Thank you very much! We are looking forward to your valuable comments.

Thanks for your advice!

1. We did a general and professional English review.

2. As for the physical relationship between the seismic and (peri) glacial processes, we have already had a supplementary discussion in Section 6.1:

Strong ground vibrations caused by earthquakes can intensify cracking within the ice/rock mass, ultimately leading to the formation of substantial failure surfaces (Kilburn and Voight, 1998). Additional loading by earthquakes and coseismic-ice/rock avalanches could destruct the englacial conduit and subglacial drainage system. These changes can cause dynamic alterations to the glacier's thermal sensitivity, exacerbating its instability (Zhang et al., 2022). As critical solid material sources, these highly active ice/rock masses caused by seismic disturbance are prone to avalanches, calving, detachment and remobilization to form glacial debris flows (Deng et al., 2017; Zhang et al., 2022).

3. The epicenter of the 1950 Assam earthquake is approximately 195 kilometers away from the ZLL Valley. Notably, this seismic event not only triggered the glacier surge-debris flow chain but also incubated the 1953 debris flow in Guxiang Valley, approximately 50 kilometers northeast of the ZLL Valley. Although the Guxiang debris flow did not occur on the same day as the earthquake, the 1950 earthquake induced co-seismic avalanches, ice falls, and rockfalls of an unprecedented scale upstream of Guxiang Valley. These events contributed a substantial volume of loose material, which subsequently led to the 1953 debris flow.

Thank you very much! These comments are very useful. We try to find more substantial links between the debris flows and earthquakes or warming events. We will improve our figures.

Thank you again!

Kaiheng Hu, on behalf of all the authors

Thanks for your advice! According to your comments, we have made the following modifications and explanations:

1. Motivation of the paper:

Recent observations indicate that episodic debris flows are predominately responsible for sediment transport from steeplands to rivers and channel erosion, being a major agent in landscape evolution in high mountain areas (Anderson et al., 2015; Kober et al., 2012; Lancaster and Casebeer, 2007; McCoy et al., 2013).  Lin et al. (2006) presented that it could last for more than 10 years after the Chi-Chi earthquake. Cui et al. (2011) predicted that the effect of Wenchuan earthquake on post-quake debris flows would last for 10-20 years, while Huang (2011) predicted that it would be from 20 to 25 years. Recent publications demonstrate that only ~ 30% portion of coseismic sediment has been evacuated by debris flows and fluvial transport 12 years after the Wenchuan earthquake (Dai et al., 2021)， and the fine-grained landslide sediment mobilized by the earthquake may stay in the affected river catchments as long as a century (Wang et al., 2015). Until now, most of previous studies have focused on the residence time and transport of earthquake-triggered landslide sediment at an orogenic scale in no-glacierized environments (Dadson et al., 2004; Dai et al., 2021; Parker et al., 2011; Wang et al., 2015). Little attentions are paid on the sediment evacuation progress by post-seismic debris flows at a catchment in glacierized environments owing to relatively low likelihood of debris flows and absence of long-term site-specific data. Consequently, this paper is driven by the motivations:

1) To describe how strong earthquakes and warming events escalate debris flows and associated sediment transport within small alpine watersheds.

2) To clarify how long the effects of an earthquake on periglacial debris flows at a glaciated catchment.

3) To examine the roles that climate and tectonics play in the development of extreme hazards in the tectonically active and climate-sensitive region of southeastern Tibet.

The reviewer's meticulous examination has highlighted an issue with our initial phrasing, which led to ambiguity.  To clarify, we deleted the sentence and changed it with: " Little attentions are paid on the sediment evacuation progress by post-seismic debris flows at a catchment in glacierized environments owing to relatively low likelihood of debris flows and absence of long-term site-specific data."

2. As mentioned above, we want to study the sediment evacuation process dominated by debris flows in glacierized environments. In order to investigate the long-term effects of earthquakes on sediment evacuation in a glaciated catchment, we select the Zelunglung catchment, a tributary of the Yarlung Tsangpo river in southeastern Tibet that has large areas of temperate glaciers and disturbed intensely by the Ms 8.5 earthquake in 1950. Moreover, the catchment has long-term remote sensing imagery for interpretating glacier changes and associated debris flows and relatively well-documented records of at least four historical periglacial debris flows in 1950, 1968, 1972, and 1984 since the 1950 Assam earthquake.

