Assessing Locations Susceptible to Landslide Initiation During Prolonged Intense Rainfall in the Lares, Utuado, and Naranjito Municipios of Puerto Rico

Baum, Rex L.; Brien, Dianne L.; Reid, Mark E.; Schulz, William H.; Tello, Matthew J.

doi:https://doi.org/10.5194/nhess-2023-185

Preprints

https://doi.org/10.5194/nhess-2023-185

Preprints

10 Nov 2023

| 10 Nov 2023

Status: a revised version of this preprint was accepted for the journal NHESS and is expected to appear here in due course.

Assessing Locations Susceptible to Landslide Initiation During Prolonged Intense Rainfall in the Lares, Utuado, and Naranjito Municipios of Puerto Rico

Rex L. Baum, Dianne L. Brien, Mark E. Reid, William H. Schulz, and Matthew J. Tello

Abstract. Hurricane María induced about 70,000 landslides throughout Puerto Rico, USA, including thousands each in three municipalities situated in Puerto Rico’s rugged Cordillera Central range. By combining a nonlinear soil-depth model, presumed wettest-case pore pressures, and quasi-three-dimensional (3D) slope-stability analysis we developed a landslide susceptibility map that has very good performance and continuous susceptibility zones having smooth, buffered boundaries. Our landslide susceptibility map enables assessment of (1) potential ground-failure locations, and (2) areas of potential landslide sources to support a companion assessment of inundation and debris-flow runout. The quasi-3D factor of safety, F₃, showed strong inverse correlation to landslide density (high density at low F₃). Area under the curve (AUC) of True Positive Rate (TPR) versus False Positive Rate indicated success of F₃ in identifying head-scarp points (AUC=0.84) and source-area polygons (0.85 ≤ AUC ≤ 0.88). The susceptibility zones enclose specific percentages of observed landslides. Thus, zone boundaries use successive F₃ levels for increasing TPR of landslide head-scarp points, with zones bounded by F₃ at TPR=0.75, very high; F₃ at TPR=0.90, high; and the remainder moderate to low. The very high susceptibility zone, with 118 landslides/km², covered 23 % of the three municipalities. The high zone (51 landslides/km²) covered another 10 %.

Received: 26 Oct 2023 – Discussion started: 10 Nov 2023

Download & links

Preprint (PDF, 15873 KB)

Supplement (20404 KB)

Download & links

Preprint (15873 KB)
Metadata XML
Supplement (20404 KB)
BibTeX
EndNote

Rex L. Baum, Dianne L. Brien, Mark E. Reid, William H. Schulz, and Matthew J. Tello

Status: closed

RC1:
'Comment on nhess-2023-185', Anonymous Referee #1, 22 Dec 2023

The authors are to be congratulated on an impressive study on assessing shallow landslide susceptibility in study areas in Puerto Rico. The study seems like the current tip of the iceberg of previous an ongoing research to analyse shallow landslides triggered by hurricane Maria in 2017. The authors present required input data including field records, a landslide inventory based on remote sensing imagery, compiled geotechnical data, a 3-step calibration procedure, a slope stability model with extened landslide geometry and new ways of presenting landslide susceptibility on large maps. I have to admit that I'm not familiar with most of the referenced studies addressing specific issues around landslides in Puerto Rico, which serve as the building blocks of the manuscript. With this in mind, my comments focus on general issues regarding slope stability modelling, datasets and paper structure. The language and the quality of the figures are excellent. The lessons learnt are summarized nicely, saving future modellers costly try & error time...
Where I see room for improvement is the paper structure, so I would recommend a minor revision. In the introduction section the objectives and main novelties should be outlined more clearly. The applied methods are well established, so more emphasis should be given to the comparison of the models used (soil depth and slope stability) and eventually the direct comparison of results based on the pre- and post-event derived LiDAR DTMs.
The M&M section should be re-structured. It is hard to connect the well-structured workflow chart in section 3.4 with the following subsections where the main processing steps are explained (despite the introduced step shortcuts). There are some overlaps (e.g. 3.4.1 Step 1, modelling soil depth VS. 3.8 Soil-depth modelling) which make it difficult to follow. Consider re-structuring without a third section-level and include subsections 3.1 to 3.3 in the workflow as well. Then you could describe the individual boxes of the workflow as individual subsections. Also the calibration workflow should be integrated more transparently. Consider summarizing the input parameter values/value ranges and property zones of each calibration/modelling step.
The results are presented well, please find minor comments annotated in the PDF. The discussion section is well elaborated, there are also only minor comments attached. The conclusions are fine.
Supplement material:

