Long-term evolution and type-specific rainy-season activity of post-seismic geohazards in Southwest China

Duan, Hongyu; Tu, Guoxiang; Liu, Han

doi:10.5194/egusphere-2026-2643

Preprints

https://doi.org/10.5194/egusphere-2026-2643

Preprints

06 Jul 2026

| 06 Jul 2026

Status: this preprint is open for discussion and under review for Natural Hazards and Earth System Sciences (NHESS).

Long-term evolution and type-specific rainy-season activity of post-seismic geohazards in Southwest China

Hongyu Duan, Guoxiang Tu, and Han Liu

Abstract. Long-term evolution of post-seismic geohazards represents a critical yet complex challenge in earthquake-affected regions. This study systematically investigates the spatiotemporal patterns of post-seismic geohazard evolution following three major seismic events: the 2008 Wenchuan earthquake (Mw7.9), 2013 Lushan earthquake (Mw6.6), and 2017 Jiuzhaigou earthquake (Mw7.0), focusing on four representative counties (Wenchuan, Beichuan, Lushan, and Jiuzhaigou County) in Southwest China. The results show that: (1) post-seismic geohazard activity showed an intermittent clustered outbreak pattern; (2) geohazards were mainly concentrated in valley-related geomorphic units, with county-level differences characterized by reactivation–expansion, spatial inheritance, localization, and contraction; (3) monthly records excluding earthquake-month records showed a clear rainy-season concentration, with 78.1 % of geohazards occurring from June to August; and (4) Beichuan and Lushan showed clearer weakening trends, whereas Wenchuan and Jiuzhaigou remained in low-level persistent activity stages and should receive continued monitoring attention. These findings provide useful insights for long-term geohazard monitoring and risk mitigation in earthquake-affected mountainous regions.

Received: 08 May 2026 – Discussion started: 06 Jul 2026

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Hongyu Duan, Guoxiang Tu, and Han Liu

Status: open (until 21 Nov 2026)

Post a comment Subscribe to comment alert

RC1:
'Comment on egusphere-2026-2643', Anonymous Referee #1, 18 Aug 2026 reply
The manuscript addresses essentially technical, scientific, and probabilistic issues, which are integral to the topics covered by NHESS. The study examines the long-term evolution and seasonal patterns of post-seismic geohazards in the mountainous regions of western Sichuan, China. It focuses on four counties (Wenchuan, Beichuan, Lushan, and Jiuzhaigou) affected by three major earthquakes (Wenchuan 2008, Lushan 2013, Jiuzhaigou 2017). The study, which covers the geological and geomorphological context of China, can, however, be considered a model for internationally applicable research. The research findings, data, and maps reported in the paper may encourage fellow scientists to experiment with this research methodology elsewhere in the world, especially in seismically active areas affected by potentially destructive earthquakes, capable of also impacting slope stability. Therefore, the conclusions are substantial. The title clearly reflects the content of the document, the abstract is complete and provides consistent information about the study, which is intended for a broader, non-specialist audience, also taking into account the implications for civil protection. The size and readability of the figures are sufficiently relevant and appropriate to the data presented. The authors give due credit to previous work, spanning a sufficient period of time. In this regard, both the number and quality of references are appropriate, and the references are accessible for use by other researchers. The overall presentation is well-structured, clear, and easy to understand, especially for technicians and scientists in the field. The length of the paper is adequate and balanced for an international scientific context. The technical language is precise and understandable even by other researchers thanks to the use of good-quality English, which is fluent, simple, and easy to read and understand, even by a broad and diverse audience. The document is not accompanied by supplementary material.
Given its characteristics, the paper can be accepted without modification, but if the authors deem it appropriate, they could enrich the manuscript with:
whether they have identified an earthquake magnitude beyond which post-seismic risk is triggered

consider site effects and any ground accelerations in relation to the types of disasters considered

indicate which types of landslides (collapse, flow, slide, etc.) are most affected by seismic events of high intensity and magnitude.

Add data from the 2008 Wenchuan, 2013 Lushan, and 2017 Jiuzhaigou earthquakes.

