Summary

In ‘Alpine hillslope failure in the western US: Insights from the Chaos Canyon landslide, Rocky Mountain National Park USA’, authors Matthew Morriss, Benjamin Lehmann, Benjamin Campforts, George Brencher, Brianna Rick, Leif Anderson, Alexander Handwerger, Irina Overeem, and Jeffrey Moore offer a comprehensive look at a large and rapid bedrock slope failure in the Rocky Mountains. The authors use a host of tools to better understand the morphology, kinematics, history, and triggering of the Chaos Canyon Landslide (CCLS), and argue that such events, which represent a potentially substantial and underappreciated hazard, may only become more common in the coming years. This manuscript offers a sort of blueprint for how to conduct reconnaissance studies of similar bedrock landslides occurring in alpine environments. However, despite all the methods thrown at this slide, the authors still struggled to outline why the CCLS failed catastrophically on June 28^th, 2022. Some of this uncertainty lies in the inherent complexity of large, creeping bedrock landslides, where the morphology and stress fields change from season to season. Moreover, some of the methods used in this paper proved more useful than others. The authors present a fantastic structure-from-motion model and accompanying morphology assessment, including precise volume estimates. The optical image-correlation based displacement history offers tremendous insight to pre-catastrophic-failure landslide behavior, and the careful historical analysis of climate and snowmelt shed light on the possible triggering mechanisms, and on how anomalous spring, 2022 was compared to past years. However, the Insar analysis and slope stability model did not seem to add much to the overall understanding of the CCLS. It was unfortunate that the Insar analysis, which could have been quite useful, suffered from unwrapping issues. I go into more detail in the following ‘main points’ regarding the slope stability modeling, but Slide2 is probably not the appropriate tool to study this type of slope failure, nor does the experimental setup of adding water to a slope with FS~1 broaden our understanding of the triggering mechanics. Finally, as I expand on in the following comments, the manuscript would benefit from a more targeted and evidence-based discussion of how the CCLS fits into the broader realm of cryosphere-related geohazards. Despite these critiques, this manuscript offers an important look at a landslide which may have broad implications for future geohazards. I recommend publication in Earth Surface Dynamics after addressing the following main points and line-by-line edits in the attached commented pdf.

Sincerely,

Dr. Sean Lahusen

Reviewer main points

1. The manuscript would benefit from more specificity and cited evidence when describing the nature of the hazard that the CCLS represents, how it fits into the global and regional records of glacier and permafrost ice loss, and in what ways we can meaningfully extrapolate that hazard to the rest of the continental U.S.

While I don’t expect the authors to be able to definitively assign the cause of the CCLS, I struggled to come away from the manuscript with a better understanding of the most likely primary underlying cause – permafrost degradation, unusually high snowmelt rates, or more inherent changes to the landslide morphology and shear surface degradation? All the above? Was anything exceptional about June 28^th, 2022? We can assume that many other fluctuations in pore-water pressure in past years had initiated or accelerated temporary pulses of CCLS deformation, so we know the slope was not in a stable configuration to begin with. And yet, these pulses of movement did not lead to catastrophic collapse in the past. A pertinent question seems to be: what was different on June 28^th? If you could prove the hydrologic conditions were historically elevated compared to past summers, this could be a convincing argument. However, it seems like the snowmelt totals where actually on the lower end compared to the last 5 years (Figure 6B). Another hypothesis the authors discuss is that the shear zone was in fact not fully developed, but had been developing over many years, and finally reached a critically weak state where an unexceptional increase in pore-water pressure was now sufficient to trigger catastrophic collapse. This seems like a reasonable hypothesis. Similarly, perhaps the landslide, which was perched on a steep slope, simply crept itself into a new, inherently unstable geometry on the slope where catastrophic failure was possible. Subtle geometric changes between the slide mass and the underlying slope can have a pronounced effect on landslide stability.

If permafrost degradation is indeed the primary conditioning mechanism for this catastrophic collapse, then to argue that these types of events are likely to become more common across the continental U.S. also requires some discussion of similar terrain in the continental U.S. How widespread is permafrost at sub-arctic latitudes? Obu et al. (2019) discusses this some, while also citing the earliest study of permafrost in the Rockies that I could find (Ives and Fahey, 1971):

'All permafrost zones occur in the Rocky Mountains of the USA with sporadic permafrost in Colorado at elevations above 3200 m asl and discontinuous permafrost above 3500 m asl, results which agree with observations by Ives and Fahey (1971). Isolated permafrost patches are modelled in the highest peaks of the Sierra Nevada in the southwestern USA, which corresponds well with the presence of active rock glaciers in the area (Liu et al., 2013)'

