Characterizing geologic and climatic controls on rockfall hazards using an inventory and integrated kinematic and runout model: Skagway, Alaska, USA

Wachino, Ian D.; Roering, Joshua J.; Cash, Reuben; Patton, Annette I.

doi:10.5194/nhess-26-1435-2026

Articles | Volume 26, issue 3

https://doi.org/10.5194/nhess-26-1435-2026

© Author(s) 2026. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/nhess-26-1435-2026

© Author(s) 2026. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 26, issue 3

Research article

|

20 Mar 2026

Research article |

| 20 Mar 2026

Characterizing geologic and climatic controls on rockfall hazards using an inventory and integrated kinematic and runout model: Skagway, Alaska, USA

Ian D. Wachino, Joshua J. Roering, Reuben Cash, and Annette I. Patton

Download

Final revised paper (published on 20 Mar 2026)
Supplement to the final revised paper
Preprint (discussion started on 01 Apr 2025)
Supplement to the preprint

Interactive discussion

Status: closed

RC1:
'Comment on egusphere-2025-1168', Dominik May, 30 Apr 2025

The manuscript offers a promising work towards a quantitative but simple assessment of rockfall sources that has potential to be valuable in geomorphology and natural hazards. There are several aspects that should be substantially improved in order to exploit the potential of this work. The details to be improved are listed in the supplement and can be summarized as a need for a more conceptual and quantitative framework in order to make the work easier to reproduce and to underpin it with a stronger numerical foundation.

Citation: https://doi.org/10.5194/egusphere-2025-1168-RC1
- AC1: 'Reply on RC1', Joshua Roering, 28 Jun 2025
  
  These reviewer comments are exceptionally comprehensive and incredibly helpful in terms of improving details as well as clarifying the big picture. Many, many thanks.
  In the attached pdf file, we've added comments to each individual reviewer comments that summarizes our response and plans for addressing the issues raised.
  
  Citation: https://doi.org/10.5194/egusphere-2025-1168-AC1
RC2:
'Comment on egusphere-2025-1168', Anonymous Referee #2, 26 May 2025

The manuscript is well organised and, except for missing information about local places in maps, also very well presented work focused on the rockfall hazard of a small area around a turistically important town in Alaska. The used methods are standard with widely used approach and software used for rockfall hazard assessment. Nevertheless, authors failed to provide enough evidence about how representative the structural data obtained from the limited part of the study area is for the rest of it. The literature review is missing some important previous work on this topic e.g., from Europe. The results combine geomorphological mapping with the rockfall hazard assessment in a meaningful interpretation, but have a nature of small-scale local study, which contribution to the research topic on a global scale is limited. Considering this and a lack of methodological innovation, the impact of this study for a global audience seems questionable.
Please consider the detailed comments in the attached file.

Citation: https://doi.org/10.5194/egusphere-2025-1168-RC2
- AC2: 'Reply on RC2', Joshua Roering, 28 Jun 2025
  
  Many thanks for the very helpful and constructive comments. The need to incorporate references from European studies is well taken and we've generated a revised literature review of the relevant work. With respect to the two major comments regarding representative joint/structural data and the need to move beyond a small-scale local study, we've done the following:
  
  1. we revisited the study area in June 2025 and gathered joint orientation data at 12 new locations that span the entire study area. in doing so, we're able to show the remarkable consistency of the primary joint orientations which support the approach we've taken here for the kinematic analysis.
  
  2. the comment about moving beyond a local study is well put and we agree that revisiting the primary research questions is necessary. To do this, we've rewritten the last introduction paragraph to focus on the conceptual approach and model novelty. We've also performed an additional analysis of the topography that uses wavelets to quantify variations in the characteristic wavelength of the topography. This analysis shows where and why rockfall propagation is diminished due to topographic controls on dissipation of energy during runout. In essence, particular wavelengths of topography are effective as dissipating rockfall runout. In addition, we've added a climate analysis of the AKDOT database to test controls on triggering. We've also more rigorously correlated the correspondence of toppling susceptible zones with downslope talus deposits. Taken together, these revisions and new methods should alleviate the concerns about the lack of generality of our results and instead emphasize innovative approaches to addressing hazard in a heavily visited tourist area.
  Please see specific comments to review comments in the attached pdf file.
  
