Evaluation of Statistical PMP Considering RCP Climate Change Scenarios in Republic of Korea
Round 1
Reviewer 1 Report
GENERAL REMARKS
This is an interesting paper, using interesting methods.
In general, the English is OK, but I gave a few suggestions for improvement.
Not all the methods are described in the methods section. Some information should be moved from other sections into the methods sections (see later remarks).
The reverse modelling of the shape factor (β) from the skewness, using equation (6) is not as obvious as the authors suggest. The method would work if the skewness is positive, but the authors should mention that and should check that indeed in all cases the skewness was positive.
I noticed an error in equation (8).
The authors have the tendency to show everything, resulting in many pages with maps. I have done suggestions to reduce the number of maps.
I think the results could be presented more clearly by using appropriate terms instead of case 1, case 2, case 3, case 4, case 5 and S1, S2 and S3.
Not all references are easily accessible and I do not read Korean (my apologies for that). I understand that for the data you need to refer to local sources, but where possible, use references from the international peer reviewed literature.
Not all references seemed correct. The referencing must be checked and improved.
Check every reference.
The paper will be publishable if the errors are removed, the method are presented in the methods section, the results are presented a bit more clearly and the referencing is corrected.
ABSTRACT
An explicit definition of the aim is missing in the abstract.
The abstract focusses too much on the methods (118 words) and too little on the outcomes/conclusions (76 words).
INTRODUCTION
Lines 81-85: This information should be in the methods. When reading Table 2 I wanted to check again what MAPE meant and could not find it in the methods. In the introduction the sentence could be a lot shorter (or omitted).
METHODS
Lines 160-169: Study area should be a part of the methods.
Line 106: What is the envelop curve?
Line 124 and 130: Could you indicate what the range is for parameters α, β and γ?
Lines 124-133: As I understand it the authors determine the skewness from the data, then calculate the shape factor (β) with the inverse of equation (6) and subsequently calculate the frequency factor (KT) with (5). This method has a problem since (6) is not a monotonic function. Therefore, the skewness (γ) is a function of β, but β is not a function of γ (see graph below). So, it is not vice versa.
If the skewness (γ) is between -0.02 and 0.00, then the shape factor (β) can have three different values ranging approximately from β = -0.05 to β = 0.28. This could yield frequency factors ranging from about 2 to 3.
The method will work if the skewness is always positive, because beyond β = 0.28 the function is monotonic. Probably this is the case. You should mention the range of skewness that you found.
Lines 138-157: It is not clear to me how the factors that go into AP are calculated (RIP, RHT, TVT, OTF). Also, it is not clear how AP is related to PMP. In short, the calculation of HPMP is not clearly described. What is the AP used for?
Line 166: How many of these sites have more than 40 years of data?
Line 188-189 (Figure 3) I do not remember that I have read how the modified nomogram was fitted to the data from the 615 sites. How did you get the PMP from the data? PMP is probable maximum precipitation (in a certain period?). If you have 40 years of observations it could well be that this probable maximum precipitation was not observed in this period.
Lines 225-240: This should have been reported in the methods.
APPLICATIONS (RESULTS AND DISCUSSION?)
Lines 160-169: Study area should be a part of the methods section.
Line 164-165 Sentence is not clear. What are the 615 sites used for and when are the 62 sites used? Besides, this information should be in the methods section.
Line 182: Equation (8) contains an error: It does not describe the red line (nor the blue line). The coefficient of the quadratic term is off by a factor 100. It should be 6.416x10-7.
Lines 196-204 and table 2: I think using this terminology with “cases” is not very clear. You could also mention them as Hershfield KM=15, Hershfield nomogram and Modified Nomogram. This might make the tables and figures easier to read.
Lines 196-204 and table 2: These results show that the Hershfield method with the variable KM grossly underestimates KM and that the Chow [28] method overestimates KM (maybe you should choose a shorter period than 60.000 years). Based on these results you could decide not to use these methods in the rest of the paper and focus on case 1, 3 and 4. It makes sense that if you fit the nomogram to your own data that it fits better then when using a nomogram form the literature. However, the differences between the original Hershfield nomogram and the modified nomogram are small.
Lines 225-240: This should have been reported in the methods.
CONCLUSION
Line 321: and ratio of 0.988 à and a SPMP/HPMP ratio of 0.988.