FIG_S1.jpg - a very nicely composed map; zooming in there are still linear artifacts which may result from tiling, but they do not interfere with the purpose of the map. Reference for REGOLITH in the box should be Baum et al. 2021

FIG_S2.jpg - reference for REGOLITH in the box should be Baum et al. 2021, consider increasing the map and reducing the size of the text around it

Fig_S3.pdf - how were the source areas mapped? In some cases they look like the runout areas are included

Citation: https://doi.org/10.5194/nhess-2023-185-RC1
- AC1: 'Reply on RC1', Rex Baum, 29 Dec 2023
  
  We thank Reviewer #1 for the thorough general assessment of the manuscript. We are pleased that the review is favorable overall.
  The suggestions about how to improve the introduction are helpful. Thank you.
  I agree that section 3, Methods and Materials, may be difficult to follow. The subsections (3.4.1, …, 3.4.4) to describe the major modeling steps were included to provide background on the models before discussing model calibration, but based on your comments, I see that those subsections create potential for confusion. Your suggested restructuring of section 3 to follow the flow diagram in fig. 6 seems straightforward and should help clarify the explanation of methods. Adding steps in sections 3.1, 3.2, and 3.3 to Figure 6 is a good suggestion. Adding those steps to the figure will require some careful thought about how to revise the figure to include them Thank you.
  Regarding the comment about greater transparency of the calibration process, we can provide additional data on the calibration, including parameter ranges and property zones considered. Such information may be well suited for a supplemental table.
  The suggestion to compare the results using the pre-event and post-event LiDAR is interesting. We expect the differences to be small between the pre-event and post-event cases. However, as we noted in lines 165-169, some mismatch exists between the pre-event and post-event LiDAR. We tried but were unable to obtain the additional data necessary to rectify the mismatch between the pre-event and post-event LiDAR. Thus, any quantitative comparison by means of differencing the pre-event and post-event factor of safety grids would have considerable uncertainty. Visual comparisons of small areas might be the best option since the viewer would tend to reference results to terrain features indicated by the hillshade, rather than the coordinate system. Comparing ROC results for pre-event and post-event topography though feasible, again is subject to some uncertainty arising from the apparent coordinate mismatch between the pre-event and post-event topography.
  Regarding the comments in the text about what to call the elevation grids used in our analysis, the LiDAR point clouds were filtered to represent the bare-earth terrain with uniformly spaced z-values. In the USA, we call such products “digital elevation models” and reserve the term “digital terrain model” for a vector dataset composed of 3D breaklines and irregularly spaced 3D mass points that characterize the shape of the bare-earth terrain (https://doi.org/10.3133/tm11b4, p. 48.) On this basis, we prefer to retain the term digital elevation model to describe our topographic data rather than changing to digital terrain model, which has two definitions. In our revision, we would add the clarification that we are using bare-earth digital elevation models with uniformly spaced z-values, created from the LiDAR point clouds, which are known in some countries as digital terrain models.
  Thank you for the many comments and questions pointing out minor clarifications and corrections needed throughout the text, figures, and supplement. We will plan to address those in our revision.
  
  Citation: https://doi.org/10.5194/nhess-2023-185-AC1
RC2:
'Comments on nhess-2023-185', Anonymous Referee #2, 16 Jan 2024

I have reviewed and carefully evaluated the manuscript “Assessing Locations Susceptible 1 to Landslide Initiation During Prolonged Intense Rainfall in the Lares, Utuado, and Naranjito Municipios of Puerto Rico”, which was submitted for publication in Natural Hazard and Earth System Sciences.
The manuscript focuses on a test site in Puerto Rico, where a physically based slope stability model is applied to assess the landslide susceptibility in the framework of an original multi-stage procedure.
The topic is relevant and well centered into the aims and scopes of the journal; the research design is appropriate and original. The manuscript is well written and would be a good contribution for the journal. However, I recommend some modifications before publication.