Reply
Citation: https://doi.org/10.5194/egusphere-2026-2643-RC1
- AC1: 'Reply on RC1', Hongyu Duan, 28 Aug 2026 reply
  
  We sincerely thank the reviewer for the positive and constructive assessment of our manuscript. We greatly appreciate the reviewer’s recognition of the study’s value, manuscript structure, figure quality, references, and overall presentation. The additional suggestions are helpful for improving the seismic background and clarifying the relationship between the initial seismic disturbance and subsequent post-seismic geohazard evolution. We have carefully considered these suggestions and will further revise the manuscript based on the available data and the scope of this study.
  Comment 1: Whether the authors have identified an earthquake magnitude beyond which post-seismic risk is triggered.
  Response:
  
  Thank you for raising this important question. We have not identified a specific earthquake-magnitude threshold for long-term post-seismic geohazard risk. This study mainly examines the post-seismic evolution associated with the 2008 Wenchuan, 2013 Lushan, and 2017 Jiuzhaigou earthquakes and their corresponding study areas; therefore, the number of major earthquake events available for comparison is limited.
  In addition, several smaller earthquakes occurred during the post-seismic periods following these major events. However, these smaller earthquakes affected slopes and loose materials that had already been disturbed by the corresponding major earthquakes. Their geohazard responses may therefore reflect both renewed seismic disturbance and the pre-existing post-seismic disturbance, making them unsuitable as independent samples for identifying a magnitude threshold.
  The available data are therefore insufficient to establish a generally applicable magnitude threshold. We will clarify this limitation in the revised manuscript. Future studies incorporating more independent earthquake events of different magnitudes and complete post-seismic geohazard records may help determine whether stable relationships or threshold behavior exist between earthquake magnitude and long-term post-seismic geohazard risk.
  Comment 2: Consider site effects and any ground accelerations in relation to the types of disasters considered.
  Response:
  
  We agree that peak ground acceleration (PGA), local ground-motion characteristics, and site effects can influence the initial disturbance of slopes and may subsequently affect post-seismic geohazard development.
  However, event-specific strong-motion records and site-response data with consistent coverage across all study areas were not systematically collected when the geohazard database was established. The publicly available strong-motion data are mainly from a limited number of discrete stations and cannot provide a consistent and comparable PGA representation for the four county-scale study areas. Therefore, the available data are insufficient to establish a quantitative relationship among PGA, site effects, and different geohazard types. We will acknowledge this data limitation in the revised manuscript and identify it as an important direction for future research.
  Comment 3: Indicate which types of landslides (collapse, flow, slide, etc.) are most affected by seismic events of high intensity and magnitude.
  Response:
  
  Thank you for this helpful suggestion. This issue is already addressed in the original manuscript. Specifically, Results Section 4.3 (Figure 10) compares the geohazard-type composition during the months of the major earthquakes. During the May 2008 Wenchuan earthquake, rock falls dominated in Wenchuan and Jiuzhaigou, whereas landslides dominated in Beichuan. During the April 2013 Lushan earthquake, landslides dominated in Lushan, while the August 2017 Jiuzhaigou earthquake was dominated by rock falls. These results show that earthquake-month geohazard peaks were generally dominated by rock falls and landslides.
  The Discussion further compares the immediate seismic response with subsequent post-seismic activity. Earthquake-month activity was mainly dominated by rock falls and landslides, whereas later rainy-season activity was dominated more by landslides and debris flows, indicating a shift from the immediate seismic response to rainfall-driven post-seismic reactivation. We agree that this conclusion can be stated more explicitly and will further highlight this type-specific transition in the revised Discussion and Conclusions.
  Comment 4: Add data from the 2008 Wenchuan, 2013 Lushan, and 2017 Jiuzhaigou earthquakes.
  Response:
  
  Thank you for this suggestion. The current manuscript already provides basic information on the 2008 Wenchuan, 2013 Lushan, and 2017 Jiuzhaigou earthquakes in the “Study Area and Geological Setting” section, but these data are presented separately in narrative form.
  To improve the clarity and comparability of the seismic background, we will add a summary table compiling the main information for the three earthquakes, including occurrence date, magnitude, epicentral information, major faulting or rupture characteristics, and corresponding study area(s).
  We again thank the reviewer for these constructive suggestions, which will help us further improve the manuscript.
  