A major point of this paper is that events like the CCLS are only going to become more common. You mention a transformational period in alpine landscapes over the last few decades and cite 'changes to the ice glaciers of the world' – but what changes are we talking about, specifically - and what rates? This message comes off as overly-vague. Replacing some of this vague language with targeted evidence would be much more compelling. The global cryosphere has undergone enormous changes in the last 115kya, including numerous glacial advances and retreats in the last 20kya. Globally, most glaciers have been receding since the end of the LIA in the middle 19^th century. Undoubtedly, some of the presently-observed glacial retreat rates, permafrost degradation, and increases in alpine rock avalanche activity are due to anthropogenic climate change over the last century. However, if you want to argue the last few decades are transformational, I would reference some of the following work: Marzeion et al. (2014) show that only 25% of global glacial mass loss from 1851-2010 was due to anthropogenic causes – but this percentage nearly tripled to 69% for the period of 1991-2010. Hugonnet et al. (2021) show a dramatic acceleration in global glacial ice mass loss and thinning from 2000-2019. Furthermore, Christian et al. (2018) suggest that much of the effect of the last few decades of warming has already been baked into mountain glaciers in the form of ‘committed retreat’, which will result in continued dramatic glacial terminus retreat, even in the absence of additional warming. Cite more specific evidence and rates of this transformational period of change over the last few decade, because the idea of a transformational period strongly suggests we have left a long period of glacial and general cryosphere stability. If so, when was this period of stability – the end of the LGM? Following retreat from the Younger Dryas? The LIA (where many glaciers in saw substantial advances and subsequent retreats)? Another way I would think about this from a geomorphological perspective is: when was the last time a landscape experienced ice-free conditions?

Finally, perhaps it’s not appropriate to lump all cryosphere processes together – mountain glaciers like those that carved chaos canyon are inherently dynamic systems. Permafrost – maybe less so. How does this dynamic glacial history compare with the record of permafrost-underlain landscapes over the same timescale? Where are the changes in the last few decades most pronounced (especially in the lower 48, as this seems to be a main point of the paper)?

1. Mohr-Coloumb limit equilibrium slope stability models like Slide2 seem insufficient to model a creeping, accelerating rock mass

Landslides like the CCLS, which have been creeping intermittently for years with total displacement >10 meters, probably have pronounced strain-rate dependent behavior. Ideally, a slope stability model could account for (or at least approximate) this strain rate dependency, in order to capture the interaction between some of the following landslide characteristics:

1. Rate-hardening or rate-weakening frictional properties of the shear zone material
2. Material dilatancy vs contraction of the shear zone material
3. Evolution of drained vs. undrained conditions along the shear zone
4. Changing landslide geometry (>10 meters of displacement since the landslide initiation)
5. External factors influencing pore-water pressure, like snowmelt

Of all these important factors, Slide2, which is typically used to model the failure of intact rock slopes, can only model the last. Of course, no model can totally capture the intricacies of a natural slope, and even if it could, ascertaining these physical properties for the CCLS may prove impossible given the lack of instrumentation prior to failure. Still, in my opinion, Slide2 is too much of an oversimplification for this very complex slope failure, and is not the appropriate tool for modeling a creeping, rate-dependent bedrock landslide - especially if interstitial ice was involved. GLE models like Slide2 typically predict a binary outcome: total stability (no slope deformation whatsoever) or catastrophic failure. This CCLS does not exemplify this use case - in all likelihood the CCLS was creeping right before it collapsed. I can see how the authors tried to replicate this condition in Slide2 by using a realistic, empirically-derived geometry and tuning the model parameters such that the FS was just barely greater than 1.0, representing a slope on the verge of collapse.

However, a slope with a FS~1 in Slide2 does not represent an intermittently creeping landslide – it represents an intact bedrock slope that has experienced no displacement. Most problematically, the addition of pore water pressure (in the form of any amount of additional water perched above the slip surface) will invariably push a modeled slope with a calculated FS of ~1.0 into a state of instability. I don’t understand what value such an analysis adds to our understanding of the CCLS. If the authors could somehow prove the slide was particularly sensitive to small perturbations in pore-water pressure, this may be a more compelling analysis. For instance, you could add the same amount of water to the slip surface as the modeled snowpack melt volume (although this would bring up the question of why previous years with higher snow melt rates did not trigger the catastrophic collapse). To be clear, after reading your manuscript I am convinced that elevated pore-water pressure, likely from snowmelt, played an important role in triggering the June 28^th collapse, but Slide2 is an inappropriate tool to study this type of landslide and does not reinforce an otherwise compelling assertion.