Lines 324-325: HPMP is expressed in mm/24h and the period in years. You cannot compare variables with different units (5 m would be less than 10 s, but 500 cm would be more). So the HPMP cannot be more or less than 60,000 years and also not equal to 60,000 years, since it is expressed in different units.
Line 330: with ratios of 0.878 and 0.726 à with SPMP/HPMP ratios of 0.878 and 0.726
MINOR CORRECTIONS
Line 10: two methods à two types of methods
Line 18: average ratio à average SPMP/HPMP ratio
Lines 58-59: The word premium is not used correctly here (I think). Maybe: “… 43 years. When the PMP was calculated based on a 30-year period, it was recommended to augment it by 10%.”
Lines 69: A stationary PMP should not be ignored à A stationary frequency analysis should not be used?
Line 89: T The à The
Line 110-111: mean of annual series à mean annual rainfall.
Lines 155-157: I could not find reference [9] on the internet and if I could have found it, it probably is in Korean, so not very useful as a reference in an international journal. In any case you could have provided a bit more detail on how the future PMPs were estimated. Was this done by running a climate model (which one?) or was it done by a statistical extrapolation? The abstract of reference [10] already gave me more insight in the way this was done. Get some of that info in your methods.
FIGURES AND TABLES
Most figure and table captions are too brief.
Line 188-189 (Figure 3) The figure caption is not clear enough. The figure caption should describe the figure. Red line is …. Blue line is …. The dots represent ….
Line 222 (Figure 4): I suggest to reformat the figures in such a way that the 6 sub figures (4a – 4f) fit on one page.
Line 222 (Figure 4): Considering the results in table 2, my suggestion would be not to show the maps for Hershfields varying KM (case 2) and for the KM from Chows model (case 5).
Line 222 (Figure 4): Also, these maps show that your case 2 and 5 are way off. But that could already be concluded from table 2.
Line 254: My first response would be to show RCP4.5 to the left and RCP8.5 to the right. So they can be compared more easily. However, if you decide to not show any further results of Hershfields variable KM method, then you do not need so many graphs. You could show two graphs: RCP4.5 to the left and RCP 8.5 to the right. Each figure showing the Hershfield nomographs as dashed lines and the modified nomograph as solid lines, using different colours for different periods (e.g. 2040 blue, 2070 green, 2100 red).
Line 266 (table 4) Also mention in the table heading that these are the results obtained by using the modified nomogram.
Line 266 (table 4): using S1, S2 and S3 for the periods does not make the discussion clearer. It would be better to label these periods with their starting year: 2040, 2070 and 2100. I think that also in the text the terms S1 – S3 can be replaced by the start years.
Line 268: Also here my advice is to put RCP4.5 to the left and RCP 8.5 to the right and replace the s1-s3 notation by the starting years.
Line 305 (Table 5): I think you can leave the results Hershfields varying KM method and Chows method in this table (no need to remove them here).
Line 307 (Figure 7) I suggest to omit the maps for case 2 and 5. Table 5 has already sown that these methos also not work properly for the future and the article already has many maps. Als label the maps by the methods instead of case 1, 3 and 4.
Line 307 (Figure 7) I suggest to omit the maps for case 2 and 5. Table 5 has already sown that these methods also not work properly for the future and the article already has many maps. Als label the maps by the methods instead of case 1, 3 and 4.
REFERENCES
Most references are in the introduction and the methods section. The “Applications” section (results and discussion?) contains few references. The authors could pay more attention to comparing their results with previous work.
In a number of cases I could see that the reference number did not refer to the correct publication. For each reference in the text, the authors should double check if it refers to the correct publication.
Line 49: Hershfield (1961) is not reference nr [14], but [15].
Lines 77-79: Insert references for these methods.
Line 90: Hersfield (1961 is not reference nr [14], but [15].
Line 121: Reference [30] refers to Lee et al. (2016). The reference for the institute of hydrology should probably be [29].
Line 142: I doubt whether reference [31] is the correct reference. And even if it is, it is in Korean, so that does not help much.
Line 154: Lee and Kim (2016) is not reference 32, but probably reference [31].
Lines 155-157: Are you sure this is reference [9], since reference [10] seems to be about estimating future PMP. Moreover, I could not find reference [9] and if I could have found it, it possibly is in Korean, so not very useful as a reference in an international journal. In any case you could have provided a bit more detail on how the future PMPs were estimated. Was this done by running a climate model (which one?) or was it done by a statistical extrapolation? The abstract of reference [10] already gave me more insight in the way this was done. Get some of that info in your methods.