  Reply
  
  Citation: https://doi.org/10.5194/egusphere-2026-2643-AC1
RC2:
'Comment on egusphere-2026-2643', Anonymous Referee #2, 05 Sep 2026 reply
I thank the editor for the opportunity to review the manuscript entitled “Long-term evolution and type specific rainy season activity of post-seismic geohazards in southwest China”
I have to say that I was not convinced by the findings presented in the manuscript. The authors claim that they observe a “clear” relationship (e.g., lines 397–398), but they do not present any sound statistical, spatial, temporal or other quantitative method to support this claim. Additionally, no research questions or hypotheses are presented, while the stated objectives of the study are only partially examined.
In detail:
The Introduction is confusing while in several places is See, for example, lines 57–68, and in particular lines 66–68. It is not clear whether the recovery period discussed by the authors refers to years, decades, centuries, or another timescale. Additionally, the authors refer to two key processes but do not explain why these processes are important or for what purpose they are being examined. Similar contradictions and phrasing issues appear in lines 45–47 and 76–85.

This brings me to another important issue. Although the study has stated objectives, I do not see clearly defined research questions or hypotheses. In addition, part of the objectives remains unclear. For example, what exactly is meant by “non-earthquake-month geohazard activity”? The authors should at least explain this term.

Throughout the manuscript, statements are made without sufficient support. See, for example, lines 96–102. This issue applies in particular to the entire Section 2 and also to parts of Section 3.

The data issue is particularly vexing. There is no reference to the source of the data and no supplementary material is provided. Additionally, the authors describe the data textually in lines 137–142. Why not present this information in a table? This would make it much easier to read and understand.

Regarding lines 143–148: is the distribution of the data normal or are extreme events distorting the results?

The first earthquake examined occurred in 2008, but the precipitation analysis begins only two years earlier, in 2006. I think this is a very short period from which to draw sound conclusions.

The findings are not sufficiently supported. Consider, for example, Figure 5 and line 185. I cannot understand the correlation or connection between precipitation and the other findings. How is this relationship demonstrated? Additionally, I cannot understand the distribution of events in the county by year. There is no statistical analysis of the proportions, and it is unclear how precipitation is linked to these patterns. The authors should revise this section and provide further explanation.

Figures 6, 7, and others present continuous data while the raw data appear to be discrete. What interpolation method was used to convert the discrete data into continuous representations?

The fact that there are more events in the year following an earthquake does not necessarily imply a causal or meaningful linkage. I would expect the authors to use a statistical or other quantitative method to support this notion. The authors currently present the distribution using bar charts, but I would expect to see something more consistent, for example an analysis based on proportions or Poisson tests.

I would also expect the authors to perform spatial and/or temporal analyses. For example, where and when can a decay in post-seismic geohazard activity be observed?

The authors reach conclusions that are not sufficiently supported by the findings of the study. See, for example, lines 384–390 and 397–400.

Reply
Citation: https://doi.org/10.5194/egusphere-2026-2643-RC2
- AC2:
  'Reply on RC2', Hongyu Duan, 16 Sep 2026 reply
  We thank the reviewer for the comments on our manuscript. We have carefully considered these comments and provide our point-by-point responses below.
  The Introduction is confusing while in several places is See, for example, lines 57–68, and in particular lines 66–68. It is not clear whether the recovery period discussed by the authors refers to years, decades, centuries, or another timescale. Additionally, the authors refer to two key processes but do not explain why these processes are important or for what purpose they are being examined. Similar contradictions and phrasing issues appear in lines 45–47 and 76–85.
  