3. Methodology:

We appreciate the reviewer's valuable comment. As you pointed out, accurate estimation of debris flow volume is hindered by the lack of sediment thickness data. The sediment volume of debris flow may be influenced by various factors, such as fan area, average fan slope and average channel (e.g. Stoffel (2010)). Nevertheless, conducting a long-term time series analysis since 1950 presents challenges in acquiring adequate data for estimating sediment volume over the past century. Borehole method can give the accurate estimation of thickness, but is too expensive for our study. Consequently, this study employs an approximate alternative approach, wherein the fluctuation in debris flow volume is inferred from changes in the area of the accumulation fan. Actually, many previous studies consider debris flow volume is empirically a function of the accumulation area (e.g. Iverson et al. (1998)). It is crucial to underscore that our emphasis lies not on the absolute volume of debris flows but on their relative trends.

4. Causes of debris flows:

The influence of a large earthquake on debris flows is a long lasting process, with variable durations of impact in different studies (Dai et al., 2021). For instance, following the Wenchuan earthquake, Tang et al. (2009) suggest a debris flow activity span of 5 to 10 years, whereas Cui et al. (2008) propose 20 to 30 years. Xie et al. (2009) indicate the possibility of strong debris flow activity persisting for 10 to 30 years, or more. Essentially, the above statement pertains to the active duration of earthquake impacts on the loose source of debris flow channels. But, when the earthquake’s effects is negligible, debris flow frequency-magnitude will resume to pre-quake level. Pre-1990 debris flow events as indicated by Figure 12d, represent a legacy of the 1950 earthquake. Of course, the post-seismic debris flows maybe directly triggered by extreme precipitation or temperature fluctuation. But the instability of the glacier/materials caused by the great earthquake is not negligible. Prior to 1990, NVA exhibited a consistent decline, whereas afterward, it displayed significant fluctuation and even a slight upward trend. This implies that the aftermath of the 1950 earthquake persisted until 1990, after which debris flow resumed activity on a relatively minor scale, influenced by climatic factors. Despite the occurrence of several small earthquakes preceding the events of 1968 and 1984 as noted by L381ff, these earthquakes were not captured by the Keefer curve considering magnitude and distance, suggesting their minimal influence on the debris flow events.

5. We appreciate your valuable suggestions. The assumption of temporal random distribution of debris flows is probably right at an orogen scale. For a specific catchment, the debris flows are not random temporally, but controlled by earthquakes or climate events. That is confirmed by the 1999 Chi-chi earthquake and the 2008 Wenchuan earthquake. Whether earthquakes and climate change will definitely increase the occurrence of glacial debris flow is indeed to be further studied, but for the southeast Tibet region where ZLL is located, this seems to be based on evidence. For example, the Assam earthquake in 1950 caused frequent debris flow activities in Guxiang Valley for decades(Du and Zhang, 1981), and the increase of glacier ablation under the influence of climate change in the past 20 years has promoted the debris flow in Tianmo Valley (Deng et al., 2017) and Sanggu Valley (Wang et al., 2023). Therefore, in order to ensure objectivity, we change the relevant statement to: “It is evident that either earthquakes or climate change may increase the occurrence of periglacial debris flows and their sediment yield in southeastern Tibet.”

Figure 12 illustrates a notable decrease in rainfall since 2000, alongside an increase in temperature, with a particularly sharp rise in 2018. Remarkably, this pattern closely resembles the fluctuation of NVA. Although extreme rainfall may also induce debris flows, from the observation data of the Linzhi meteorological station, the average maximum and minimum temperature from 15 August to 14 September in 2020 were 27 ◦C and 13 ◦C and the daily rainfall in this period ranged from 0 to 17.5 mm/d. According to an eyewitness video at 18:55 on 10 September 2020, the steel bridge deck was dry, which indicated that the precipitation was light on the event day (Peng et al., 2022). Therefore, combining the long-term climate trend and the daily value data before the event, we believe that the debris flow is caused by warming.

