Non-Monotonic Relationships between Return Periods of Precipitation Surface Hazard Intensity

Round 1

Reviewer 1 Report

Manuscript Title: “Non-monotonic relationships between return periods of precipitation surface hazard intensity”

 

Overview:

The authors present a statistical analysis of the relationship between trigger and surface hazard intensities for South-Eastern Grand-Bay catchment within the Caribbean island of Dominica. The attribution of the results seems to be more theoretical-oriented rather than supported by the data. The topic addressed is of practical relevance; however, the manuscript has several shortcomings that need to be addressed before publication in the journal. I recommend a MAJOR REVISION.

 

Major Comments:

1. The introduction needs to be revised. In the present form, it lacks proper linkages between paragraphs, which needs to be improved. Sub-section 1.1 can be removed and that paragraph can be merged in Section 1 only.
2. Section 2.1: Why only GEV distribution was considered? Other probability distributions were not evaluated?
3. Page 7 Lines 164-166: Have the authors performed analysis to make the claims reported in this sentence or is that statement being cited from the previous studies?
4. Are Figures 4 and 5 representing similar results? If so, then remove Figure 5 and mark those locations in Figure 4 itself.
5. Figure 6: It would be better if a bivariate scatter plot is adopted to represent the relationship between hazard and precipitation return periods and fitted with a trend line (linear or non-linear) for better visual inference and comparison. The present form (as time series plot) seems to be inappropriate. Also, what is the rationale behind the comparison of sequential locations (such as 1 and 2, 3 and 4 and so on) considering that some locations (such as 7 and 8, 3 and 4) do not appear to be spatially dependent or related?
6. Figures 7 and 8: The results indicated in these figures can take only discrete (five) values mentioned in the respective figures. For instance, in Figure 8, can’t the rank correlation be 0.25 or 0.33? A stretch scale bar must be adopted for representation in such a case.
7. Conclusions: The authors state that the relationship between trigger and hazard intensity deviates in the areas prone to landslide dam and reservoir formation. This claim does not have any statistical confirmation from the analysis reported in this study.
8. The authors must carefully examine the manuscript for spelling errors and correct them. Also, the manuscript needs language correction.

 

Minor Comments:

1. Page 2 Lines 25-26: Rephrase the sentence for better clarity.
2. Page 2 Lines 29-30: The description in the parenthesis should be only never decreasing. Remover ‘monotonic’ from the parenthesis.
3. Page 2 Line 30: Correct the word ‘relationshused fip’.
4. Page 2 Line 32: It should be ‘relationship’ (singular).
5. Page 2 Line 34: It should be ‘increase’.
6. Page 3 Line 79: It should be ‘volcanic’.
7. Figure 1: The figure on the left shows DEM in the backdrop but the legends for the same is missing. The scale of all three sub-figures shown are the same but their dimension appears to be different. Please check. Also, the notations for soil class may be expanded in full for better clarity.
8. Page 4 Line 106: What is full GEV curve? It is just a GEV curve. Please correct it.
9. Figure 2: In the caption, the authors write Gumbel distribution, while in the text they mention GEV. Correct the mismatch.
10. Figure 3: The abbreviations (such as NDVI, LAI) listed in the figure must be explained. Two arrows are connecting flow properties and two-phase flow. Please check the flow of activities.
11. Table 1: There are several abbreviations that are not expanded at their first occurrence in the text.
12. Table 1: Check the description written against the soil depth. Correct it.
13. Figure 4: The coordinates shown are local coordinates. The authors must express their maps in projected global coordinates. Scale bar is also missing. Same for other figures, wherever applicable.
14. Page 7 Line 161: It would be beneficial to mention the year in which hurricane Maria occurred.
15. Page 10 Line 204: It should be ‘2 and 7 years’.
16. Figure 7: The values in the legends represent what?
17. Page 13 Line 240: It should be ‘shown in Table 3’.

Author Response

We would like to thank the reviewer for the helpful comments. We have adapted the manuscript according to the requested changes and comments, and hope these changes will be considered sufficient.

1: The introduction has been adapted for a more clear link between paragraphs.
2: While we attempted to use other probability distributions (weibull, Gumbell in particular) those methods showed a poor fit to data due to the presence of several types of storm systems: Tropcial cyclones and non-cyclone storms. The general extreme value distribution was, as is considered in the literatue, the best option for such a dataset, and this method was consequently used. See also:

Papalexiou, S. M., & Koutsoyiannis, D. (2013). Battle of extreme value distributions: A global survey on extreme daily rainfall. Water Resources Research, 49(1), 187-201.

3: These causes of slope instability are observed in the physically-based model (as opposed to failure due to seismic activity, lateral ground water movement, or deep-seated landslides due to structural weaknesses, the last two of which are in principle viable to occur within our simulations).  Field observations of shallow soil slips, and heavy over-saturation support these causes for slope instability.
We have clarified this in the text. If desired, we could add pictures of the slope failures to the article. For length considerations, we did not do so yet.

4: The requested change was made

5: The rationale behind the double locations was to provide at least 2 examples of each type of trend. This would provide better evidence. We have altered the figure to a scatter plot, with trend line.

6: The data was plotted with continuous values, but the legend was (because of software limitation) discrete. We have tried to clarify this with the legend.

7: This conclusion was drawn from figure 8 in particular, and the locaitons where the correlation is smaller than 0.5 or negative. To provide a more objective verification if this, we calculated correlation with landslide locations, which were manually mapped from high-resolution sattelite imagery.

8: The manuscript was edited for language and grammer mistakes.

All specific minor comments were adressed in the manuscript.

Reviewer 2 Report

I think that temporal changes and the effect of time should discuss more.  Intensity-duration-frequency (IDF) is the main relation among rainfall characteristics, In this study, is the volume of rainfall (or flood) major consideration?
In figure 3, what is the final output? Probability was used in which step? 
The estimation of vulnerability is not clear.
Figures 3 and 4 could be merged.
What are the colors (legend) in fig 7? 
Based on selected return periods, is it result in that return periods more than 100, the hazard is definitely?

Author Response

Dear Reviewer,

Thank you for the helpfull comments and your review.
We have adapted the manuscript to all your suggestions and feedback.

Please find the new version attached.
We hope the changes are considered sufficient.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The authors have addressed the comments. However, since the authors' rebuttal does not provide the details of modifications (e.g., Page and Line numbers) made in the revised submission, it is difficult for the reviewer to identify what changes were made vis-a-vis the previous version. Overall the revised version of the manuscript reads well and can be accepted for publication after receiving clarification on the suggested minor corrections. My queries are:

1. Figure 5: Why the value of peak flow height for the return period = 29 years is zero at all locations? This seems to be unusual. Check it.
2. Any limitations of the study must be highlighted at the end of the discussions or mentioned in the conclusions section.

Author Response

Dear Reviewer,

Our apologies for the confusion on the revisions.
We were not aware of the option to submit a highlighted manuscript to the journal.

For now, we have adressed your comments.
Please find a new figure 5 (the data for rp 29 was indeed incorrectly absent).
Additionally, at the end of the discussion a paragraph was added referring to limitations of the method.

Kind regards,
The authors