GENERAL COMMENTS
The introduction is very oriented on the specific test site. The feeling is to be reading a technical report. That would be normal for a “technical note” or a “test site description”, but I think NHESS publishes only research papers. Consequently, some improvements are needed in this part to better follow the standards of research papers. I recommend starting by stating a general problem and briefly drafting a state of the art on the topic. In your case, that should be the reliable application of physically based models (I would keep away from the huge literature about susceptibility studies based on statistical and machine learning methods: it’s a completely different approach). Maybe also a short focus on the use of hydrological, geotechnical and soil depth parameters in those models. (See e.g. Palacio Cordoba, J., Mergili, M., & Aristizábal, E. (2020). Probabilistic landslide susceptibility analysis in tropical mountainous terrain using the physically based r. slope. stability model. Natural Hazards and Earth System Sciences, 20(3), 815-829. - Medina, V., Hürlimann, M., Guo, Z., Lloret, A., & Vaunat, J. (2021). Fast physically-based model for rainfall-induced landslide susceptibility assessment at regional scale. Catena, 201, 105213. - Tofani, V., Bicocchi, G., Rossi, G., Segoni, S., D’Ambrosio, M., Casagli, N., & Catani, F. (2017). Soil characterization for shallow landslides modeling: a case study in the Northern Apennines (Central Italy). Landslides, 14, 755-770.). Afterward, I suggest clearly addressing some open research questions and highlighting the contribution of your work. I would move all the description about the site and the meteorological event in section 2 (e.g. the figure of the test site, by definition, should be in the section bout the test site). In the introduction I would just state that these features make a perfect case of study to test your original approach.
I think this is a very good paper, with just a few minor defects. One of them is that it is long and difficult to read at once. Also, it is affected by redundancies and repetitions, because of the chosen structure. Some sections (and related contents) should be reorganized. In chapter 3 some sections have 3rd level sections that are longer than 2nd level sections (e.g. 3.10 is only 4 lines). I suggest being as concise as possible when reviewing the manuscript, to adopt a linear and smooth structure that explains your the workflow you followed, avoiding repetitions.
The attention in incorporating soil depth into the model is noteworthy. I was puzzled to see that in some cases a constant-depth approach provides the best results, but you adequately discussed this outcome. This is a feature that should be highlighted in the introduction as many works about slope stability still focus on the geotechnical model, using oversimplifications to input soil depth (or other parameters). I recommend highlighting this in your improved introduction. I would also stress that your approach is (at least it seems to me by reading the paper) a good compromise because between the swift but simplistic approaches and the most complicated (thus, time consuming) approaches based on complex modeling schemes (see e.g. Xiao, T., Segoni, S., Liang, X., Yin, K., & Casagli, N. (2023). Generating soil thickness maps by means of geomorphological-empirical approach and random forest algorithm in Wanzhou County, Three Gorges Reservoir. Geoscience Frontiers, 14(2), 101514.)

SPECIFIC COMMENTS
L15-16: the difference between point 1 and point 2 is not clear at this time. I suggest rephrasing.
Section 3.2: as many data as possible are put together to have a consistent dataset for model calibration. This is correct in principle, but did you consider that the same parameters measured with different instruments or protocols may bring to problems of inconsistency and dishomogeneity in your dataset?
L224: computing how? This comes very later in the manuscript.
L526: by writing this (minor revisions to correct errors), you automatically trigger some curiosity… Can we have some example?
L618-629. This is confusing. I’m ok that above 60° you don’t’ have landslides… but maybe this is because soil does not form or does not stay in place in such steep conditions? You basically should set soil depth to zero here and verify this assumption…
L770 these three factors could be downsides (missing features) of your approach that could be implemented in future research improvements?
L801 this series of citations would be better placed in a state of the art review in the introduction (maybe with the addition of some recent works).
L872 Same as above.

