Evaluating Curve Number Implementation Alternatives for Peak Flow Predictions in Urbanized Watersheds Using SWMM

Round 1

Reviewer 1 Report

This paper deals with the application of the well-known Curve Number approach for infiltration losses in urban catchments. In particular, the main goal of the work is to discuss it in relation to different choices that can be made to treat the level of imperviousness of the catchment. This issue has been addressed in the light of applications using two software (SWMM and WinTR-55).

Though the paper has some elements of practical interest, I don't think it is suitable for publication. In its present form, it looks more like a report of practical activities than a scientific paper. It can be viewed as a technical note, as considered in some international journals.

Basically, the main limitations of this work are reported below.

1. Purposes of the work: the authors reported them at lines 78-85. In my opinion, they are of very limited importance for the literature so that the novelty introduced by this work is not so significant. It would have been better to underline the soundness of the research in a framework much wider than the effects on SWMM. The question you raised (the use of fully-composite CN or different strategies) is not strictly linked to SWMM but it is more general. For example, I listed 5 papers at the end of my report in which this matter is faced from different kinds of applications and purposes. My suggestion is to rethink the introduction, underlying a "scientific" problem and not a "technical problem", otherwise the scientific interest of the work is limited. Please observe that the references are quite dated. The number of papers in the last five years is 3. This is a confirmation that your study can be affected by a lack of interest. I think that those references (and other similar ones you can find accordingly) could help you to improve the scientific soundness of your study. 

2. Discussion: a discussion is missing. This point is linked to the previous one. What is the interest of your study of literature? What are the expected impacts in this field of research and, mostly, the implications for other studies in which the CN approach has been used (again see references 1-5) ? I suggest adding a section like this, to put emphasis on the scientific value of your research. 

3. Conclusions: I cannot see general conclusions. First of all, they look like a (long) summary of the work. It is useless to repeat concepts already expressed in the introduction (see lines 390-411). Please substantially reduce this part and focus more on lines 411-414 and 420-426 that seem to me the most interesting ideas to stress. 

Therefore, I encourage the authors to made efforts to improve the overall presentation of the work, highligthing the novelty and the interest of this study for the related literature.

Minor comments:

please improve the quality of all the figures (except from figures 11 and 12)

Cited works

[1] DOI: 10.1007/s00477-021-02049-2

[2] DOI: 10.1016/j.jenvman.2022.116814

[3] DOI: 10.1016/j.jhydrol.2022.127870

[4] DOI: 10.1016/j.scitotenv.2021.151591

[5] DOI: 10.1016/j.jhydrol.2020.124704

Author Response

Though the paper has some elements of practical interest, I don't think it is suitable for publication. In its present form, it looks more like a report of practical activities than a scientific paper. It can be viewed as a technical note, as considered in some international journals.

Thanks for your comment, we agree that there are elements of practical interest that, to our knowledge, have not yet been published in a technical paper, such as a direct comparison between WinTR-55 and SWMM 5 predictions in the context of an urbanized watershed. We have performed changes, as detailed below, following your suggestions. The manuscript was submitted in the format of an “Article” for MDPI Water as the other options “Reviews” and “Data Descriptor” were not applicable.

Basically, the main limitations of this work are reported below.

1. Purposes of the work: the authors reported them at lines 78-85. In my opinion, they are of very limited importance for the literature so that the novelty introduced by this work is not so significant. It would have been better to underline the soundness of the research in a framework much wider than the effects on SWMM. The question you raised (the use of fully-composite CN or different strategies) is not strictly linked to SWMM but it is more general. For example, I listed 5 papers at the end of my report in which this matter is faced from different kinds of applications and purposes. My suggestion is to rethink the introduction, underlying a "scientific" problem and not a "technical problem", otherwise the scientific interest of the work is limited. Please observe that the references are quite dated. The number of papers in the last five years is 3. This is a confirmation that your study can be affected by a lack of interest. I think that those references (and other similar ones you can find accordingly) could help you to improve the scientific soundness of your study.

