the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 27 May 2026
Brief communication：Delayed Ice Avalanches Triggered by Earthquakes: A Strain-rate Dependent Strengthening Mechanism
Abstract. Earthquakes have long been regarded as one of the major triggers of ice avalanches (IAs). However, their role in IA initiation remains controversial. Here, we compiled 169 IA events that occurred between 1941 and 2022 together with their associated earthquake records, and found that only 9 IAs occurred on the same day as the earthquake. This indicates a pronounced delayed response of IAs to seismic forcing. Taking the Hailuogou, Yanzigou, and Dagongba glaciers on Mt Gongga as examples, and using the 5 September 2022 Luding earthquake as the dividing point, we analysed glacier surface velocities during September-December 2021 and September-December 2022. The results show that glacier motion reached its maximum 1 month after the earthquake. By comparing meteorological data from the pre- and post-earthquake periods, we excluded climate variability as the primary cause of the observed glacier acceleration, thereby confirming the statistical pattern of delayed earthquake-induced IAs. We further propose a new contact model showing that seismic loading can transiently enhance the strength of ice. This strengthening effect is interpreted as the fundamental reason for the delayed occurrence of IAs after earthquakes. This study provides a new theoretical framework and fresh insights into the failure mechanism and hazard mitigation of earthquake-induced IAs.
- Preprint
(1338 KB) - Metadata XML
-
Supplement
(276 KB) - BibTeX
- EndNote
Status: open (until 31 Aug 2026)
- RC1: 'Comment on egusphere-2026-2367', Anonymous Referee #1, 01 Jul 2026 reply
-
RC2: 'Comment on egusphere-2026-2367', Anonymous Referee #2, 25 Aug 2026
reply
This Brief Communication investigates the delayed response of ice avalanches (IAs) to earthquakes by combining a global IA–earthquake dataset, satellite-derived glacier velocities, and a strain-rate-dependent ice-strength model. The topic is interesting and potentially relevant for understanding earthquake-induced glacier instability. However, several methodological aspects and interpretations require further clarification and stronger support before the main conclusions can be fully justified. In particular, the statistical relationship between earthquakes and delayed IAs, the attribution of glacier acceleration to seismic forcing, and the link between the proposed model and the observed delay require further investigation.
The major comments are provided below.
- The lack of comprehensive temperature and precipitation data represents an important limitation, as climatic variability or short-term meteorological events could act as confounding factors in the observed delay times. Therefore, the statement that this limitation “does not undermine the statistical significance of our findings” should be reconsidered. While it may not affect the distribution itself, it does limit the interpretation of a systematic earthquake-related triggering mechanism.
- The statement that earthquakes below Ms 3.5 “appear incapable of triggering IAs” seems too strong. The absence of observed events does not necessarily demostrate the absence of triggering, as this may also depend on the number of low-magnitude earthquakes considered, their distance from unstable glaciers, and possible observational limitations. Additional information should be provided, or the conclusion should be softened.
- The statement that IA delay time is “strongly correlated with earthquake magnitude” does not appear sufficiently supported by Fig. 1c, which shows substantial scatter. A quantitative statistical analysis should be provided to support both the strength and the proposed non-linear nature of this relationship. The potential influence of other variables, particularly epicentral distance, should also be considered.
- The comparison between September–December 2021 and 2022 is insufficient to attribute the observed velocity increase to the earthquake, as glacier velocities may show significant interannual variability. A longer time series would be needed to assess whether the 2022 acceleration was actually anomalous. Moreover, the statement that all three glaciers reached their maximum velocity approximately one month after the earthquake is inconsistent with the reported results, as Hailuogou Glacier peaked in November.
- The POT methodology used to derive glacier surface velocities is not sufficiently described. Since the observed velocity changes represent a key element of the proposed post-earthquake response, the methodology used to calculate these velocities should be briefly described.
- The climatic analysis considers only air temperature and precipitation, which may not fully represent the environmental controls on glacier velocity. Moreover, the analysis is limited to three glaciers from the same region and affected by the same earthquake. This limits the generalization of the proposed delayed response to different climatic and seismic conditions.
- The proposed model describes the strain-rate dependence of ice strength and supports the hypothesis of transient strengthening during seismic loading. However, it is not clear how the model quantitatively explains the observed delay of weeks to months between the earthquake and IA occurrence. This link should be better explained, particularly regarding the post-seismic evolution from transient strengthening to delayed failure.
- The model is developed and calibrated using experimental ice-strength data and a simplified particle-contact representation. Further discussion would be useful on how this behaviour can be transferred to glacier-scale conditions and ultimately to ice-avalanche failure, including the main limitations of this scale transition.
Minor comments
Line 26. The citation Wood et al. (2024) does not appear to fully support the statement for which it is used. The reference should be checked and, if necessary, replaced or complemented with a more appropriate citation.
Lines 33–38. Literature review. The discussion of previous studies on earthquake-induced IAs is rather limited. The text refers to “some studies” and “other studies”, while only a few individual cases are presented. Since the earthquake–IA relationship represents the main motivation of the study, this section would benefit from additional references and case studies addressing both direct and delayed/indirect earthquake triggering.
Lines 50–52. It is unclear whether the IA dataset was compiled exclusively from the TPDC database or also from additional literature and open-access repositories. If additional sources were used, they should be clearly identified and referenced.
Line 55. The choice of Ms ≥ 3.0 based on the “high strain-rate sensitivity of ice” would benefit from a supporting reference or further justification.
Line 105. The statement that faster glacier motion implies lower glacier stability should be supported by appropriate references and/or expressed more cautiously.
Citation: https://doi.org/10.5194/egusphere-2026-2367-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 86 | 52 | 16 | 154 | 13 | 8 | 8 |
- HTML: 86
- PDF: 52
- XML: 16
- Total: 154
- Supplement: 13
- BibTeX: 8
- EndNote: 8
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This manuscript aims to address an interesting and important question: whether earthquakes can trigger ice avalanches with a time delay. The research question is original and relevant for hazard assessment in High Mountain Asia. However, the manuscript has several major issues with the dataset, statistical methodology, and theoretical framework. In my opinion, these issues are sufficient to recommend rejection.
First, the dataset contains factual errors and inconsistent event classifications, which reduce confidence in the analyses. Second, the statistical approach has several major problems. The epicentral distance threshold is derived from a pre-selected subset of events and then applied to the full dataset. Several key parameters are introduced without justification or sensitivity analysis, and Keefer's empirical relationship is applied to ice avalanches without explanation. The calibration is based on only 20 events; the analysis does not compare post-seismic ice avalanche occurrence with the background frequency of ice avalanches, and climatic controls such as temperature and precipitation are not available but are instead misinterpreted. As a result, the observed temporal association between earthquakes and delayed ice avalanches cannot be interpreted as evidence of causation. Third, the mechanical model operates at the microscale and is based entirely on previously published experimental data; it does not explain the observed delay times and is not directly connected to the observational analysis or the specific events discussed in the manuscript.
Overall, the manuscript reads as though it consists of loosely connected components rather than a single integrated study. The original contribution is limited, and the observational and modeling components are not sufficiently developed for publication in The Cryosphere.