  Citation: https://doi.org/10.5194/egusphere-2025-1168-AC2

Peer review completion

AR – Author's response | RR – Referee report | ED – Editor decision | EF – Editorial file upload

ED: Reconsider after major revisions (further review by editor and referees) (08 Jul 2025) by Andreas Günther

AR by Joshua Roering on behalf of the Authors (19 Aug 2025) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (27 Aug 2025) by Andreas Günther

RR by Dominik May (13 Sep 2025)

RR by Anonymous Referee #3 (29 Oct 2025)

Suggestions for revision or reasons for rejection

Stepping late and as an additional referee into this review process I think the authors did reply well to some of the original reviewer concerns, and the quality of the manuscript improved accordingly. However, I made some additional observations and think that this manuscript needs further revisions to become up to standards for possible publication:
Section 3.6: If I correctly understand, the authors measured 405 joint orientations and then tested each DEM pixel with each of this data for topple failure to express the terrain pixel susceptibility as the percentage from the 405 orientations fulfilling the kinematic criteria at this location? I am sorry, but I think this approach is inadmissible because you cannot use a failure analysis on individual joints (from which you suppose they are present in exactly the same orientation and number all over the area) for spatially distributed analysis. I would highly recommend to calculate (or at least estimate from the density plot) mean orientations of the two orthogonal joint sets together with a confidential cone (can be done with all stereonet softwares) and then do the testing with the DEM data using the mean orientations and their variance defined by the confidential cone. However, this would only be a global analysis neglecting that mean joint orientations (and sometimes also the number and presence of joint sets!) often vary throughout a study area. For a valid spatial analysis, a sufficient number of orientations of any joint set present at a particular location should be measured and mean orientations should be calculated at each sampling location. The spatial distribution of the mean orientations can then be used for kinematic testing. In this context, I wonder why the authors did not test for wedge toppling (as described in Günther et al. 2012) since it can be supposed that when having steep, orthogonal discontinuities, this mechanism may also be identified for some locations. Last, I think at least some justification of the applied residual friction coefficients (expressed as friction angles) of the joint sets for kinematic testing must be given. This could be done using literature values, or by applying some appropriate field analyses.
Section 3.7: I think some illustration of the data used for RAMMS block size parameterization should be given, even though cited. I am not sure if talus fragments are suitable for estimating rockfall trajectories, this must be justified.
Section 4.1: Was the inventory only exploited for rockfall triggering conditions? Couldn´t it be used for parametrization of RAMMS when doing some frequency/size statistics to estimate block sizes? Why is the inventory lacking information along the railway line (except for one location)?
Section 4.3: This is not convincing, as also stressed above. In addition to the above: how did you identify the clusters? Why are some shallow dipping and moderately dipping discontinuities only present in the synoptic diagram? Why did you exclude sample locations from the fabric analyses (as in Fig. 8 some sectors are plotted but many sample locations are outside of those)? What is the rationale behind the “heat map” (as these only shows how many data was measured and you may always measure more)? Please also plot the traces of major faults in your maps since it is often the case that joint spacing reduces and persistence increases in the vicinity of fault zones, enhancing rockfall susceptibility.
Section 4.4: I think this is inadmissible. You cannot define a susceptibility index by calculating percentages of some measured orientations fulfilling failure criteria. To do so, you have to measure ALL joint orientations (at least at a specific observation scale and for a defined moving window where you assume bulk rock mass homogeneity) at many outcrops, this would result in an enormous workload. Alternatively (and statistically admissible) you may measure a number of orientations of each joint set present at a particular location and compute mean orientations and their scattering using some vector statistics available in common stereonet softwares. Then you can do some kinematic testing. In this context, I wonder why the authors did not collect data on discontinuity persistence, spacing, roughnes, fillings etc. This would be the data important for failure susceptibility evaluations.
Section 4.5.: I think it would be good to have some idea on the influence of the settings of the different RAMMS parameters. Please give some insights into the parameter spaces. Additionally, why are the RAMMS models not even capable for reproducing the only documented rockfall event along the railway line? I think the presentation is missing some (even if only qualitative) evaluations of the modelled rockfall trajectories.
To conclude, I think the paper needs additional work and data to become up to standards for publication in NHESS, or at least some major improvements of the applied analyses. However, I am aware that stepping into a running review process as a third referee is problematic, and the original referees have been more optimistic. Consequently, I would leave it up to the editor to decide if this manuscript has the potential for publication in a revised version.