The authors would like to thank reviewer 1 very much for pointing to these references, which include various applications, from the response of watersheds using shallow water equation models to approaches to estimating flooding thresholds and extent. In common, all these recent studies have used CN in their methodology, and adding these to our reference section has improved our manuscript in contextualizing the recent application of CN in hydrologic studies. Pointing these studies to readers will enrich our manuscript. We also agree with Reviewer 1 that the application of CN is not strictly linked with SWMM and we have edited the manuscript introduction to reinforce this point. However, we would like to reiterate that study brings useful information to Water readers given the popularity of SWMM as a hydrologic model and the availability of CN data to compute infiltration.

Regarding the point of a scientific problem: in our own interactions with researchers and practitioners in the US that apply SWMM as their hydrologic model, the CN implementation is not entirely clear. As explained in our manuscript, the SWMM 5 Hydrology Manual outlines the procedure for CN use, combining pervious and impervious areas for computing CN. However, the manual also states that if someone partitions their subcatchments into pervious and impervious areas, they will have to adjust the CN values to remove the effects of imperviousness.” Such an approach is also indicated by seasoned SWMM modelers [Refs. 23-25]. To perform an objective comparison between these two approaches to assign CN to SWMM subcatchments, we proposed the idea of a CN Cut-off approach and the comparison between SWMM and the modeling tool that applied CN, the WinTR-55.

2. Discussion: a discussion is missing. This point is linked to the previous one. What is the interest of your study of literature? What are the expected impacts in this field of research and, mostly, the implications for other studies in which the CN approach has been used (again see references 1-5)? I suggest adding a section like this, to put emphasis on the scientific value of your research.

That is another good point. Our manuscript has grouped the “Results and Discussion” in a single section, in which we discuss and analyze the findings of the field measurements, and from the application of SWMM 5 and WinTR-55 models. We have added a text that attempts to articulate how the present study can contribute to the existing literature, as suggested by Reviewer 1.

3. Conclusions: I cannot see general conclusions. First of all, they look like a (long) summary of the work. It is useless to repeat concepts already expressed in the introduction (see lines 390-411). Please substantially reduce this part and focus more on lines 411-414 and 420-426 that seem to me the most interesting ideas to stress.

Thanks for your comment. We have summarized the conclusions and refocused the discussion following your suggestions and the ones from the other reviewers. We hope that the conclusions section in the new version meets your expectations.

Therefore, I encourage the authors to make efforts to improve the overall presentation of the work, highlighting the novelty and the interest of this study for the related literature.

Thanks for the constructive comments.

Minor comments:

please improve the quality of all the figures (except from figures 11 and 12)

The manuscript figures were extensively edited to improve their quality.

Cited works -

[1] DOI: 10.1007/s00477-021-02049-2

[2] DOI: 10.1016/j.jenvman.2022.116814

[3] DOI: 10.1016/j.jhydrol.2022.127870

[4] DOI: 10.1016/j.scitotenv.2021.151591

[5] DOI: 10.1016/j.jhydrol.2020.124704

Thanks again for the suggested references contextualizing the recent use of CN in hydrologic modeling. They are now part of the manuscript introduction.

Reviewer 2 Report

The authors utilize flow discharge and water depth for a small watershed in Southern USA to evaluate procedures to assign curve number (CN) values for different land uses using automated GIS procedures.  The authors consider the implications of defining impervious areas separately, or averaging CN assignments without specifically identifying impervious areas and specifying as impervious areas with CN greater than a threshold.

The authors present hydrographs and hyetographs computed primarily with the SWMM model.

The paper is fairly well written and does not require changes in that respect.

Looking at the presented results, however, it seems that the runoff volumes are not correctly modeled, as the tail end of hydrographs show a significant discrepancy (clearly shown in Fig. 9). It appears that it could be a combination of infiltration and roughness/storage or other factors that control hydrograph recession.

The main issue is how the definition of impervious areas for SWMM is considered in its calculations.  This is not clearly explained in the manuscript.  If it's just a shortcut to say CN = 100, if this defined as impervious or manually set to an equivalent value (98 as in some user manuals) should produce small changes that would negate the need for this study.  Clarification of this aspect is necessary.