Citation: https://doi.org/10.5194/nhess-2023-185-RC2
- AC2: 'Reply on RC2', Rex Baum, 22 Feb 2024
  
  We thank anonymous Referee #2 for carefully evaluating our manuscript and for the helpful suggestions to improve it. Referees #1 and #2 are in general agreement that the introduction and the methods section (section 3) are the main areas needing improvement. There seems to be sufficient overlap in their suggestions for improving these sections to make possible straightforward revisions that satisfactorily address both reviews. In addition, their specific comments throughout the text and supplements identified details that we can easily clarify.
  REPLY TO GENERAL COMMENTS
  Regarding Referee #2’s comments about the introduction, we agree that rewriting the introduction as suggested would provide a context for readers to understand the contributions of our work. The suggested outline (problem statement, summary of the state of the art, open research questions, and highlighting contribution of our work) is logical and appropriate. The five topics suggested for the state-of-the art summary (reliable application of physically based models, model parameterization, probabilistic landslide susceptibility analysis of tropical mountain terrain, fast physically based models for rainfall-induced landslide susceptibility assessment at regional scale, and soil-depth characterization for shallow landslide modeling) are relevant to our manuscript. We agree that descriptions of the site and meteorological event can be moved to section 2.
  Regarding length of the manuscript and structure of section 3. We agree that the paper is long and that the structure of section 3 is complex and may be difficult to follow. Referee #2’s suggestion to shorten section 3 and adopt a linear and smooth structure is consistent with the suggestions of Referee #1. We believe that improving the consistency between the workflow chart and the text of Section 3 as suggested by Referee #1 will help improve the structure, clarify, and shorten Section 3. We believe that doing so would help eliminate the redundancies noted by Referee #2.
  Thank you for the suggestions to highlight in the improved introduction both the incorporation of soil depth into the model and that our approach reaches a good compromise between swift, simplistic approaches and approaches based on complex modeling schemes.
  REPLY TO SPECIFIC COMMENTS
  L15-16: We will work to clarify the differences between the two points; however, that might not be fully achieved in the 200-word abstract. We are trying to describe two different applications of the susceptibility mapping, (1) assessing potential for ground failure which is hazardous in and of itself or (2) considering areas that have significant potential for ground failure as potential sources for debris flows, which create hazards for areas downslope. Since the assessments of inundation and debris-flow runout are treated in a separate paper, we acknowledge that the distinction may not be obvious here.
  Section 3.2, Thank you for raising the question about nonuniformity of measurements. We can briefly address consideration of differences resulting from instruments and measurement protocols.
  L 224. We can briefly explain that the factor of safety was computed using the infinite slope equation of Taylor (1948) and Iverson (2000).
  L526. An example of code revisions resulting from inspections is removing a code block that treated trial landslides centered on ridge crests and gently sloping ground differently from trial landslides elsewhere. This code block produced some spurious spots of low factor of safety identified by the map inspections.
  L618-629. Regarding soil depth on slopes greater than 60°. We can clarify with slight revision to the text here that the modeled soil depth is zero (or arbitrarily small) on slopes steeper than 60°. The discussion of soil strength parameters on lines 620-626 applies only where the slope is flatter than 60°.
  L770. We agree that the contributions of the three factors, soil suction, root resistance, and lateral stress to slope stability is worth investigating and can add a brief statement suggesting these as topics for future research.
  L801. Thank you. The suggestion to move this series of citations to the state-of-the-art summary in the improved introduction makes sense.
  L872. Yes, we agree that these citations would fit well into a state-of-the-art summary in the introduction.
  