Günther, A., Wienhöfer, J., Konietzky, H. (2012): Automated mapping of rock slope geometry, kinematics and stability with RSS-GIS. Natural Hazards, 61:29-49, DOI 10.1007/s11069-011-9771-2

Hide

ED: Reconsider after major revisions (further review by editor and referees) (10 Nov 2025) by Andreas Günther

AR by Joshua Roering on behalf of the Authors (02 Jan 2026) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (02 Feb 2026) by Andreas Günther

RR by Anonymous Referee #3 (17 Feb 2026)

Suggestions for revision or reasons for rejection

The authors provided an extensive reply letter together with their sparsely revised manuscript that incorporates a nice little review on probabilistic analysis of rock slope kinematics, which is, unfortunately, not very helpful for the justification of the applied approach characterizing the structural configuration of the area for distributed rock slope kinematic testing. Even though the authors reply that their kinematic analysis is (apparently?) not based on the assumption that all 405 orientations are present at each terrain location, their susceptibility analysis is since it is stated in the manuscript that “Specifically, we applied the stability criteria for each failure mode at each pixel in the merged lidar DEM by estimating the fraction of the 405 joint orientations that are predicted to be unstable given the topographic aspect and slope angle of that pixel” (p12, l296-298). This implies that at each terrain element nothing is expected to change but the slope orientation known from the DEM data. For a probabilistic analysis, the authors should at least estimate the probability of a specific orientation to occur at a certain terrain location (btw, I´ve checked the supplement and wonder what is reported with the listed 89 orientation data? Maybe you can check this and give some hint).
Anyway, just to make the point again: One may think of rock slope discontinuities as planar but they are not in nature. The authors may test this by conducting a number of orientational measurements on one large discontinuity. They will most probably receive a cluster of plane poles undulating around a mean orientation when plortted in a stereonet (a center of gravity of the unit vectors) – similar when measuring a number of discontinuities belonging to the same discontinuity set (or family). As a consequence, the orientation of a specific discontinuity set at a specific location can best be described measuring a suitable number of orientations of a specific joint set and then calculating a center of gravity and an associated confidential cone with an opening angel depending on the scattering of the data. In rock slope engineering and geology, we commonly use this information for kinematic testing (or the emplacement of stabilizing measures). However, for failure susceptibility analysis much more information on rock slope discontinuities would be necessary than solely orientation information.
To conclude, I must admit the authors failed to convincingly discuss their approach for structural characterization and failure susceptibility analysis given their rather unsystematic and very limited dataset, pointing out the limitations of the analysis and the necessity for further improvements. Therefore, I am sorry but I still cannot recommend publication of the paper in its current form.

Hide

ED: Publish subject to technical corrections (03 Mar 2026) by Andreas Günther

AR by Joshua Roering on behalf of the Authors (06 Mar 2026) Manuscript

Download

Article (16096 KB)
Full-text XML

Short summary

Rockfalls are a common hazard in steep mountain valleys, especially near Skagway, Alaska, where recent events have threatened public safety and infrastructure. This study identifies zones prone to rockfall by analyzing rockfall records, rock formations, past rockfall deposits, and computer models predicting how rocks travel downslope. Our findings highlight high-risk areas and provide insights to improve hazard mitigation, helping protect communities and tourism in the region.