6. L105: Just visually. Fig. 2 doesn’t particularly look like precipitation is increasing at all. How did you calculate that increase?

As depicted in Figure 2b, the precipitation at Linzhi Station exhibited significant fluctuations from 2000 to 2021. The linear regression analysis revealed a minimal growth rate of rainfall, specifically 0.065 mm/year, which was not readily discernible from the raw data. To enhance clarity, we have included fitted trend lines for both temperature and precipitation in figures 2a and b.

7. L132: I would start the methods section with a summary sentence of the measurements that you are trying to make.

Thanks for your suggestion. We changed the structure of this section. We set the field surveys as a part of the methodology, and divided the original methodology section into two parts, they are NVA interpretation and Drone image interpretation. And we add a summary paragraph at the beginning of the sub methods section, as described below:

The study utilizes a combination of field surveys, aerial drone photography, and satellite imagery analysis to investigate debris flow events in the Zelunglung region.  High-resolution orthoimages and digital surface models are generated to assess terrain changes, while non-vegetated area (NVA) serves as a proxy for sediment volume for time series analysis.  The integration of these methods offers a detailed insight into the debris flow history and its influencing factors.

8. L159ff: I suggest to move the entire section on glacier changes before the methods. This is not your work as far as I understand, so it’s a bit odd to have that as part of the other results. You could move it together with the study area section

Thanks for your suggestion, we have integrated this section into the study area section.

9. L336: Can you explain where the interpretation of “some kind of dilute or hyper-concentrated flow” comes from?

Thank you for your review. We apologize for the oversight; this sentence was part of our original draft and was mistakenly left in, leading to a misunderstanding. We have now deleted it.

10. L351ff: A lot of that section reads like a discussion. I suggest to move it there.

Thanks for your suggestion, we moved the last 2 paragraph this section to the discussion section (new section 6.1), and named it with “The dominant factor for debris flows and sediment yield”. As a consequence, the section 6.2 was renamed to "The future risk".

11. Another note on this paragraph: You write of three surges (L168) but then only explicitly note two of them (L169ff and L172ff). I guess the third surge is the one you describe in L179? Can you make that explicit?

Thanks for your comment. Yes, the third surge occurred in 1984, and we wrote this date in the article.

12. Figures:

We re-examined and redesigned the all the figures according to the your comments.

13. Figure 12: Can you give some more information in the figure caption about where the ‘on-site’ sample is and when the sample was taken?

The ‘on-site’ sample was taken at the alluvial fan on September 11, 2020 (the day after the 2020 event). We have included this figure into figure 8, and numbered it as figure 8d. The location of the on-site sample was marked in figure 8a-1.

Thanks for your advice! According to your comments, we have made the following modifications and explanations:

1. Further considerations are needed to strengthen the conclusions, particularly the conclusion referring to the influence of the 1950 earthquake in further debris flows until 1990.

Thank you for your appreciation! As you mentioned, the debris flow activity in glaciated areas is more complex than in non-glaciated areas. Our case study attempt to shed light on how tectonic and climatic factors influence sediment evacuation processes via debris flows. Lin et al. (2006) presented that it could last for more than 10 years after the Chi-Chi earthquake. Cui et al. (2011) predicted that the effect of Wenchuan earthquake on post-quake debris flows would last for 10-20 years, while Huang (2011) predicted that it would be from 20 to 25 years. Recent publications demonstrate that only ~ 30% portion of coseismic sediment has been evacuated by debris flows and fluvial transport 12 years after the Wenchuan earthquake (Dai et al., 2021), and the fine-grained landslide sediment mobilized by the earthquake may stay in the affected river catchments as long as a century (Wang et al., 2015). Compared with the 1999 Chi-chi earthquake and the 2008 Wenchuan earthquake, the influence period of the 1950 earthquake is longer. We think the earthquake effects last longer in glaciated areas.

Thus, we modified the third point of the conclusion and added a sentence: The non-vegetated area of the Zelunglung’s fan reduced from 0.78 km² in 1950 to 0.067 km² in 1990, and kept at a stable low value until 2020, indicating the influence of the 1950 earthquake on the debris-flow sediment transportation could last 40 years. Compared with the 1999 Chi-chi earthquake and the 2008 Wenchuan earthquake in non-glaciated areas, the influence period of the 1950 earthquake is much longer.