  Response:
  We re-examined the statements in Lines 45–47, 57–68, and 76–85 as pointed out by the reviewer. The discussion of “recovery” in Lines 57–68 was intended to emphasize the long-term persistence of post-seismic geohazard activity and the substantial differences in persistence among different earthquake-affected regions, rather than to define specific recovery periods for the four study areas. The studies cited in the original manuscript indicate that post-seismic geohazard activity may persist for years to decades. In the revision, we will make this timescale clearer and avoid presenting the “recovery period” as a fixed stage defined in this study.
  The progressive removal of unstable material and vegetation recovery mentioned in Lines 66–68 were possible explanations proposed in previous studies for the long-term evolution of post-seismic geohazard activity, rather than two variables specifically examined in this study. The expression “two key processes” in the original manuscript may have overstated their importance and their relevance to the present analysis, and we will therefore remove this wording. The discussion of vegetation disturbance in Lines 45–47 is intended to illustrate the geological and ecological disturbance caused by strong earthquakes, and we will revise the wording and paragraph transitions to clarify its role as background information. The research objectives and research questions referred to in Lines 76–85 are further addressed in our response to Comment 2.
  We will also reorganize the overall logic and transitions of the Introduction following the sequence: “strong earthquakes trigger geohazards and cause structural and ecological damage—post-seismic damage may develop into new geohazards under subsequent external disturbances—long-term observations show that post-seismic geohazards may persist for extended periods—rainfall acts as an important external triggering factor affecting post-seismic geohazards—research gaps and the research questions of this study.” This revision will clarify the connection between the research background and the objectives of the present study.
  This brings me to another important issue. Although the study has stated objectives, I do not see clearly defined research questions or hypotheses. In addition, part of the objectives remains unclear. For example, what exactly is meant by “non-earthquake-month geohazard activity”? The authors should at least explain this term.
  
  Response:
  The original manuscript stated three research objectives but did not explicitly formulate them as research questions. Previous studies have shown that geohazard activity in earthquake-affected regions may persist for extended periods after strong earthquakes, making it important to understand its long-term evolution for both process interpretation and long-term risk management. Meanwhile, existing studies have often focused on individual earthquakes, local representative areas, or specific geohazard types, whereas comparative analyses of multiple counties based on long-term continuous records within a unified framework remain relatively limited. In the revision, we will clarify this research background and research gap at the end of the Introduction and reformulate the objectives as the following three research questions: (1) What temporal changes and representative-year spatial patterns are observed in geohazards across the four study areas during the long-term observation period? (2) After excluding geohazard records from the months in which the relevant earthquakes occurred, what seasonal distribution patterns and type-specific differences are observed during the remaining months? (3) How does the monthly probability of geohazard occurrence change with elapsed post-earthquake time across the different study areas?
  The term “non-earthquake-month geohazard activity” in the original manuscript refers to geohazards recorded during the remaining months after excluding the months in which the relevant earthquakes occurred. This treatment was intended to separate the concentrated coseismic records that may occur during earthquake months from geohazards recorded during other months, so that the seasonal distribution of the latter could be examined. In the revision, this term will be explicitly defined when it first appears in the Methods section.
  Throughout the manuscript, statements are made without sufficient support. See, for example, lines 96–102. This issue applies in particular to the entire Section 2 and also to parts of Section 3.
  
  Response:
  We re-examined the supporting information for the statements in Lines 96–102 and throughout Section 2. Lines 96–102 involve previously reported seismogeological information, including the faulting styles, surface rupture lengths, coseismic displacements, and tectonic kinematics of the Wenchuan, Lushan, and Jiuzhaigou earthquakes. Appropriate references will be added in the revised manuscript. Necessary references will also be added for the regional tectonic setting, major fault systems, earthquake-induced rock-mass fracturing, vegetation disturbance, and subsequent reactivation of earthquake-damaged materials discussed in Section 2.
  Figures 2 and 3 present direct field observations from this study, including fractured and loosened rock-soil masses and representative landslides, rockfalls, and debris flows. In the revision, we will further clarify their observational basis so that direct observations from this study can be clearly distinguished from previously reported regional background information.
  Regarding the reviewer’s concerns about Section 3, because Comment 4 raises more specific questions regarding data sources and data presentation, we address these issues collectively in our response to Comment 4 and will revise the corresponding section accordingly.
  The data issue is particularly vexing. There is no reference to the source of the data and no supplementary material is provided. Additionally, the authors describe the data textually in lines 137–142. Why not present this information in a table? This would make it much easier to read and understand.
  