Lines 165-166: Lee and Kim is not [10, 32], but [10, 31]
Line 175: Publication [29] is not a publication by Hershfield.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This study lacks innovation and its content is consistent with the journal. The key information and the research process of the paper are not clearly expounded, which cannot reflect the repeatability of the research, the scientific nature and the reliability of the research results. This is a major problem in this research. At present, the study needs at least a major overhaul before it is publishable. In addition, there are still some minor issues to be corrected. The specific issues are as follows:
ISSUES:
1. The spelling of parameter "Km" in the abstract is incorrect. Please check for modification.
2. Please check the first statement of 2.1. Hershfield Method. "T" seems to be superfluous, please confirm.
3. ‘Hershfield initially suggested a fixed value of 15 (Km= 15)’.Please provide references.(Line 96-97)
4. Reference [28] mark position is inappropriate, suggest modification.
5. The description of parameters in equation (3) is not comprehensive. What does the "T" in line 117 mean?
6. It is suggested that the author carefully check the manuscript and do not make unnecessary mistakes.
7. ‘coefficients for the normal and Gumbel distributions are 0 and 1.1396, respectively’.Please provide references.(Line 137)
8. ‘In this study, the SPMPs were estimated by applying the five cases listed in Table 2, and the precipitation data up to 2020 for each site were used. ‘The key information is not clear. Scientific research should be repeatable. Such a way of explanation cannot reflect the repeatability of this research, nor can it reflect the scientific and reliability of the research results.(Line 191-192)
9. The structure of the article is not clear and the logic is not good.
10. ‘T=60000 years’.Please check for accuracy.(Line 213)
11. The conclusions of this study are of little scientific significance and of little application value.
12. The third part of the paper is tedious and unclear.
13. The conclusion is not concise enough.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
GENERAL REMARKS
The manuscript has improved considerably since the last version. However, a few things still need to be improved.
· In my view the focus of the abstract should be more on the outcomes of the research.
· Equation 6 must be corrected.
· The rounding of the numbers in the tables should be improved.
· There are some minor errors that should be corrected (see below).
Other than that, I made a few suggestions, that you could use to improve the manuscript (in my view), but you can disagree or improve in another way.
If the error mentioned above are corrected, then the manuscript can be published according to me.
ABSTRACT
Structure of this abstract:
Introduction 52 Words – OK
Objective 31 words – OK
Methods 124 words – Too long
Results 77 words – Too Short
When I read an abstract, I am usually interested in the motivation (introduction), the objective and the outcomes (results/conclusions). I can read the details about the methods in the paper later.
The focus of the abstract should be on the results and conclusions. The main conclusions should be mentioned in the abstract.
You abstract is focused on the methods (124 words). The most important part of the abstract, the outcomes, is only briefly discussed (77 words).
Based on the outcomes I will decide whether it is worthwhile to read the rest of the publication. If these are not properly presented, I might not read your paper at all.
I suggest to summarize the methods part of the abstract and write more about the most important results and conclusions.
INTRODUCTION
Maybe the introduction is a bit long (in my view), but the structure is OK.
METHODS
Good that study area and some other parts that were outside the methods are now included in the methods.
Lines 115-123: The datasets are now described more clearly.
Lines 124-133: The modelling is now described more clearly.
Line 149: It looks like the section number (3) is missing).
Line 150: It looks like the section number (3.1) is missing).
Line 152: [154] à [15]
Line 169: It looks like the section number (3.2) is missing).
Line 177 (Fig. 2) It is a bit confusing that the red line in figure 2 coincides with the blue line in this figure 3 and not with the red line in figure 3. Maybe the easiest change is to change the colours of the lines in Figure 2 (this figure).
Lines 180-198: If I understand correctly your methods is as follows:
1. You determine the skewness (γ) with the standard moments method
2. Then you use (6) to find out which shape factor (β) would give this skewness.
3. Subsequently you use (4) to calculate the frequency factor (KT).
4. This goes into (3) to get XT.
Do I understand correctly if I say that this XT is your PMP for a given return period.
If this is correct maybe you could describe this procedure clearly.
If this is not correct then you should certainly describe it more clearly.
Lines 204-206: I understand that you are not using these two-parameter models. Maybe you can omit the remark on the two-parameter models.
