province in the lower-left map, this was originally intended to provide a quick administrative reference for the study area, but we acknowledge that it could be misleading by overemphasizing Yunnan and obscuring the broader context of the Sichuan–Yunnan border reservoir region. To avoid ambiguity, we have removed the single-province highlighting and replaced it with a neutral basemap, while delineating the Baihetan Reservoir study area using consistent symbols/boundaries (with provincial boundaries shown in a uniform style where necessary) and clarifying these elements in the legend. These revisions have been incorporated into the revised Figure 1 and its caption.
Comment 4: Figure 7. I keep my query about the polygons in Figure 7. I think some of the polygons are not the boundaries of landslides.
Response: Thank you very much for your careful observation. In Figure 7, the red polygons denote inventory-based landslide boundaries, which were manually delineated by interpreting geomorphological evidence from high-resolution optical imagery and the DEM (e.g., head scarps/tension cracks, lateral margins, and depositional features). These polygons therefore represent the geomorphic extent of mapped landslides rather than boundaries derived directly from the InSAR LOS velocity field using a deformation threshold. Because InSAR measures only the line-of-sight (LOS) component and is constrained by coherence, landslides may show spatially heterogeneous deformation, local decorrelation, layover/shadow effects, or motion directions partly insensitive to the LOS. Consequently, the observable deforming pixels may not fully cover the entire landslide body and may not coincide exactly with the mapped geomorphic boundary, which can create the impression that some polygons do not match the deformation pattern. To address this potential ambiguity, we have carefully re-checked the polygons against the optical imagery and DEM and revised the caption to explicitly state that the polygons are inventory-based landslide boundaries, thereby making the relationship between landslide extent and InSAR deformation information clearer.
Comment 5: Again, Figure 8. Especially, the polygon has showing in the Figure 8a. I believe that it’s just a deep valley but not a landslide. Because I did not see the typical landslide scarp in this figure.
Response: Thank you for your valuable feedback. We understand your concern that the "typical arcuate rear scarp" may appear less distinct in the single nadir optical image of Figure 8a, given the target's location on a steep, deeply incised valley slope affected by shadows, illumination, and surface erosion. To address this, we have re-evaluated the polygon by integrating multi-temporal high-resolution optical imagery and DEM-derived data (e.g., slope, hillshade). Although the rear scarp is subtle in the original context, the area still exhibits a consistent set of landslide geomorphic evidence—including an upper slope break, relatively clear lateral boundaries, slope texture disturbance, and lower accumulation/frontal features—supporting its identification as a landslide. To improve clarity, we have optimized Figure 8a by using a clearer base map (with optional supplemental hillshade/slope maps), enhancing annotations of key features, and slightly adjusting the polygon boundary, ensuring the landslide characteristics are more intuitively presented in the revised figure.
The above are our point-by-point responses and explanations of the revisions made based on your valuable comments. We believe the manuscript's quality has been significantly enhanced by incorporating your suggestions. Once again, we express our deepest gratitude for the time and expert guidance you have devoted. We kindly ask you to review the revised content.
Sincerely,
Wenfei Xi (on behalf of all authors)
Corresponding Author Email: wenfeixi@ynnu.edu.cn
Date: January 31, 2026
Citation: https://doi.org/10.5194/egusphere-2025-3054-AC6
Data sets
InSAR baseline optimization data Hong Wenyu https://doi.org/10.57760/sciencedb.22558
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 1,224 | 140 | 46 | 1,410 | 36 | 37 |
- HTML: 1,224
- PDF: 140
- XML: 46
- Total: 1,410
- BibTeX: 36
- EndNote: 37
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
感谢您对我们关于植被适应加权相干阈值法 （WCTM） 的研究感兴趣。该方法旨在通过明确纳入植被物候动态来解决植被茂密峡谷地区的季节性去相关挑战，这是传统 InSAR 基线优化中经常被忽视的因素。在白鹤滩库区的案例研究中，WCTM显著提高了监测精度和滑坡检测能力。
我们希望与其他研究人员一起讨论WCTM方法的潜在适用性。您认为，WCTM 方法是否也能在其他类型的地表覆盖区域中发挥价值，例如季节性积雪覆盖地区、集约化农业区或快速城市化环境？此外，除了我们研究采用的C波段Sentinel-1数据和NDVI指数之外，结合L波段（例如ALOS-2/4）或P波段SAR数据，或直接利用SAR反向散射系数来缓解不同类型的去相关源，有哪些前景和挑战？
我们认为，WCTM框架通过明确考虑动态表面覆盖变化，为InSAR基线优化提供了一种新方法，在提高低相干性区域的相位展开质量方面具有特别的优势。未来工作的一个关键问题是评估其在不同地理环境和传感器配置中的可转移性和适应性。我们也非常有兴趣听听您对该方法的潜在局限性的看法，以及您对其扩展到多源遥感数据融合的建议。
我们相信，深入探讨该方法的通用性和局限性，将极大地推动InSAR技术在复杂环境监测中的广泛应用。我们非常重视您的见解，并期待从您的专业知识和实践经验中学习。