1. The presented historical aerial imagery is a valuable source of information that could be explored to strengthen the manuscript. For example, if available, the source and affected area of the historical periglacial debris flows (1950,1968,1972,1984) could be identified.

We show our gratitude to your valuable feedback. Regrettably, we lack the necessary resources to support this research. Obtaining high-resolution images of both the source and affected areas, especially from immediately before and after the events, presents significant challenges due to the lack of advanced satellite remote sensing technology during that time period. Apart from the 1972 event, the occurrence of the other three events is well-documented in previous literature (as we quoted in my manuscript).

1. I missed in the manuscript a clear explanation about why you consider that the 1950 earthquake has a stronger influence than the glacier surges in the triggering of the debris flows after 1950 and, therefore, impacting the vegetation changes between 1950 and 1990 (line 484). In lines 187-188, it is stated that the debris flows were triggered by the glacier instability. Also, the 1950 debris flow coincides with a glacier surge.

As Zhang and Shen (2011) have stated: “The Chayu violent earthquake (Ms 8.5, on Aug 15, 1950) evoked Zelongnong glacier surge (Zhang 1985). After the rapid motion the ice cube carrying a mass of solid matter moved to the lower reaches. Ice block thawed and evolved into debris-flows during this process, a great deal of ice block and grit deposited at the convergence mouth.” Therefore, the events triggered by the 1950 earthquake constituted a chain reaction of glacier surge and debris flow, wherein the glacial process was integral to the debris flow mechanism. The impact of such a seismic event does not dissipate suddenly. Pre-1990 debris flow events, in our view, represent a legacy of the 1950 earthquake.  While their occurrence may have been directly triggered by glacial advancement or extreme precipitation, the root cause is the continued understability of the glacier/materials caused by the great earthquake. Prior to 1990, NVA exhibited a consistent decline, whereas afterward, it displayed significant fluctuation and even a slight upward trend.  This implies that the aftermath of the 1950 earthquake persisted until 1990, after which debris flow resumed activity on a relatively minor scale, influenced by climatic factors. Therefore, we have modified the expression of lines 187-188: Four large-magnitude debris flows accompanied by glacier instability occurred in 1950, 1968, 1973, and 1984.

1. The photo interpretation from Planet Lab images shows an ice-rock residual under the detachment area of the 2020 debris flow. The authors conclude that the entrained volume is at least 16 times the initiated volumes (line 227), consistent with the 1.14 Mm3 previously presented by Peng et al., (2022). This is highly relevant for the calibration of Debris Flow models.

Thank you for your suggestion! Yes, this data can be used to calibrate ice-rock avalanche and debris flow models.

1. The proposed methodology involved a considerable amount of remote sensing data manual interpretation and general information on the uncertainty of the manual mapping of non-vegetation areas was given. Regardless, I missed a short sentence on the reconstruction of the DSM from the UAV surveys (e.g., software use, ground control points, alignment), and the uncertainty on the elevation change that propagates to the presented volumes (line 346).

The reconstruction and differencing of DSMs are carried out in Pix4DMapper and Arcmap10.8. Since we did not deploy ground control points during drone photography, we generated DSM and DOM of September 9 in Pix4DMapper, and then selected 20 relatively stable points that were not affected by debris flow events as control points in Arcmap with DOM of September 9 as reference. These control points were then used in Pix4DMapper to generate the September 11 DSM and DOM.

To determine the uncertainty for our UAV DSMs of difference (DoD) differencing result we follow methods outlined in Shugar et al. (2021). We identified a series of fifteen stable areas on old debris flow terraces adjacent to the valley floor (Mainly roads and unseeded farmlands) and retrieved the standard deviation of DoD values within these areas and used these to estimate a two-sigma DoD uncertainty.  The uncertainty was ±0.493 m.

We have included a description of above methodology in section 3.2.3 “Drone image interpretation”.

