the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 27 Oct 2025
Assessing the predictability of rockfall simulations constrained to simple objective input parameters
Abstract. Rockfall simulations are used to predict runout distances in case potentially unstable rock compartments would eventually fail. Transit simulated values such as the bounce heights and involved energies are also useful for hazard and risk assessments and for mitigation design tasks. However, it has been shown that the predictions from simulation results can vary significantly from user to user and from site to site. This highlights the need for simulation models with quantified accuracy and precision, low parametric subjectivity, and with good performance at predicting the transit values. The objective of this work is to present a validation methodology for rockfall simulation models and to objectively evaluate the predictive performance of stnParabel freeware simulation model when used with the Rolling friction rebound model. For this purpose, numerous mapped observations from a combination of back analyses of rockfall experiments and real events involving different remote sensing techniques were gathered. They cover twelve sites of various characteristics and geometries. The extensive collected observations include several hundred mapped deposited rock fragments of known dimensions and respective source locations. Each individual rock’s dimensions and masses were repetitively simulated without any other parameter adjustments in order to minimize the subjectivity of the simulation approach. In complement to the systematic objective process-based simulations, the runouts were also predicted for all sites with simulated trajectories from two additional process-based models for comparisons. Moreover, runout extents were also obtained geometrically with a commercial software and with a common geometrical approach for comparison. The results showed that the runout prediction accuracy from our process-based simulated trajectories is generally stable from site to site. Moreover, the runout precision of the simulations with stnParabel is improved by 2× to 3× compared to those of all other methods tested. And this is achieved with limited errors on the predicted transit values such as the bounce heights and translational velocities.
- Preprint
(10799 KB) - Metadata XML
-
Supplement
(11691 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-4635', Anonymous Referee #1, 26 Dec 2025
-
AC1: 'Reply on RC1', François Noël, 17 Feb 2026
Dear Reviewer,
We would like to sincerely thank you for the time and effort you dedicated to reviewing our manuscript. We are grateful that our work was found to be scientifically relevant and of clear practical value to the rockfall community. Your constructive comments and suggestions are highly appreciated, and we are confident that they will help us improve the quality of the manuscript.
We plan to address the remarks as follows:
1. Form and Content
We will significantly shorten the manuscript, primarily by reworking the Results section. Several figures will be moved to the supplementary material, and only an overview of the predictive performance results will be retained in the main text. Content from the Introduction that is not subsequently discussed will be removed, and the Methodology section will also be shortened by removing any repetitions.
2. Novelty
We will clarify the novelty of the present validation work, making a clearer distinction between contributions that are new and those building on previous studies.
3. Fragmentation and Forest
We will provide detailed clarifications on how fragmentation was emulated in the simulations and how fragments were mapped in the inventories. As forest cover was not systematically mapped in the inventories, it was neglected in the current analysis. We will explicitly state this limitation and discuss its potential implications. Given the large rock fragments and relatively thin forest at several sites, we believe the simplification has limited impact. Where possible, forest cover information from the datasets will be added to the site characteristics table.
4. Velocities and Bounce Heights
We acknowledge that conclusions regarding velocities and bounce heights cannot be generalized beyond the two sites where data were available. While this already exceeds the validation scope of most existing studies, we will ensure that this context is clearly stated whenever relevant conclusions are drawn. This will be contrasted with the broader spatial dataset of a dozen sites used for runout and lateral deviation validations.
5. Inventory Compilation and Limitations
Although these aspects are partially addressed in the Methodology and Appendix, we will add as much detail as possible to clarify the compilation process and its limitations, while still aiming to shorten the manuscript. Given the topic of the special issue, particular attention will be given to these points in the revised Discussion section.
Once revisions are complete, we will provide a detailed response outlining all changes made in response to your comments.
Thank you once again for your insightful and constructive review.
Sincerely,
François Noël and Synnøve Flugekvam Nordang
Citation: https://doi.org/10.5194/egusphere-2025-4635-AC1
-
AC1: 'Reply on RC1', François Noël, 17 Feb 2026
-
RC2: 'Comment on egusphere-2025-4635', Anonymous Referee #2, 29 Dec 2025
General comments
The manuscript is about an interesting topic and the problem itself is relevant.Also the results could be useful. However, in the present form, the paper is very difficult to read and it is not always clear what the main contribution really is.
The first issue is the length of the manuscript. The paper is very long and, in my opinion, this is not justified by the scientific content. Many parts are repeated, figures are many and often similar, and there is a lot of overlap between methods, results and discussion. Because of this, reading the paper becomes tiring and the main messages are not easy to follow. I think the manuscript should be strongly reduced, probably by about half.