  Response:
  The source of the geohazard inventory was already described in Lines 121–123 of the original manuscript. The inventory was mainly derived from long-term monitoring records of the Sichuan Provincial Geological Environment Monitoring Station and was verified by professional geological investigation teams. The precipitation data were obtained from the same monitoring system. In the revised manuscript, we will provide further information on the data formats, major fields, and data-processing procedures for both the geohazard and precipitation datasets.
  The Data Availability statement in the original manuscript already indicates that the datasets used and analyzed in this study are available from the corresponding author upon reasonable request. These original monitoring data are not part of a publicly accessible database but were obtained through project-based research collaboration, and therefore were not directly provided as Supplementary Material.
  To improve the readability of the information currently presented in Lines 137–142, we will summarize the total number of geohazards and the numbers of landslides, rockfalls, and debris flows for the four study areas in a table and correspondingly streamline the textual description.
  Regarding lines 143–148: is the distribution of the data normal or are extreme events distorting the results?
  
  Response:
  The multi-year mean used in Lines 143–148 of the original manuscript was introduced mainly to reduce subjectivity in identifying annual high-value years and was not used in subsequent statistical analyses. We agree that a small number of pronounced high-value years may influence the long-term mean. Because this criterion is not necessary for the subsequent analyses, we will remove it in the revision. Instead, the long-term annual records will be used directly to describe episodic high values and their intermittent occurrence. Representative years will then be selected based on earthquake years, years with clearly elevated annual geohazard counts, and representative precipitation conditions.
  The first earthquake examined occurred in 2008, but the precipitation analysis begins only two years earlier, in 2006. I think this is a very short period from which to draw sound conclusions.
  
  Response:
  We would like to clarify that the precipitation records from 2006–2007 were not used as a pre-Wenchuan-earthquake precipitation baseline, nor did this study conduct a pre- versus post-earthquake comparison of precipitation conditions. The 2006–2023 precipitation dataset was used only for within-period comparison with the contemporaneous geohazard records. At the annual scale, we preliminarily examined whether years with high geohazard counts coincided in time with years in which annual precipitation exceeded the mean level of the observation period. This comparison was not used to determine a rainfall-triggering threshold or to assess whether post-earthquake precipitation differed from pre-earthquake conditions. At the monthly scale, the analysis was further used to examine the seasonal concentration of geohazards during the main rainy season. Therefore, the conclusions of this study do not depend on having only two years of pre-earthquake precipitation records.
  The findings are not sufficiently supported. Consider, for example, Figure 5 and line 185. I cannot understand the correlation or connection between precipitation and the other findings. How is this relationship demonstrated? Additionally, I cannot understand the distribution of events in the county by year. There is no statistical analysis of the proportions, and it is unclear how precipitation is linked to these patterns. The authors should revise this section and provide further explanation.
  
  Response:
  Figure 5 presents the annual number of recorded geohazards, the composition of different geohazard types, and the corresponding annual precipitation for the four study areas. We would like to clarify that the statement concerning precipitation in Line 185 was intended only as a descriptive annual-scale observation: some years with clearly elevated geohazard counts coincided with years in which annual precipitation was above the mean level of the observation period. We did not use this observation to attribute the geohazard peaks or intermittent outbreak behavior to precipitation, nor did we interpret it as a statistically significant precipitation–geohazard relationship. In the revision, we will adjust the wording around Line 185 to avoid interpreting this temporal coincidence as a definite precipitation–geohazard linkage. Section 4.3 subsequently examines the seasonal concentration of geohazards during the main rainy season and the composition of different geohazard types at the monthly scale.
  Regarding the composition of geohazard types, the stacked bars in Figure 5 already show the annual number and type composition of geohazards in each study area. To make these results more explicit, we will add the overall proportions of landslides, rockfalls, and debris flows for each study area to the data summary table and provide quantitative descriptions of the type composition for the major high-value years in the main text.
  Figures 6, 7, and others present continuous data while the raw data appear to be discrete. What interpolation method was used to convert the discrete data into continuous representations?
  