  Citation: https://doi.org/10.5194/nhess-2023-185-AC2

Status: closed

RC1:
'Comment on nhess-2023-185', Anonymous Referee #1, 22 Dec 2023

The authors are to be congratulated on an impressive study on assessing shallow landslide susceptibility in study areas in Puerto Rico. The study seems like the current tip of the iceberg of previous an ongoing research to analyse shallow landslides triggered by hurricane Maria in 2017. The authors present required input data including field records, a landslide inventory based on remote sensing imagery, compiled geotechnical data, a 3-step calibration procedure, a slope stability model with extened landslide geometry and new ways of presenting landslide susceptibility on large maps. I have to admit that I'm not familiar with most of the referenced studies addressing specific issues around landslides in Puerto Rico, which serve as the building blocks of the manuscript. With this in mind, my comments focus on general issues regarding slope stability modelling, datasets and paper structure. The language and the quality of the figures are excellent. The lessons learnt are summarized nicely, saving future modellers costly try & error time...
Where I see room for improvement is the paper structure, so I would recommend a minor revision. In the introduction section the objectives and main novelties should be outlined more clearly. The applied methods are well established, so more emphasis should be given to the comparison of the models used (soil depth and slope stability) and eventually the direct comparison of results based on the pre- and post-event derived LiDAR DTMs.
The M&M section should be re-structured. It is hard to connect the well-structured workflow chart in section 3.4 with the following subsections where the main processing steps are explained (despite the introduced step shortcuts). There are some overlaps (e.g. 3.4.1 Step 1, modelling soil depth VS. 3.8 Soil-depth modelling) which make it difficult to follow. Consider re-structuring without a third section-level and include subsections 3.1 to 3.3 in the workflow as well. Then you could describe the individual boxes of the workflow as individual subsections. Also the calibration workflow should be integrated more transparently. Consider summarizing the input parameter values/value ranges and property zones of each calibration/modelling step.
The results are presented well, please find minor comments annotated in the PDF. The discussion section is well elaborated, there are also only minor comments attached. The conclusions are fine.
Supplement material:

FIG_S1.jpg - a very nicely composed map; zooming in there are still linear artifacts which may result from tiling, but they do not interfere with the purpose of the map. Reference for REGOLITH in the box should be Baum et al. 2021

FIG_S2.jpg - reference for REGOLITH in the box should be Baum et al. 2021, consider increasing the map and reducing the size of the text around it

Fig_S3.pdf - how were the source areas mapped? In some cases they look like the runout areas are included

Citation: https://doi.org/10.5194/nhess-2023-185-RC1
- AC1: 'Reply on RC1', Rex Baum, 29 Dec 2023
  
  We thank Reviewer #1 for the thorough general assessment of the manuscript. We are pleased that the review is favorable overall.
  The suggestions about how to improve the introduction are helpful. Thank you.
  I agree that section 3, Methods and Materials, may be difficult to follow. The subsections (3.4.1, …, 3.4.4) to describe the major modeling steps were included to provide background on the models before discussing model calibration, but based on your comments, I see that those subsections create potential for confusion. Your suggested restructuring of section 3 to follow the flow diagram in fig. 6 seems straightforward and should help clarify the explanation of methods. Adding steps in sections 3.1, 3.2, and 3.3 to Figure 6 is a good suggestion. Adding those steps to the figure will require some careful thought about how to revise the figure to include them Thank you.
  Regarding the comment about greater transparency of the calibration process, we can provide additional data on the calibration, including parameter ranges and property zones considered. Such information may be well suited for a supplemental table.
  The suggestion to compare the results using the pre-event and post-event LiDAR is interesting. We expect the differences to be small between the pre-event and post-event cases. However, as we noted in lines 165-169, some mismatch exists between the pre-event and post-event LiDAR. We tried but were unable to obtain the additional data necessary to rectify the mismatch between the pre-event and post-event LiDAR. Thus, any quantitative comparison by means of differencing the pre-event and post-event factor of safety grids would have considerable uncertainty. Visual comparisons of small areas might be the best option since the viewer would tend to reference results to terrain features indicated by the hillshade, rather than the coordinate system. Comparing ROC results for pre-event and post-event topography though feasible, again is subject to some uncertainty arising from the apparent coordinate mismatch between the pre-event and post-event topography.
  Regarding the comments in the text about what to call the elevation grids used in our analysis, the LiDAR point clouds were filtered to represent the bare-earth terrain with uniformly spaced z-values. In the USA, we call such products “digital elevation models” and reserve the term “digital terrain model” for a vector dataset composed of 3D breaklines and irregularly spaced 3D mass points that characterize the shape of the bare-earth terrain (https://doi.org/10.3133/tm11b4, p. 48.) On this basis, we prefer to retain the term digital elevation model to describe our topographic data rather than changing to digital terrain model, which has two definitions. In our revision, we would add the clarification that we are using bare-earth digital elevation models with uniformly spaced z-values, created from the LiDAR point clouds, which are known in some countries as digital terrain models.
  Thank you for the many comments and questions pointing out minor clarifications and corrections needed throughout the text, figures, and supplement. We will plan to address those in our revision.
  
