Next Article in Journal
Irrigation Requirement for Eucalyptus pellita during Initial Growth
Previous Article in Journal
Sediment Transport Mechanisms in a Lagoon with High River Discharge and Sediment Loading

Article
Peer-Review Record

Experimental Investigation of Flood Energy Dissipation by Single and Hybrid Defense System

Water 2019, 11(10), 1971; https://doi.org/10.3390/w11101971
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Water 2019, 11(10), 1971; https://doi.org/10.3390/w11101971
Received: 15 July 2019 / Revised: 13 September 2019 / Accepted: 17 September 2019 / Published: 21 September 2019

Round 1

Reviewer 1 Report

The authors used physical flume experiments to study different models of overbank flow defense systems with different combinations of simulated vegetation, dykes and moats.  As a statistician, I will focus on some of the statistical aspects of the experiment.

In several places--Equations (4)-(6), the researchers present regression equations and claim that they were developed "with 95% confidence interval".  The equations appear to be loglinear regressions, converted to the original response scale, but that information is unavailable in the paper; this should be made clearer.  I cannot tell if 95% confidence refers to confidence intervals for the regression curves, or the individual slopes/exponents for the two explanatory variables.  No confidence intervals are presented to help explain the claim; these issues need to be clarified and the referenced confidence intervals should be presented in the paper, or included as a supplement.

In Figures 10(b) and 10(c), the underlying relationship between the response and the explanatory variable appear to be curvilinear for DVI and DTMVS, respectively, suggesting the need for a nonlinear fit.  In fact, loglinear models were used in Equations (4)-(6).  The linear fits in Figures 10(a)-(c) are inconsistent with the loglinear models presented in Equations (4)-(6); why are nonlinear models presented in Equations (4)-(6), while linear models are used in Figure 10?

This is a minor point, but the x and y axes should have the same scale in Figure 11.  Given the pattern of the scatterplot in 11(a) and 11(c) in particular, a log-log transformation may provide a better presentation of the data (and lead larger r-squared values, incidentally).

The graphs in Figure 12 suggest tiny differences in treatments for given Froude numbers for both Delay Time and Relative Delay Time.  In order to judge the significance of these differences, standard error bars should be added to the graphs at the very least (multiple comparisons of pairwise differences within each level of Froude number would be a more thorough approach).  In the absence of any esimation or inference, categorical statements about the relative effectiveness of different treatments should not be made.

Author Response

Author Response File: Author Response.docx

Reviewer 2 Report

The theme of the manuscript is interesting, when dealing with the implementation of flood mitigation measures. And the results shown in the manuscript are interesting.

However, I observe two critical points in the work presented. First, I believe that the manuscript suffers from a section of critical discussion of the results obtained. This is so despite the fact that Section 3 is called "Results and Discussion", since throughout that section I only observe a description of the results obtained, but in no case do I observe a discussion of them, or in relation with the results or experiences of other authors. In addition, I consider that in general Section 3 needs a revision of its wording, because in some moments it is difficult to follow the line of argument of the authors

Secondly, the analysis raises a question that I have not been able to solve or understand throughout the whole work. In Figure 1d, the flow direction goes from top to bottom of the figure, and the fenders are located on the left side of the channel (according to the direction of flow). However, in the physical model described by the authors, the flow is directed transversely or perpendicular to the defense structures. That is something that does not fit me with the scheme in Figure 1d, and that I think can be a serious problem of approaching the physical model.

Apart from these basic aspects, there is another series of small comments (such as those related to the format of bibliographic references within the text) that are included in the attached document.

Author Response

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Point 1. The authors explain why confidence intervals are used; this new language is unnecessary and should be removed. They still fail to explain for which parameters they computed confidence intervals, and the manuscript does not provide any confidence intervals. The authors did try to respond, but they missed the point.

Point 2. I appreciate the authors’ updated graphics and fitted nonlinear curves.

Point 3. I appreciate the updated models and graphics.

Point 4. It is good to see error bars—the authors should make it clear whether the bars represent a single standard error in either direction or confidence intervals. While non-overlapping confidence intervals imply differences, you cannot say that overlapping confidence intervals imply no difference (a search on “overlapping confidence intervals” provides numerous references and counterexamples).

I do not see how the ANOVA test is a suitable substitute for pairwise comparisons within Froude number. It may be possible that pairwise comparisons aren’t feasible if there were not replicates within each level of Froude number, in which case I would withdraw this particular concern.

Author Response

Author Response File: Author Response.docx

Reviewer 2 Report

I consider that the manuscript has improved compared to its previous version. However, I think there are still a number of points that should be improved.

I consider that the quality and format of some of the figures are clearly improvable (Table 1, Figure 5, Figure 9, Figure 11, Table 3, Table 4). In fact, I consider that some of the tables do not adapt to the format proposed by Water.

Authors should review the format of bibliographic references, both within the text and in the corresponding epigraph. This is because my impression is that the format of some of the references does not comply with the format required by Water journal.

The authors have included a Discussion section that I believe was necessary. However, I think this section needs a revision. In addition, within the discussion the authors compare results obtained from the modeling of Tsunamis with the results obtained in the present work in a river environment. I don't think it's the best possible comparison given the differences in the processes associated with each environment.
The authors clarify which flow is what they intend to model. However, the modeling design I think would be adjusted only to very specific conditions. This is because the authors consider the direction of water flow as perpendicular to the channel at the time the river overflows its waters. In my opinion, in nature the direction of flow of the waters that overflow the channel is not perpendicular to it, but rather oblique, since the flow has a certain inertia of progress towards downstream. Therefore, I believe that only in a situation where the slope of the floodplain is negligible and the banks of the canal are already limited by defense works, it is possible that the direction of flow of water when overflowing the canal is perpendicular the same.

I think that these characteristics should be taken into account by the authors, and should be indicated in the introduction or methodology sections. In this regard, I believe that authors should try to try other configurations of their model. Or at least define under what conditions of flow and configuration of the channel the waters that overflow the channel will behave (in terms of their direction of advance) as proposed by the authors.

Author Response

Author Response File: Author Response.docx

Round 3

Reviewer 1 Report

Response 1. The authors apparently have a method for computing confidence bands for the best fit equation, rather than confidence intervals for parameters, but their approach is still unclear.  In particular, they report the width of the confidence interval as the confidence interval itself--this is a serious error.  Perhaps they have software that can compute confidence bands and include those confidence bands on Figure 10a, b, and c.  I.e., in addition to graphing the fit line, graph the confidence bands for each line as well.  And then simply report that you computed and plotted 95% confidence bands for each of your equations (4)-(6).

Response 2. Fine

Response 3. Fine

Response 4. The authors' second response has resolved the issue here; a brief and understandable explanation of the error bars has been provided.

Response 5.  The authors' second response has resolved the issue; pairwise comparisons were carried out and explained sufficiently well.

Author Response

Author Response File: Author Response.docx

Reviewer 2 Report

In my opinion, the manuscript has improved again in this second revision. Although some aspects of it do not quite convince me 100%, with special emphasis on what would be the behavior of the defense measures raised against a flow that reached them in an oblique direction to them. Well, I consider that this oblique direction is probably the most common in the overflow processes of a river channel.

Author Response