1. The paper suggests tackling climate-driven alpine slope stability hazards on a national scale, but offers no new related data or analyses outside of the Chaos Canyon Landslide

I think the scope of this paper in its current form is totally sufficient and represents a fantastic addition to our understanding of an important landslide - potentially a harbinger of future, similar events. Unfortunately, the authors also suggest they intend to examine similar processes at the scale of the continental U.S. The author’s stated goals for this study are to: 1.‘Develop the tools necessary to analyze those events’ and 2. ‘Understand the stability of similar alpine slopes within a warming climate regime by characterizing the relation between permafrost, topographic and climate forcings, and slope instabilities.’ (lines 46-48). The manuscript title is also suggestive of this broader national-level analysis. While it is totally reasonable and very intriguing to speculate about extrapolating the implications of this single landslide to a national scale, this study does not offer any new analyses at that scale. Original data and analyses in this paper are confined to the CCLS. So, while such speculation is an important element of the paper’s discussion that should remain, it does seem like the paper promises but then does not deliver on new analyses or data pertinent to the broader hazard. I suggest subtle rewording throughout the manuscript to reflect this.

References Cited

Christian, J. E., Koutnik, M., & Roe, G. (2018). Committed retreat: controls on glacier disequilibrium in a warming climate. Journal of Glaciology, 64(246), 675-688.

Obu, J., Westermann, S., Bartsch, A., Berdnikov, N., Christiansen, H. H., Dashtseren, A., ... & Zou, D. (2019). Northern Hemisphere permafrost map based on TTOP modelling for 2000–2016 at 1 km2 scale. Earth-Science Reviews, 193, 299-316.

Hugonnet, R., McNabb, R., Berthier, E., Menounos, B., Nuth, C., Girod, L., ... & Kääb, A. (2021). Accelerated global glacier mass loss in the early twenty-first century. Nature, 592(7856), 726-731.

Zemp, M., Huss, M., Thibert, E., Eckert, N., McNabb, R., Huber, J., Barandun, M., Machguth, H., Nussbaumer, S.U., Gärtner-Roer, I. and Thomson, L., 2019. Global glacier mass changes and their contributions to sea-level rise from 1961 to 2016. Nature, 568(7752), pp.382-386.

Bolch, T., Kulkarni, A., Kääb, A., Huggel, C., Paul, F., Cogley, J.G., Frey, H., Kargel, J.S., Fujita, K., Scheel, M. and Bajracharya, S., 2012. The state and fate of Himalayan glaciers. Science, 336(6079), pp.310-314.

Marzeion, B., Cogley, J.G., Richter, K. and Parkes, D., 2014. Attribution of global glacier mass loss to anthropogenic and natural causes. Science, 345(6199), pp.919-921.

The manuscript by Matthew Morriss et al. provides an interesting case study of the Chaos Canyon Landslide (CCL) in Rocky Mountain National Park, including the character of the slope failure and its drivers.  The manuscript is valuable for its application of a range of observational and modeling techniques to understand the event.  The application of a wide range of techniques helps the authors to develop a comprehensive picture of this bedrock landslide, including the evolution of pre-failure creeping, and the volume of the slide mass.  The authors explore the connection between the CCL and climate change, a topic of scientific and land-management interest.  The authors generally do not overstate the significance of their results, which can only be speculatively connected to a climatic driver, but the manuscript would benefit from more critically assessing the potential mechanisms by which warming could have caused the CCL event.  In addition to this point, I have several questions about the methods used and how the results are integrated, although I find no problems that jeopardize the validity of the authors’ conclusions.  In general, the manuscript presents a valuable case study and methodology that I believe will be of interest to the community and is appropriate for publication in ESurf after revision. 

The weakest point of the manuscript is the connection between climate change and a CCL trigger.  The occurrence of the CCL near the hottest time of the year is compelling in suggesting (speculatively) that temperature played a role in triggering the event.  What is less clear is why the CCL occurred in the year that it did, which prevents clear establishment of a climate change-related mechanism.  There is not a straightforward relationship between the behavior of the CCL and annual temperature, as 2022 was not an exceptionally warm year compared to the previous ~10 years, and the data do not resolve when pre-failure creep began.  The authors explore both a reduction in interstitial ice and a meltwater-induced rise in groundwater as possible mechanisms for triggering the CCL.  The hypothesis of significant interstitial ice reduction seems inconsistent with the authors’ permafrost models, which indicate a maximum melting depth of <2 m in Summer 2022, an order of magnitude less than the depth of erosion (and thus minimum slip plane depth) indicated by the SfM analysis.  Thus, the vast majority of the slide mass would still have been subject to freezing temperatures immediately before failure.  The meltwater hypothesis is shown to be feasible through a simple factor of safety analysis, although the data and analyses presented don’t establish how the magnitude and rate of meltwater produced in the area are likely to have fluctuated over the years, preventing assessment of any temporal trends that might explain the timing of CCL failure.  It would be useful to see curves of annual precipitation and/or modeled meltwater production over time, if possible.  Of course, pre-failure deformation of the CCL mass may have contributed to the timing of failure more than the specific conditions of 2022, but it is difficult to attribute this pre-failure activity to climatic forcing without better constraints on when it began.  I don’t see this issue as a fatal flaw for the manuscript, and the authors generally do a good job of stating that a climate forcing mechanism for the CCL is speculative.  However, given the interest in this topic, I think the manuscript would be enhanced (and made more impactful) by a more in-depth discussion of potential climatic forcing mechanisms.