  Citation: https://doi.org/10.5194/nhess-2023-185-AC1
RC2:
'Comments on nhess-2023-185', Anonymous Referee #2, 16 Jan 2024

I have reviewed and carefully evaluated the manuscript “Assessing Locations Susceptible 1 to Landslide Initiation During Prolonged Intense Rainfall in the Lares, Utuado, and Naranjito Municipios of Puerto Rico”, which was submitted for publication in Natural Hazard and Earth System Sciences.
The manuscript focuses on a test site in Puerto Rico, where a physically based slope stability model is applied to assess the landslide susceptibility in the framework of an original multi-stage procedure.
The topic is relevant and well centered into the aims and scopes of the journal; the research design is appropriate and original. The manuscript is well written and would be a good contribution for the journal. However, I recommend some modifications before publication.

GENERAL COMMENTS
The introduction is very oriented on the specific test site. The feeling is to be reading a technical report. That would be normal for a “technical note” or a “test site description”, but I think NHESS publishes only research papers. Consequently, some improvements are needed in this part to better follow the standards of research papers. I recommend starting by stating a general problem and briefly drafting a state of the art on the topic. In your case, that should be the reliable application of physically based models (I would keep away from the huge literature about susceptibility studies based on statistical and machine learning methods: it’s a completely different approach). Maybe also a short focus on the use of hydrological, geotechnical and soil depth parameters in those models. (See e.g. Palacio Cordoba, J., Mergili, M., & Aristizábal, E. (2020). Probabilistic landslide susceptibility analysis in tropical mountainous terrain using the physically based r. slope. stability model. Natural Hazards and Earth System Sciences, 20(3), 815-829. - Medina, V., Hürlimann, M., Guo, Z., Lloret, A., & Vaunat, J. (2021). Fast physically-based model for rainfall-induced landslide susceptibility assessment at regional scale. Catena, 201, 105213. - Tofani, V., Bicocchi, G., Rossi, G., Segoni, S., D’Ambrosio, M., Casagli, N., & Catani, F. (2017). Soil characterization for shallow landslides modeling: a case study in the Northern Apennines (Central Italy). Landslides, 14, 755-770.). Afterward, I suggest clearly addressing some open research questions and highlighting the contribution of your work. I would move all the description about the site and the meteorological event in section 2 (e.g. the figure of the test site, by definition, should be in the section bout the test site). In the introduction I would just state that these features make a perfect case of study to test your original approach.
I think this is a very good paper, with just a few minor defects. One of them is that it is long and difficult to read at once. Also, it is affected by redundancies and repetitions, because of the chosen structure. Some sections (and related contents) should be reorganized. In chapter 3 some sections have 3rd level sections that are longer than 2nd level sections (e.g. 3.10 is only 4 lines). I suggest being as concise as possible when reviewing the manuscript, to adopt a linear and smooth structure that explains your the workflow you followed, avoiding repetitions.
The attention in incorporating soil depth into the model is noteworthy. I was puzzled to see that in some cases a constant-depth approach provides the best results, but you adequately discussed this outcome. This is a feature that should be highlighted in the introduction as many works about slope stability still focus on the geotechnical model, using oversimplifications to input soil depth (or other parameters). I recommend highlighting this in your improved introduction. I would also stress that your approach is (at least it seems to me by reading the paper) a good compromise because between the swift but simplistic approaches and the most complicated (thus, time consuming) approaches based on complex modeling schemes (see e.g. Xiao, T., Segoni, S., Liang, X., Yin, K., & Casagli, N. (2023). Generating soil thickness maps by means of geomorphological-empirical approach and random forest algorithm in Wanzhou County, Three Gorges Reservoir. Geoscience Frontiers, 14(2), 101514.)