1. Line 130-131: Did you present results on the surveys in 2021 and 2022 in this manuscript?

We conducted three field surveys, as described in lines 68-69 of the manuscript. The initial two were scheduled one day before and one day after the 2020 event, on September 9 and 11, 2020, respectively. Predominantly, these surveys involved drone photography before the disaster and included drone photography, measurements, and sampling in the post-disaster assessment. The third survey took place on December 21, 2022, and was specifically designed to capture aerial photography of the ZLL basin's upper reaches, an area inaccessible in the prior surveys due to adverse weather conditions.  The outcomes of the 2022 survey are detailed in Figures 1d, 4, and 6c, providing insights into the upstream channel and the debris flow initiation zone.  We did not execute a field survey in 2021.  We offer our apologies for any confusion that our initial errors may have caused.

1. Figure 4: Can you add a north arrow and a scale? Maybe you can try to use a different font color. The years of the images are hard to see.

Thanks for your suggestion. We have added north arrows and scalebars in all sub-figures, and changed all the texts in to yellow to make it easier to recognize.

1. Figure 9: c) Which distance is presented on the x-axis? distance from the outlet?

The x-axis in figure c represents the distance from P1 to P2 (i.e., the farthest end of the alluvial fan) along the main channel in Figure a-1. NOTE: To distinguish them from T1 and T2 in Figure 5, we have modified the original labels in Figure a-1 to P1 and P2, respectively.

1. Figure 11: Could you please extend the cross-section before the bridge to include the deposition areas? You could also include some information on the deposited particle sizes to exemplify.

Following your suggestion, we have extended the cross-section before the bridge to include the deposition areas. As shown in Figure 10c, the left bank edge of the channel has significant deposition, while the right bank platform actually has some fine particles or slurry deposited, but the graphic result is weak erosion, which may be due to the bias between the two phases of DSM. Due to the limitation of DOM resolution and the fact that debris flow slurry and particles with small size are mainly deposited in this area, particles with particles smaller than 50cm are difficult to be separated from DOM, and only a few coarse particles with size ＞ 50cm are detected on the left bank edge, which are not enough to support our further analysis.

1. Line 341. The maximum erosive depth of 20.47m is in the main channel or correspond to lateral erosion? Please include the mean erosive depth at the channel.

The maximum erosive depth of 20.47m is in the main channel. The mean erosive depth at the channel is 4.17m (The calculation area is the upstream area of the Zhibai bridge as shown in Figure 10).

1. Line 346. Could you please discuss how much underestimated is the volume you are presenting compared to Peng et al., (2022) and what is the expected error from the photogrammetric workflow?

Our estimated final deposit volume is 12.8×10⁴ m³, with an uncertainty of ±0.493 m according to DoD, and the uncertainty of deposit volume is ±1.85×10⁴ m³. Peng et al., (2022) estimated the final deposit volume is 37.5× 104 m³, and our result is 65.8% smaller than Peng et al., (2022). It is noteworthy that our DSMs were derived from data collected in September, during a period of high water levels in the Yarlung Tsangpo River. This situation could result in the neglect of certain sediment, potentially submerged by the water. In contrast, Peng et al., (2022) utilized data from December, a time characterized by low water levels in the Yarlung Tsangpo River. During this period, previously submerged sediments became exposed, and the area of the deposit, as delineated by Chen et al., was nearly twice the size of ours. This discrepancy accounts for the primary reason behind the underestimation of volume in our study. Furthermore, Peng et al. utilized data from the Ziyuan-3 satellite, which has a coarse resolution (2.5m). This lower resolution may also impact volume estimations.

We have included the above discussions in the end the 2nd paragraph of section 4.2.3.

1. Figure 12: The figure is not referenced in the text. Maybe you can merge Figure 12 and Figure 9 and include the location of the on-site sample.

Thanks for the reviewer's suggestion. We have included the graph of Cumulative grain size distribution into figure 8, and numbered it as figure d. The location of the on-site sample was marked in figure 8a-1. We referenced the figure 8d in the 2nd paragraph of section 4.2.3.

1. Figure 13: Can you add a north arrow?

Thanks for your suggestion. We added the scale-bars and the north arrows in the Figure.

1. Figure 14: Could you also include in Figure d) the other 4 debris flows for comparison?

Thanks for your suggestion. We have added the other 4 debris flows in Figure d for comparison.