The color schemes for figs. 6 & especially 7 do not let readers see differences in CN clearly.  Perhaps using a ramp with more colors would help.

Author Response

The authors utilize flow discharge and water depth for a small watershed in Southern USA to evaluate procedures to assign curve number (CN) values for different land uses using automated GIS procedures.  The authors consider the implications of defining impervious areas separately, or averaging CN assignments without specifically identifying impervious areas and specifying as impervious areas with CN greater than a threshold. The authors present hydrographs and hyetographs computed primarily with the SWMM model. The paper is fairly well written and does not require changes in that respect.

Thanks for the clear summary of our work and your comments.

Looking at the presented results, however, it seems that the runoff volumes are not correctly modeled, as the tail end of hydrographs show a significant discrepancy (clearly shown in Fig. 9). It appears that it could be a combination of infiltration and roughness/storage or other factors that control hydrograph recession.

Reviewer 2 is correct in stating discrepancies in the recession limb. We have attributed this discrepancy to the absence of the aquifer object in the SWMM models that were constructed, which is aligned with the hypothesis of storage (i.e., underground storage) by Reviewer 2.

The main issue is how the definition of impervious areas for SWMM is considered in its calculations. This is not clearly explained in the manuscript. If it's just a shortcut to say CN = 100, if this is defined as impervious or manually set to an equivalent value (98 as in some user manuals) should produce small changes that would negate the need for this study. Clarification of this aspect is necessary.

We are not entirely clear on what is implied in this comment. We believe you meant evaluating whether SWMM sub-catchments with CN=98 or fully impervious are comparable. While this is not the focus of the research, our results permitted us to compare such conditions. We also have edited the manuscript to explain better the implications of having SWMM with subcatchments with a fraction of impervious areas or with higher CN values but fully pervious. Evaluating the fraction of impervious areas in subcatchments typically involves visual evaluation of land use, which can be tedious and subjective. Rather than adopting this approach, we have proposed using large CN values as a surrogate to determine if an area is impervious or not. We hope that this version is clearer, and with that, we have addressed your concern.

The color schemes for figs. 6 & especially 7 do not let readers see differences in CN clearly. Perhaps using a ramp with more colors would help.

Thanks for your suggestion. However, the other color ramp options yielded by QGIS were not as clear as what we have presented, in our opinion. Yet, we have edited all figures in the manuscript to improve their quality, including those showing the CN distribution in the watershed. More importantly in Figure 6 (formerly Figure 7), the importance is to contrast the increase of white areas (i.e., impervious areas) with the lowering of the CN Cut-off values. This was highlighted in the current manuscript version.

Reviewer 3 Report

The authors used SWMM model associated with curve number method to calculate peak flow in urbanized watersheds. Agreement of computed results compared with field observations and those by WinTR-55 model shows the application of this model for the peak flow in urbanized watersheds. Some comments on the draft are list below.

1.    For Fig. 9 and Fig. 10, the computed peaks are examined to be slightly lower than the observations and have a time lag.  The underestimation statistics of the peak value of each watershed can be added to serve as a reference for flood protection and engineering design. Time delay is also an important factor for urban drainage, but it is not discussed in the draft. If it cannot be further analyzed, possible reasons can be deduced in the reversion.

2.    The performance by fully composite computation for the Hamilton Road sub-watershed in Table 5 and 6 shows the worst among all CN-computations. However, the performance by fully composite computation for the Champions Blvd sub-watershed are the best among all CN-computations. Should the result using the Fully Composite that averages CN values from all areas, including impervious areas with high CN values be between the results using three thresholds of CN Cut-off?

Comments for author File: Comments.pdf

Author Response

The authors used SWMM model associated with curve number method to calculate peak flow in urbanized watersheds. Agreement of computed results compared with field observations and those by WinTR-55 model shows the application of this model for the peak flow in urbanized watersheds. Some comments on the draft are listed below.

1. For Fig. 9 and Fig. 10, the computed peaks are examined to be slightly lower than the observations and have a time lag. The underestimation statistics of the peak value of each watershed can be added to serve as a reference for flood protection and engineering design.