SPECIFIC COMMENTS
L15-16: the difference between point 1 and point 2 is not clear at this time. I suggest rephrasing.
Section 3.2: as many data as possible are put together to have a consistent dataset for model calibration. This is correct in principle, but did you consider that the same parameters measured with different instruments or protocols may bring to problems of inconsistency and dishomogeneity in your dataset?
L224: computing how? This comes very later in the manuscript.
L526: by writing this (minor revisions to correct errors), you automatically trigger some curiosity… Can we have some example?
L618-629. This is confusing. I’m ok that above 60° you don’t’ have landslides… but maybe this is because soil does not form or does not stay in place in such steep conditions? You basically should set soil depth to zero here and verify this assumption…
L770 these three factors could be downsides (missing features) of your approach that could be implemented in future research improvements?
L801 this series of citations would be better placed in a state of the art review in the introduction (maybe with the addition of some recent works).
L872 Same as above.

Citation: https://doi.org/10.5194/nhess-2023-185-RC2
- AC2: 'Reply on RC2', Rex Baum, 22 Feb 2024
  
  We thank anonymous Referee #2 for carefully evaluating our manuscript and for the helpful suggestions to improve it. Referees #1 and #2 are in general agreement that the introduction and the methods section (section 3) are the main areas needing improvement. There seems to be sufficient overlap in their suggestions for improving these sections to make possible straightforward revisions that satisfactorily address both reviews. In addition, their specific comments throughout the text and supplements identified details that we can easily clarify.
  REPLY TO GENERAL COMMENTS
  Regarding Referee #2’s comments about the introduction, we agree that rewriting the introduction as suggested would provide a context for readers to understand the contributions of our work. The suggested outline (problem statement, summary of the state of the art, open research questions, and highlighting contribution of our work) is logical and appropriate. The five topics suggested for the state-of-the art summary (reliable application of physically based models, model parameterization, probabilistic landslide susceptibility analysis of tropical mountain terrain, fast physically based models for rainfall-induced landslide susceptibility assessment at regional scale, and soil-depth characterization for shallow landslide modeling) are relevant to our manuscript. We agree that descriptions of the site and meteorological event can be moved to section 2.
  Regarding length of the manuscript and structure of section 3. We agree that the paper is long and that the structure of section 3 is complex and may be difficult to follow. Referee #2’s suggestion to shorten section 3 and adopt a linear and smooth structure is consistent with the suggestions of Referee #1. We believe that improving the consistency between the workflow chart and the text of Section 3 as suggested by Referee #1 will help improve the structure, clarify, and shorten Section 3. We believe that doing so would help eliminate the redundancies noted by Referee #2.
  Thank you for the suggestions to highlight in the improved introduction both the incorporation of soil depth into the model and that our approach reaches a good compromise between swift, simplistic approaches and approaches based on complex modeling schemes.
  REPLY TO SPECIFIC COMMENTS
  L15-16: We will work to clarify the differences between the two points; however, that might not be fully achieved in the 200-word abstract. We are trying to describe two different applications of the susceptibility mapping, (1) assessing potential for ground failure which is hazardous in and of itself or (2) considering areas that have significant potential for ground failure as potential sources for debris flows, which create hazards for areas downslope. Since the assessments of inundation and debris-flow runout are treated in a separate paper, we acknowledge that the distinction may not be obvious here.
  Section 3.2, Thank you for raising the question about nonuniformity of measurements. We can briefly address consideration of differences resulting from instruments and measurement protocols.
  L 224. We can briefly explain that the factor of safety was computed using the infinite slope equation of Taylor (1948) and Iverson (2000).
  L526. An example of code revisions resulting from inspections is removing a code block that treated trial landslides centered on ridge crests and gently sloping ground differently from trial landslides elsewhere. This code block produced some spurious spots of low factor of safety identified by the map inspections.
  L618-629. Regarding soil depth on slopes greater than 60°. We can clarify with slight revision to the text here that the modeled soil depth is zero (or arbitrarily small) on slopes steeper than 60°. The discussion of soil strength parameters on lines 620-626 applies only where the slope is flatter than 60°.
  L770. We agree that the contributions of the three factors, soil suction, root resistance, and lateral stress to slope stability is worth investigating and can add a brief statement suggesting these as topics for future research.
  L801. Thank you. The suggestion to move this series of citations to the state-of-the-art summary in the improved introduction makes sense.
  L872. Yes, we agree that these citations would fit well into a state-of-the-art summary in the introduction.
  
