A Calibrated, Watershed-Specific SCS-CN Method: Application to Wangjiaqiao Watershed in the Three Gorges Area, China

Round 1

Reviewer 1 Report

The paper entitled " A calibrated regional specific SCS-CN method: Application to WangJiaQiao watershed in the Three Gorges Area, China” contributes to the improvement of SCS-CN methodology with an original perspective. The paper is well concise and provides useful information for the readers of the journal. The paper can be published with minor revisions:

The introduction section needs revision. Review of the state of the art regarding more recent works done in 2017-2019, their strengths, weaknesses (past contributions) and research gaps (current needs of improvement) should be expressed in Introduction section. Explain/present the logical scheme of the algorithm proposed in Chapter 2 (The Proposed Calibrated Regional Specific SCS-CN Method). The results section is well written looking convincing. The references cited are mostly decade old. I advise to cite the latest papers in every context to show the relevance of the problem in the present time.

Author Response

Please see the attachment. Thank you for reviewing our manuscript.

Author Response File: Author Response.pdf

Reviewer 2 Report

In the paper ‘A calibrated regional specific SCS-CN method: Application to WangJiaQiao watershed in the Three Gorges Area, China’ various Curve Numbers were considered and various methods of estimation of runoff depth were discussed.

The problem is very important from the point of view of designing hydro-technical structures, preparing maps of flood risk etc. The Authors properly justified the need of the study because the choice of lambda and S is very complex and should be carefully performed. The use of various advanced methods is also promising.  In my opinion, however, the paper is of poor quality and cannot be published in the present form. I encourage the Authors to improve and resubmit the paper.

The rationales are as follows:

A model should be named ‘regional’ when the analysis is conducted for the number of watersheds located near or not far from each other whereas the model was designed for one watershed only. Thus the title is not adequate.
Similarly, the ‘regional specific S’ in section 332 is in fact the local S. The objective of the paper was not clearly explained and the reader may feel confused why various methods were involved and which one is the recommended one. The novelty of the method should be questioned at this stage because the Authors used the bootstrap technique which is a commonly used method in statistical inference. The major disadvantage is that the equation (12) was not verified on a new sample. There is a lack of a clear guidance how to obtain Q when we have new P data for this watershed. To use lambda=0.043 and S=260.081?   Why was the calibrated model compared to conventional model through regression on P (eq. 14)? Why the model with the asymptotic CN was not adjusted? It is not clear which model was finally recommended; the calibrated or adjusted. The paper looks as a ‘discussion of various methods of determining Q using the CN method’. The confusion of various method is not clearly explained. There are many statistical tests used in the paper which are not clearly explained. Examples:

- test for R^2 line 221;

- test for standard error; how is the p-value calculated (Figure 3). The whole paper should be revised accordingly.

There are errors or disagreements, eg.

(line 136) the normality of the two-dimensional variable (lambda, S) was not tested. The Shapiro-Wilk test is used only for one-dimensional variable.

(Table 2) the lower CI limit for the mean is higher than the upper one,

(line 280) ‘the conventional SCS-CN model becomes (…) not statistically significant’; how do the Authors understand this exactly? There is no test for this, and the word ‘significant’ should be used with caution.

(line 298-299) ‘the newly calibrated model had the lowest residual variance and standard deviation’ but the standard deviation model error and variance of residual are larger for calibrated (eq 12) than for corrected (eq 15) in Table 2.

The Authors mix the confidence level with the significance level eg. in line 323 they show the interval range of 99% confidence and a p-value, but what is p? No test is used here. The whole paper should be revised carefully because there are more similar mistakes in statistical language.

Many parts of the paper are unnecessary, eg equation (8) is known in the literature and the derivation of the formula is useless. The distribution of the S parameter, estimated from empirical values, is known to be skewed thus the considerations in lines 90-98 and others, and statistical calculations related to normality should be removed. Instead of this, the empirical S values should be presented on empirical pdf plot.

 

Author Response

Please see the attachment. Thank you for reviewing our manuscript.

Author Response File: Author Response.pdf

Reviewer 3 Report

The topic is interesting but I think the research motivation is not clearly provided in the manuscript. Especially, there are many other conditions that affect the prediction using the SCS-CN method. The authors need to fully clarify this. In this regard, I would recommend a major revision. Please find my comments below:

Lines 32-33 How do the authors can be sure about that the proposed method can be applied to different areas with different conditions? At least, the authors should provide assumptions such as, for example, the same optimizing conditions.

Lines 38-39 I agree with the authors but there exist many other arguments about this. I would suggest the authors to include more references about this but keep in mind that model complexity basically depending on the purpose of the modeling and no model is perfect in inself. There are many ways such as data assimilation to make prediction more accurate and precise.

Lines 122 It is interesting to see Equation 8 but I am not quite sure how much important this is compared to other conditions such as antecedent moisture condition, which was not mentioned at all in this study. In addition the curve number itself is empirical and can produce different results depending on location. In addition if the authors use a conceptual hydrologic model in this study, the model requires a calibration and validation procedures inherently and CN itself needs calibration.

Author Response

Please see the attachment. Thank you for reviewing our manuscript.

Author Response File: Author Response.pdf

Reviewer 4 Report

The authors present comprehensive  approach to improve the performance of the Soil Conservation Service Curve Number method in the regional scale on the basis of a measured depth of rainfall-runoff data pairs. The key of this method is optimizing maximum potential water amount S and initial abstraction ratio λ with use of bootstrapping for inferential statistics, BCa procedures and normality tests.

The optimized λ does not match the traditional value of 0.2 thus the Authors obtain the relationship between S_0.2 and S_λ in order to convert from 0.2-based CN to λ-based CN.

The authors have calculated the runoff of the WangJiaQiao watershed with use of proposed model and three other models proposed earlier by other authors. For the purpose of measuring the runoff  predicting an error of every model they have applied residual analysis and proved that the newly proposed model is the most accurate.

The paper presents a topic that is of interest for this journal. The paper is original and the new method is described in much detail and the results are well presented and discussed. I enjoyed reading this manuscript, because the method of Curve Number to runoff prediction is extensively used especially in not controlled catchments but some key issues of it are still quite ambiguous.

My only suggestion is that the formula (8) does not have to be derived here (this is quite trivial) because it was previously presented in other publications, e.g.:

Hawkins R. H., (1993), Asymptotic Determination of Runoff Curve Numbers From Data, Journal of Irrigation and Drainage Engineering 119(2), p.334-345, DOI: 10.1061/(ASCE)0733-9437(1993)119:2(334),

Fu S., Zhang G., Wang N., Luo L. (2011), Initial Abstraction Ratio in the SCS‐CN Method in the Loess Plateau of China, American Society of Agricultural and Biological Engineers, Vol. 54(1): 163-169

Author Response

Please see the attachment. Thank you for reviewing our manuscript.

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

I think the raised questions were answered properly and recommend publication in the corresponding journal. Further study would be interesting incoporating the antecedent moisture conditions.

Author Response

Please see the attachment. Thank you for reviewing our manuscript again. 

Author Response File: Author Response.pdf