  Response:
  We would like to clarify that no spatial interpolation was applied to the geohazard data shown in Figures 6–9. As stated in the Methods section of the original manuscript, these maps were generated in ArcGIS by directly plotting the recorded locations of individual geohazard events. In some areas, the high density of events, together with the map scale and symbol size, causes individual points to overlap and visually appear as a more continuous cluster. However, the geohazard data are represented throughout as discrete event points.
  The fact that there are more events in the year following an earthquake does not necessarily imply a causal or meaningful linkage. I would expect the authors to use a statistical or other quantitative method to support this notion. The authors currently present the distribution using bar charts, but I would expect to see something more consistent, for example an analysis based on proportions or Poisson tests.
  
  Response:
  We agree that an increase in the number of geohazards during a particular year or period after an earthquake does not, by itself, establish a direct causal relationship with the earthquake. Post-seismic geohazards may also be influenced by precipitation, progressive weathering, gravitational adjustment, and other environmental factors, and a specific geohazard peak may therefore result from the combined effects of multiple factors.
  Rigorous attribution of a specific geohazard peak to earthquake effects would require sufficiently long and consistently compiled pre- and post-earthquake geohazard records, together with consideration of other major influencing factors. However, the four study areas lack systematic long-term pre-earthquake geohazard records, and the available data are therefore insufficient to further separate the relative contributions of earthquake legacy effects and other factors to a specific geohazard peak.
  In the revision, we will further distinguish observational results from mechanistic interpretation and remove or revise expressions such as “delayed response” that may imply a direct causal linkage. For the anomalously high number of rockfalls observed in Lushan County in December 2014, outside the main rainy season, we will describe this as a concentrated non-rainy-season occurrence and discuss, against the background of earthquake-induced rock-mass damage, the possible roles of progressive weathering, gravitational adjustment, and other factors in promoting the gradual instability of damaged rock masses. Relevant references will also be added to support this mechanistic interpretation.
  I would also expect the authors to perform spatial and/or temporal analyses. For example, where and when can a decay in post-seismic geohazard activity be observed?
  
  Response:
  We agree that the original assessment of long-term geohazard activity in the different study areas relied mainly on annual geohazard counts and representative-year spatial patterns and was therefore largely descriptive. In particular, these observations alone make it difficult to distinguish a systematic long-term decline from temporary low-activity intervals between concentrated outbreaks.
  To address this issue, we performed an additional monthly-scale binary Logistic regression analysis, which will be added to the revised manuscript. Specifically, each month was classified according to whether at least one geohazard was recorded (geohazard occurrence = 1; no geohazard occurrence = 0). Elapsed time since the relevant major earthquake was used as the main explanatory variable, while contemporaneous monthly precipitation was included in the model and annual sine and cosine terms were used to account for seasonality. This analysis therefore tests whether the probability of geohazard occurrence changes systematically with post-earthquake time after accounting for monthly precipitation variability and seasonal effects.
  The results show that the monthly probability of geohazard occurrence decreased significantly with post-earthquake time in Wenchuan and Jiuzhaigou, whereas no significant decrease was detected in Beichuan or Lushan. Based on these results, we will revise the original interpretations of “decay” and “persistence” for the different study areas. Representative-year spatial contraction or decreases in annual geohazard counts will no longer be used directly as criteria for inferring activity decay, and the corresponding statements in the Results, Discussion, and Conclusions will be revised accordingly.
  The authors reach conclusions that are not sufficiently supported by the findings of the study. See, for example, lines 384–390 and 397–400.
  