  Citation: https://doi.org/10.5194/nhess-2023-185-AC2

Rex L. Baum, Dianne L. Brien, Mark E. Reid, William H. Schulz, and Matthew J. Tello

Supplement

https://doi.org/10.5194/nhess-2023-185-supplement

Data sets

Engineering soil classification and geotechnical measurements in Lares, Naranjito, and Utuado, Puerto Rico R. L. Baum and A. C. Lewis https://doi.org/10.5066/P9UXTQ4B

Model input and output data covering Lares Municipio, Utuado Municipio, and Naranjito Municipio, Puerto Rico, for landslide initiation susceptibility assessment after Hurricane Maria R. L. Baum, D. L. Brien, M. E. Reid, W. H. Schulz, M. J. Tello, and E. C. Bedinger https://doi.org/10.5066/P9C1U0LP

Field observations of landslides and related materials following Hurricane Maria, Puerto Rico W. H. Schulz, E. K. Jensen, C. R. Cerovski-Darriau, R. L. Baum, M. A. Thomas, and J. A. Coe https://doi.org/10.5066/P9T9KZ6T

Model code and software

Slabs3D—A Fortran 95 program for analyzing potential shallow landslides in a digital landscape R. L. Baum https://doi.org/10.5066/P9G4I8IU

REGOLITH--A Fortran 95 program for estimating soil mantle thickness in a digital landscape for landslide and debris-flow hazard assessment R. L. Baum, E. C. Bedinger, and M. J. Tello https://doi.org/10.5066/P9U2RDWJ

Rex L. Baum, Dianne L. Brien, Mark E. Reid, William H. Schulz, and Matthew J. Tello

Viewed

Total article views: 289 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	Supplement	BibTeX	EndNote
206	61	22	289	44	16	13

HTML: 206
PDF: 61
XML: 22
Total: 289
Supplement: 44
BibTeX: 16
EndNote: 13

Views and downloads (calculated since 10 Nov 2023)

Month	HTML	PDF	XML	Total
Nov 2023	24	5	0	29
Dec 2023	62	16	7	85
Jan 2024	33	7	3	43
Feb 2024	34	15	1	50
Mar 2024	33	10	3	46
Apr 2024	20	8	8	36

Cumulative views and downloads (calculated since 10 Nov 2023)

Month	HTML	PDF	XML	Total
Nov 2023	24	5	0	29
Dec 2023	62	16	7	85
Jan 2024	33	7	3	43
Feb 2024	34	15	1	50
Mar 2024	33	10	3	46
Apr 2024	20	8	8	36

Viewed (geographical distribution)

Total article views: 272 (including HTML, PDF, and XML) Thereof 272 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 26 Apr 2024

Download

Preprint (15873 KB)
Metadata XML

Short summary

We mapped potential for heavy rainfall to cause landslides in part of the central mountains of Puerto Rico using new tools for estimating soil depth and quasi-3D slope stability. Potential ground-failure locations correlate well with the spatial density of landslides from Hurricane Maria. The smooth boundaries of the very high and high ground-failure susceptibility zones enclose 75 and 90 percent, respectively, of observed landslides. The maps can help mitigate ground-failure hazards.


Total:	0
HTML:	0
PDF:	0
XML:	0