I have outlined some additional significant but less important points that would benefit from clarification below, in no particular order, with line edits following.  In addition to these points, the manuscript would benefit from a close rereading to identify typographic errors, ensure that figures are consistent with their captions, revisit the order of figure calls, and ensure that all figure panels are referenced in the text.  The authors should also consider making the field photos, analyzed satellite imagery, and SfM model available for the sake of reproducibility.

Other points

1. Composition of the CCL mass. The text is somewhat ambiguous as to whether the pre-failure mass was bedrock or regolith.  Section 1.2 states, “The slide occurred along the contact between the Middle Proterozoic Silver Plume Granite and the early Proterozoic biotite schists.” Does this mean that the slip plane is inferred to be the contact between these two units, or only that this is the geographic location where the slide occurred?  The foliations mapped on Fig. A1 have dips similar to or less than the stated pre-failure surface slope of 40 degrees, consistent with CCL slip along a pre-existing foliation plane.  Whether the pre-failure CCL material was bedrock or unconsolidated sediment would have implications for the failure mechanism—interstitial ice is probably less significant in igneous & metamorphic bedrock than sediment and foliation planes in bedrock might provide conduits for meltwater transport, so it is worth being more explicit about the composition of the slide mass.  If the bedrock is important, consider adding the geology to Figure 2.
2. Intercomparability of the image correlation results. The image correlation techniques have an important role in the study, producing the only results that establish pre-failure movement of the slide material.  Because of this, it would be valuable to see the image correlation results presented more systematically, including having similar figures for the different approaches taken with the Google Earth and PlanetScope imagery, such that the reader could evaluate the consistency of the results from the two imagery sources and distinct methods. 
3. Structured residuals in the SfM model. The difference model between the post-slide SfM model and the pre-slide topography (Figure 8) shows coherent differences outside the slide area.  Areas downslope from the slide have negative differences and areas higher on the slope outside the slide have positive differences.  I wonder if this reflects distortion in the SfM model, problems with registration to the DEM, or something else.  This should be addressed in the text, along with any implications for the eroded and deposited volume estimates.

Line edits

54-55 – reformat citation

73 – redefine ‘SfM’ as this is its first use in the body of the paper

125 – how was the environmental lapse rate calculated?

141 – how were the 305 photos collected from the 9 photo points (e.g., mosaic from each photo point location)?

238-239 – rephrase to communicate greater confidence/reproducibility, e.g., “independent measurements of displacement of large boulders identified visually in the images are consistent with displacement magnitude inferred from image analysis”

249 – change “1/velocity” to “inverse-velocity” or similar

253 – rephrase “out of the ordinary” to “atypical” or similar

263 – reintroduce what is being shown in this difference map and how it was obtained using a new topic sentence

277-278 – revisit for syntax, degree symbol

293 – add citations

315 – I suggest eliminating the first clause as it is very different from where the paragraph is going

346 – it appears that Fig. 12A only shows one mobility index, L/H

350-351 – incomplete sentence

352 – I think something is missing from the parenthetical note

354 – reformulate to avoid use of contraction

364 – citations needed or eliminate the reference to other scientists

Figures

3 – panels C, D not called or discussed in text

5 – what are the thin grey lines? How is velocity (panel B) calculated?  It appears somewhat different from what I expect based on the slope of the displacement measurements in A.  Also, are the points plotted at the date of the analyzed image each year?

6 – replace “or” in first line of the caption with “, ” if 3,668 m is the elevation of the top of the slide; revisit entire caption for spelling, capitalization.

9 – the date of the slide is stated as June 29, in contrast with June 28 in the rest of the paper; I suggest adding something to state that the beginning of the hydrological year (0 on panel B x-axis) is not the same as the beginning of the calendar year

10 – colors did not come through for the version I received; caption states that the thick dashed line is limit of landslide material—is all the highlighted material in A the pre-collapse material or not?

11 – define PDDS (both the acronym and how it is calculated); because snowmelt depends on both temperature and precipitation, it would be valuable to see annual precipitation plotted as well

A1 – include geologic unit symbols

I enjoyed reading the manuscript and think it will make a valuable contribution, and I encourage the authors to contact me with any questions or for clarification about the review.

Sincerely,

Alex Tye

alex.tye@utahtech.edu