  Response:
  For the statements in Lines 397–400 concerning activity decay and persistence in the different study areas, the supporting evidence is closely related to the issue raised in Comment 10. As described above, these interpretations will be revised based on the newly added quantitative analysis.
  The discussion in Lines 384–390 was not intended to present mechanisms directly demonstrated by observations in this study, but rather to interpret the observed differences in rainy-season activity among different geohazard types. Increased pore-water pressure, reduced effective stress and shear strength due to rainfall infiltration, and the erosion and entrainment of loose slope and channel materials by rainfall-generated runoff are mechanisms widely discussed in previous studies for landslides and debris flows. In the revision, we will further verify and supplement the relevant references.
  To avoid the impression that these mechanisms were directly demonstrated by the present study, we will clarify the interpretive nature of this paragraph. For example, the paragraph may begin with: “The observed differences in rainy-season activity among different geohazard types can be interpreted in light of mechanisms reported in previous studies,” followed by discussion of the different responses of landslides, debris flows, and rockfalls.
  We again thank the reviewer for the constructive comments, which are very helpful for further improving the manuscript.
  
  Reply
  
  Citation: https://doi.org/10.5194/egusphere-2026-2643-AC2
CC1:
'Comment on egusphere-2026-2643', dan wang, 28 Sep 2026 reply
Thank you for submitting this regionally significant study. Based on geohazard records from four counties in Sichuan from 2006 to 2023, this paper examines the long-term evolution, spatial distribution, and rainy-season activity of post-seismic geohazards. The topic falls within the scope of NHESS, the data span a relatively long period, and the comparison among the four counties has certain potential value. However, after careful review, I believe that the current version still has fundamental shortcomings in terms of data representativeness, statistical methods, separation of coseismic and post-seismic events, rainfall triggering scales, quantitative spatial analysis, and mechanistic interpretation. These issues cannot be resolved through minor revisions at the textual or formatting level; rather, they are key flaws that directly affect the reliability of the core conclusions. Therefore, although the topic is valuable, in its present form I am afraid I cannot recommend publication in NHESS. The specific comments below are provided for the authors’ reference, and I hope that after supplementing the necessary data and conducting a new analysis, the authors can produce a more robust and mechanistically insightful study.

Event counts were not normalized, so cross-county comparisons lack comparability. The four counties differ greatly in area, population, road density, geohazard-prone area, and number of monitoring stations. Directly comparing 1,281 events in Wenchuan, 1,384 in Beichuan, 493 in Lushan, and 881 in Jiuzhaigou has no clear physical meaning. They should at least be standardized by area, road length, population, or geohazard-prone area. It is recommended that normalized indicators be established separately for each county and then compared with one another.

The definition of a “clustered outbreak year” is overly subjective and statistically not robust. Defining an outbreak year as a year in which the annual number of geohazards exceeds the county mean is highly sensitive to extreme values. For example, 621 events in Wenchuan in 2008 would substantially raise the mean, causing subsequent normal years to be misclassified as low-activity years. No median, MAD, quantile, change-point detection, or Poisson/negative binomial model was used, and the threshold lacks a statistical basis.

Coseismic and post-seismic hazards are not strictly separated, leading to conceptual confusion. The title and abstract emphasize post-seismic, but the annual and monthly analyses include earthquake months. The coseismic peaks in May 2008, April 2013, and August 2017 dominate the statistics. If coseismic events are not excluded, conclusions regarding post-seismic decay, rainy-season concentration, and type shifts will be seriously confounded.

Monthly and annual rainfall scales cannot establish triggering relationships. The authors report that 78.1% of hazards occurred from June to August after excluding earthquake months, but they do not compare this with the proportion of annual rainfall occurring from June to August. If rainy-season rainfall itself accounts for 60%–70% of annual rainfall, a concentration of 78.1% may not be significant. An expected-proportion comparison, chi-square test, or bootstrap confidence intervals must be conducted.

There is a lack of daily rainfall, antecedent rainfall, and threshold analysis. Landslide and debris-flow triggering generally depends on short-duration intense rainfall, antecedent rainfall, and I–D thresholds. Monthly rainfall alone cannot support mechanistic explanations such as pore-water pressure, infiltration, and reduced shear strength. At a minimum, daily rainfall, 3/7/15-day antecedent rainfall indices, or event rainfall thresholds should be used.