An Error in equation (6)
Line 198: the power of the numerator should be 3/2, not 2/3.
Below I inserted a graph of the relation between Skewness (γ) and the shape factor (β), according to equation (6) in the manuscript with a power of 2/3 (red line in the figure) and a corrected version of equation 6 with a power of 3/2 (blue line).
Although their notation is a bit different Muraleedharan (2009) also use a power of 3/2.
Muraleedharan G, Guedes Soares C and Lucas C (2009) Characteristic and Moment Generating Functions of Generalised Extreme Value Distribution, Chapter 13 in: Wright LL (ed.) Sea Level Rise, Coastal Engineering, Shorelines and Tides. Nova Science Publishers, Inc. 9 pp.
Or look for a better reference yourself.
APPLICATION AND RESULTS
Line 276 (eq. 8): the equation now is correct. Thank you for looking at my comment critically and correcting my error. It was indeed off by a factor 100, but the number that I gave was wrong. My apologies.
Line 278: Daily precipitation should be in mm/24 h, not in 24 h/mm.
Line 283 (Fig. 3): See my comment at Figure 2.
Line 289: Table 4 à Table 3.
Line 294: Maybe change “The modified Hershfield’s nomograph (Case 4) shows” into “The original Hershfield’s nomograph (Case 3) and the modified nomogram (case 4) show”, since the differences between these two nomograms were very small.
Lines 288-303: My summary: In short, the best two methods were the modified nomogram and the original nomogram, whereas Hershfield’s method with a fixed KM=15 also gave acceptable results. Case 2 (Hershfield’s method with a varying KM) and case 5 (Chow’s method) provided PMP which were either far too low (Case 3) or far too high (Case 5).
Line 305 (Table 4): See my comments on the rounding of the number at Table 6.
Lines 310-311: “the result of SPMPs (except Case 2 and 5)” à “only the result of SPMPs of case 1, 3 and 4”.
Lines 318-327: You are here discussing figure 4, which does not contain a map for Case 2 and Case 5. So, there is no need to discuss these cases here. If you still want to discuss case 5, you should do that where you discuss Table 4 (previous page).
Line 328 (fig. 4): The four maps on one page now provide a nice overview and can be compared to each other.
Line 370 (Fig. 5): Since you have already decided that Hershfield’s method with the varying KM severely underestimates KM, it might be better to omit the dots from this graph completely. The reader can than concentrate on the methods that do provide reasonable results.
Line 384-385 (Table 5): Table 54 à Table 5.
Line 384-385 (Table 5): See my comments on the rounding of the number at Table 6.
Line 384-385 (Table 5): Why stick to this unclear S1 – S2 – S3 notation. Would it not be better to just use the starting years. The table caption can explain that the years are the starting years.
Line 392 (Fig. 6): I suggest that in the legend - 200 is replaced by 0-200 or by < 200. 2500 - can be replaced by > 2500 (-200? Negative precipitations do not occur)
Line 433 (Table 6): The table caption states “HPMP (2100)”, but the column heading says “HPMP (2020)”. I am confused. Which one should it be?
Line 433 (Table 6): Rounding of the numbers: I suggest to provide all PMPs (max, mean and min) without decimals, so round to whole numbers. Lees is more. You will provide a better overview by omitting meaningless details. Also MAE, MAPE and RMSE can be given as rounded numbers (no decimals). For the KM values one decimal makes sense, since some of the KM values are less then 10. Note that all KM values should be rounded in the same way, so all with one decimal. SPMP/HPMP rations can be provided in three decimals, like they are now.
Lines 435-443: I understand that this part is for the RCP 4.5 scenario. You should mention that when you start discussing the results.
Line 444: start a new paragraph when you start discussing the results for the RCP 8.5 scenario.
Lines 448-449: “and the spatial patterns were indicated the difference between SPMPs and HPMP”. This is not a correct sentence. Correct it.
Lines 458-459: “decreased from 0.878 for the” à “was smaller than the 0.878 for the”
CONCLUSION
The conclusion is a bit long. Maybe provide less details on the numbers.
Line 478: “(Case 4) shows” à “(Case 4) was considered the best method, since it has”
Line 483: Replace the Korean symbol (ê¹€).
REFERENCES
During a quick check I did not encounter errors in the cross referencing.
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
The revision is more careful, I think there is no problem.
Author Response
We would like to express our gratitude again for your valuable time and consideration in offering feedback on our manuscript.