This problem is related to the structure of the paper. At the moment, the organization is confusing. Methods, results and discussion are often mixed together. In the Results section, there are many comments and evaluations that look more like discussion. On the other hand, in the Methods section there are long explanations that are not always useful, while some important concepts are treated very fast. A clearer structure would help a lot.
The Results section contains too many figures. Several figures show very similar information and do not really add new elements. The captions are very long and often repeat what is already written in the text. This makes the manuscript even longer and harder to read. Fewer figures and shorter captions would improve the paper
Another point is the repetition of the research questions. At the beginning of the Results section, research questions are presented again, even if they were already defined in the Introduction. This is confusing and not necessary.
The Introduction is also very long and includes figures. This is unusual and, in this case, not very convincing. The Introduction should be shorter and more focused on explaining the problem, the aims of the study, and what is new in this work.
About novelty, this is a very important issue. It is not clear what is really new compared to the authors’ previous papers. The authors say that a paper from 2025 already included validation on 11 additional sites, and these sites seem to be the same shown here. Because of this, it is difficult to understand what the added value of the present manuscript is. This point needs to be clarified.
The Methods section also needs revision. Even if it is long, it does not always help the reader. Some key ideas are explaied in a very short and superficial way. For example, the part where simulated trajectories are converted into hazard zoning using probabilities, frequencies and thresholds is very important, but it is described too quickly. At the same time, other parts of the Methods section are repetitive and could be shortened.
Another issue is related to the special issue. The paper is submitted to a special issue on landslide inventories, but the connection with inventories is weak. The modelling results are not clearly discussed in relation to the quality, completeness and methods used to build the inventories. Given the topic of the special issue, this aspect should be better addressed.
If it is not possible to strongly reduce and reorganize the manuscript, the authors could consider splitting the work into two papers, one more focused on methods and validation and one on applications. However, this should be considered only as a second option.Additional general remark
I do not list minor comments here, because many small issues will probably be solved during a strong revision of the manuscript. However, I note a general problem with wording. The manuscript often uses expressions like “our model”, “our impact detection algorithm” and “our methodology”. It is not clear what “our” refers to. It could mean only the two authors of this paper, or also previous works with the same first author and different co authors. In addition: it refers to which algorithm? which model? These things should be clarified to avoid confusion and to better understand the contribution of this study.
Citation: https://doi.org/10.5194/egusphere-2025-4635-RC2 -
AC2: 'Reply on RC2', François Noël, 17 Feb 2026
Dear Reviewer,
We sincerely thank you for the time and effort you dedicated to reviewing our manuscript, particularly considering the challenges posed by its length and complexity. Your constructive comments are greatly appreciated and will contribute significantly to improving our work.
We are sorry to hear that the novelty of the validation study was not immediately clear. To our knowledge, no 3D rockfall simulation model has ever been systematically and quantitatively validated across multiple site inventories for runout, lateral dispersion, velocities, and bounce height predictions. In this work, we performed 60 comparison analyses involving 3D and 2.5D simulations against observational data from 12 sites, using three process-based and two geometrical models for the first two metrics. For rockfall dynamics, we also present what we believe to be the first quantitative validation using reconstructed 3D trajectories from two separate sites. These analyses were made possible by extensive inventories, including mapped events and reconstructed trajectories from both real-size experiments and documented rockfall events. We hope that these novel results will provide the rockfall research and practitioner communities with a clearer understanding of the strengths and limitations of commonly used simulation tools.
Based on your comments, we understand that the main criticisms relate primarily to the form and structure of the manuscript, while content-related concerns seem to arise from the complexity of the current presentation. We will therefore focus our revisions on improving the manuscript’s clarity, reducing redundancy, and shortening its overall length. Our planned revisions include:
1. Manuscript Length:
We will streamline the results by providing concise overviews of the validation outcomes, supported primarily by overview figures (histograms, pie charts, and precision/accuracy targets). Detailed figures focusing on individual metrics will be moved to the supplementary material, as is already the case for four of the five simulation methods. This approach will shorten the manuscript without compromising transparency. The discussion will be slightly expanded to better connect the results to the introduction, particularly regarding the influence of simulation parameters on prediction performance. Any topics not directly addressed in the results or discussion will be removed from the introduction to maintain conciseness.
2. Paper Structure:
We will ensure that each metric is clearly treated in the methodology, results, and discussion sections without unnecessary repetition. The restructured manuscript will be easier to follow and logically aligned with the validation objectives.