Spatial evolution patterns have no operational definitions or quantitative indicators. The paper proposes four patterns—reactivation–expansion, spatial inheritance, localization, and contraction—but these are mainly based on visual interpretation of representative years. Quantitative indicators such as kernel density overlap rate, standard deviational ellipse, and coseismic–post-seismic hazard overlay rate should be calculated.

The selection of representative years is not transparent. Only spatial maps for earthquake years, outbreak years, and rainfall-rich years are presented, while maps for all years or supplementary materials are not provided. Readers cannot judge whether the spatial patterns are robust, nor can they rule out the possibility that the authors selected specific years to strengthen their conclusions. All years should be provided, or at least complete annual spatial statistics should be included.

There is no pre-earthquake baseline, making it impossible to assess “recovery” or “decay to pre-earthquake levels.” The data begin in 2006, leaving only 2006–2007 as the pre-earthquake period, and the pre-earthquake windows differ among earthquakes. Without a reliable baseline, it cannot be determined whether geohazard activity has returned to pre-earthquake levels, nor can the long-term decay rate be quantified.

There are no statistical models or hypothesis tests, making it impossible to distinguish trends from random fluctuations. The paper relies mainly on descriptive statistics and visual interpretation and does not use regression, change-point detection, survival analysis, time-series analysis, or Bayesian hierarchical models. Therefore, conclusions such as “weakening trend,” “low-level persistence,” and “intermittent outbreak” lack statistical significance.

The confounding effects of aftershocks, climate change, and anthropogenic activities are not considered. Aftershocks, post-disaster reconstruction, road relocation, tourism development, mining, and slope cultivation can all affect slope stability and hazard reporting. Tourism in Jiuzhaigou and road reconstruction in Wenchuan and Beichuan are particular.

The discussion in Section 5 essentially remains a textual restatement of the results in Section 4 and a general citation of global literature; it does not transform the observations from the four counties into testable mechanistic explanations. The authors propose patterns such as “intermittent clustered outbreaks,” “spatial inheritance,” “reactivation–expansion,” “localization,” and “contraction,” but they do not answer the following questions: What processes control the time intervals between outbreaks? What controls spatial expansion and contraction? Why does the dominant type shift from coseismic rock falls/landslides to rainy-season landslides/debris flows? Why do Beichuan and Lushan show decay, whereas Wenchuan and Jiuzhaigou remain persistent? Therefore, the discussion is merely a description of distribution patterns and a naming of patterns; the mechanistic explanation is insufficient, and it cannot support the authors’ core conclusions regarding long-term evolution, regional differences, and type-specific responses. It is recommended that the authors either supplement the above mechanistic analysis or substantially reduce the causal strength of their conclusions by explicitly limiting them to descriptive statistical results based on a risk-oriented inventory.

Reply
Citation: https://doi.org/10.5194/egusphere-2026-2643-CC1

Hongyu Duan, Guoxiang Tu, and Han Liu

Viewed

Total article views: 324 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
222	72	30	324	24	36

HTML: 222
PDF: 72
XML: 30
Total: 324
BibTeX: 24
EndNote: 36

Views and downloads (calculated since 06 Jul 2026)

Month	HTML	PDF	XML	Total
Jul 2026	62	13	6	81
Aug 2026	40	13	5	58
Sep 2026	92	28	14	134
Oct 2026	28	18	5	51

Cumulative views and downloads (calculated since 06 Jul 2026)

Month	HTML	PDF	XML	Total
Jul 2026	62	13	6	81
Aug 2026	40	13	5	58
Sep 2026	92	28	14	134
Oct 2026	28	18	5	51

Viewed (geographical distribution)

Total article views: 293 (including HTML, PDF, and XML) Thereof 293 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 25 Oct 2026

Short summary

Earthquakes can leave mountain slopes unstable for many years. This study examines records of landslides, rock falls, and debris flows in four counties of Southwest China after three major earthquakes. The results show that these hazards tend to occur in repeated active periods and are concentrated near valleys, especially during the rainy season. The findings can help identify areas that need long-term monitoring and risk reduction.


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