Reviewer 3 is correct in stating discrepancies in the recession limb. We have attributed this discrepancy to the absence of the aquifer object in the SWMM models we created, which is aligned with the hypothesis of storage (i.e., underground GW storage) put forward by Reviewer 3.

Time delay is also an important factor for urban drainage, but it is not discussed in the draft. If it cannot be further analyzed, possible reasons can be deduced in the reversion.

Reviewer 3 is also correct in stating that, for different rain events, the models predicted an earlier rise of the hydrograph. Following the reviewer’s suggestion, we have provided a few possible explanations for the earlier hydrograph rise in the manuscript section 3.2.

2. The performance by fully composite computation for the Hamilton Road sub-watershed in Table 5 and 6 shows the worst among all CN-computations. However, the performance by fully composite computation for the Champions Blvd sub-watershed are the best among all CN-computations. Should the result using the Fully Composite that averages CN values from all areas, including impervious areas with high CN values be between the results using three thresholds of CN Cut-off?

Reviewer 3 observations match ours regarding the performance of the Fully Composite computation. The Fully Composite approach has the highest CN threshold (i.e., a Cut-off CN of 100) since as the CN Cut-off value increased, more areas were considered pervious. As such, we do not believe that the results with larger CN should be between the lower CN Cut-off thresholds.  

Reviewer 4 Report

The paper looks good, but it needs further improvements before it is going to be published

• The research questions and objectives should be combined and written in a one paragraph 

• The quality of some figures is not suitable and needs improvement

• Please improve the introduction section and focus more on the previous works and the advantages of the applied Moi ethos

• The paper sections and figures are very long, and I suggest moving some of the sections in the appendixes

• Please separate the conclusion from the recommendations and limitations 

• According to the presented figure 11, I can sometimes see the value of NSE is very low. Please provide some insightful reasons

• Key results should be mentioned in the abstract 

Author Response

The paper looks good, but it needs further improvements before it is going to be published

We appreciate your comments and constructive suggestions. We hope that the new version addresses your concerns and suggestions.

• The research questions and objectives should be combined and written in a one paragraph

We have edited the introduction to improve the manuscript clarity following your suggestion and other reviewers’. However, we maintained the research questions in a numbered list and the objectives in a separate paragraph. We attempted to combine all these in a single paragraph, per your suggestion, but the resulting paragraph was long, and clarity was lost. We hope that the changes in the introduction have improved the clarity of this portion of the manuscript.

 The quality of some figures is not suitable and needs improvement

Per your suggestion, and other reviewers’, all figures were edited and improved.

• Please improve the introduction section and focus more on the previous works and the advantages of the applied Moi ethos* (* we assumed the reviewer intended to write Methods).

Following your suggestion, and from reviewers, we have added more references and discussion providing better contextualization of the current use of CN and have added in the Results and Discussion a subsection in which we evaluate the potential advantages of the current approach.

• The paper sections and figures are very long, and I suggest moving some of the sections in the appendixes

Following your suggestion, we have eliminated Figure 5 in the previous manuscript version. Also, we have summarized the conclusions of the manuscript to make the manuscript shorter. We have not created appendices because we thought that by doing this, we impact the manuscript’s clarity. We hope this can be acceptable to Reviewer 4.

• Please separate the conclusion from the recommendations and limitations

We have rewritten the manuscript conclusion. Future work recommendations were shortened, but preserved, in the conclusions.

• According to the presented figure 11, I can sometimes see the value of NSE is very low. Please provide some insightful reasons

Following your suggestion, we have improved our discussion on the results from Figures 10 and 11, as well as in Table 5. As explained in the manuscript, the lowest CN Cut-off creates an unrealistically large fraction of impervious areas, which in turn systematically creates larger predicted peak flow depths. The results with larger Cut-off values (90 and 93) performed much better in terms of NSE, and when the watershed had the largest average CN, the fully-composite results were the best ones.

 Key results should be mentioned in the abstract

Thanks for the suggestion. The abstract is now improved with more discussion of the research results.

Round 2

Reviewer 1 Report

The authors provided suitable modifications to the manuscript according to all my comments. Theire replied are convincing and satisfactory.

I think that the paper can be published in its present form.