3. Figures and Captions:
Complex or repetitive figures with lengthy captions will be moved to the supplementary material. This will make the main text more accessible while preserving the comprehensive data for interested readers.
4. Methods Section:
We will revise the methods to focus on the comparison and validation procedures, removing content related to broader hazard assessment applications, which are beyond the scope of this study. Interested readers will be directed to Noël and Nordang (2025) for detailed discussions of hazard applications.
5. Connection to the Special Issue:
With these revisions emphasizing the extensive use of multi-site inventories to validate simulation models, some of which are commonly applied in practice, we believe the manuscript will more clearly align with the aims of this special issue.
After completing the revisions, we will provide a detailed response letter that outlines the changes made in direct response to your comments.
Thank you again for your thoughtful and constructive feedback. It will help us significantly enhance our work.
Sincerely,
François Noël and Synnøve Flugekvam Nordang
Citation: https://doi.org/10.5194/egusphere-2025-4635-AC2
-
AC2: 'Reply on RC2', François Noël, 17 Feb 2026
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 276 | 310 | 33 | 619 | 70 | 29 | 24 |
- HTML: 276
- PDF: 310
- XML: 33
- Total: 619
- Supplement: 70
- BibTeX: 29
- EndNote: 24
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
General
The manuscript presents a well-conceptualized and thoroughly executed study evaluating the predictive performance of rockfall simulations. The approach integrates the Rolling Friction Rebound model within the stnParabel framework and contrasts its outcomes with process-based and geometric methods, offering a methodological comparison. Validation is carried out across twelve sites.
Although rockfall modelling is a widely investigated topic, this work provides a valuable contribution by demonstrating that reliable predictions of runout distances, lateral dispersion, and kinematic metrics can be achieved without extensive site-specific calibration. However, the Methods section would need more concise explanations and a clearer distinction between novel contributions and previous work. Additionally, limitations related to inventory collection, vegetation filtering, and boulder fragmentation effects should be discussed in depth to contextualize the results.
Overall, the study is scientifically relevant and of clear practical value for rockfall hazard management, and it has strong potential for publication once the suggested clarifications and improvements are incorporated.
Specific Comments
Introduction
Line 115 – In this line, the authors already indicate that validation for the Mel de la Niva case was extended to 11 additional sites in their previous work (Noël & Nordang, 2025), among other objectives, to compare runout distances and assess the predictive performance of stnParabel under varying geometries and conditions. It would be recommendable to include a clarification that clearly differentiates what was achieved in the previous study and what constitutes the novelty of the present work.
Approach
Line 183 – In this line, it is mentioned that no artificial roughness or subjective adjustments to terrain material properties were applied, and the conclusions indicate that this was done to facilitate comparability. However, if this is the case, it remains unclear how the effect of vegetation was considered, particularly if any of the selected sites included areas with dense vegetation (no information is provided on this). If high or dense vegetation was present in any case, it is necessary to explain how this was considered, given its influence on perceived surface roughness and, consequently, on the model’s predictability, including runout distances and lateral spreading.
Line 208, 455 and others – Please clarify in the text how boulder fragmentation was considered in the trajectory comparisons. This clarification should also be applied to other sections of the manuscript, for example, explaining how this was addressed in the inventory and in the simulation.
Line 243 – The validation of velocities and bounce heights was performed at only two sites, as these are the only locations where such data are available. It is therefore important to discuss whether this limited sample is sufficient to support general conclusions, particularly when contrasted with the validation of runout distances, which considers a broader range of sites and geomorphological contexts.
Line 245 and 383 – For validation, the results are compared against Berger and Dorren (2006). However, given the strong influence of terrain model resolution on the analysis of velocities and bounce heights, it is essential to discuss the limitations and challenges of such a comparison. Berger and Dorren (2006) relied on a contour line–based model with a 2.5 m resolution, whereas the present study uses high-resolution DTMs with a 0.2 m resolution.
Simulation results and analyses
Lines 444, 610 and Appendix – The article should address the limitations associated with the inventory, as these may have contributed to deviations in some results. Furthermore, the Appendix reveals a wide range of approaches used to compile the inventory, from exhaustive field campaigns to image-based reconnaissance. The inventory was assembled by different institutions, which may have applied varying criteria during data collection, such as the minimum block size included or the spatial extent considered part of the studied rockfall event. All these aspects should be clearly discussed in the manuscript, explaining how these points were managed.
Appendix
As previously recommended, Tables A1 and A2 could also include brief information on the existing vegetation within the transit zone.