Modeling Reference Crop Evapotranspiration Using Support Vector Machine (SVM) and Extreme Learning Machine (ELM) in Region IV-A, Philippines

Round 1

Reviewer 1 Report

2.Major concerns:

1). There is lack of introduction of research background of development and evaluation of different models to estimate the ETo worldwidely and in Philippines, which makes it unclear about the importance and necessity of this study. In fact, this is the prerequisite before any idea or research topic can be determined. Do include this in Sect 1 (Introduction).

2)Normally, the input observational data of the training set and evaluation set are chosen to represent different climate zones in a country. In this study, 8 stations of 3 climate types are included, giving opposite evaluation conclusions for different models even for stations belonging to the same climate zone (Line 24-26 in P18). This is confusing, thus the authors are suggested to re-consider the inclusion of the stations for training and evaluation set.

3) The quantative evaluation of monthly daily ETO estimated by different models are recommanded in Sect 3.2. Right now, the dot-line figures for all stations are insufficient to illustrate the performance of these models in a quantative way.

3.Specific comments

• There is too much introduction of SVM and ELM (referring Line 84-134) in Sect 1.Normally these information should be in Sect 2 Methods.
• The structure of this paper should be explained at the end of Sect1.
• Line 52: Eto ->ETo
• Line 55: please reconstruct the sentence: using a lysimeter using a simple…
• Line 57: delete ‘to use’
• Line 67: delete ‘ and’ before Priestley
• Line 71-73: Logical conjunctions are suggested, like ‘On one hand, on the other hand’
• Line 132: is there a letter missing after ‘where’?
• Line 154: error in puntuation marks, like extra brackets and comma.
• Line 159-164 and Figure 1: The name of type of climate zones don’t match in paragraph and Figure 1. There are 4 types of zones in Figure 1 while only 3 of them are mentioned in the paragraph. Apart from that, it is named as ‘Type I’ in line159 while labeled as ‘1^st Type’ in legend of Figure 1. Please write in consistent way.
• Line 164: Table 1->Table 2
• Line 191: error writing after ‘As shown in’.
• Line 226-230: format error, please rewrite this.
• Line 250: extra full stop mark at the end of this line, delete it.
• Table 5~Table 11: As suggested in Major concerns, listing all the calculated results for every station in 7 tables is not necessary. Please re-consider the representiveness of stations in different climate zones and reduce the number of tables. Normally, only the most representative results are included in the paper while other similar results can be listed in supporting file or supplementary file.
• Figure 2~8: Similar suggestion. Please reduce the number of figures showing similar information and then provide other figures illustrating the performance of these models in a quantative way (see the 3^rd Major concern).

Author Response

Greetings, 

Kindly see the attached document containing the consolidated point-by-point responses to the reviewer's comments. 

Thank you and keep safe. 

Author Response File: Author Response.pdf

Reviewer 2 Report

I have added some questions and suggestions to the text. Please find the manuscript.

Comments for author File: Comments.pdf

Author Response

Greetings, 

Kindly see the attached document containing the consolidated point-by-point responses to the reviewer's comments. 

Thank you and keep safe. 

Author Response File: Author Response.pdf

Reviewer 3 Report

I enjoyed the manuscript. I only have several concerns:

-Move all equations  and Table 1 in introduction to methods section.

-Only T average based Oudin et al. (2005) is missing from the benchmark Table 1 (https://doi.org/10.1016/j.jhydrol.2004.08.026). 
It can be easily incorporated to the this study and compare the results with AI methods (https://webgr.inrae.fr/en/models/evapotranspiration-model/)

-Page 4/29 Line 146: Only research gap was given here. Add research objective(s) and research approach in this paragraph.

-P6L191:  Error! Reference source not found.

-Table 5 to 12: FAO-Penman results should be added too. Fig 2-8 have Penman results which is good to compare.

-Page 27 Line 129: "daily reference crop evapotranspiration" all figures are monthly, why? Please show daily figures for a selected year.

-Table 4: why too many similar metrics are used? For example R2 and NSE are really similar. MAE and MAPE are also redundant.
Please add SPAEF metric to compare the patterns in the time series. 

Author Response

Greetings, 

Kindly see the attached document containing the consolidated point-by-point responses to the reviewer's comments. 

Thank you and keep safe. 

Author Response File: Author Response.pdf

Reviewer 4 Report

This is an interesting paper in ET and the methodology you have used to accomplish this research work. But please consider the following notes before the paper gets published

General notes

1. I noticed a little explanation about the importance of ET₀ calculation in section 1 (introduction). Please add more studies and literature review that support the significant importance of ET₀ estimation in hydrology and agricultural management. Also, the explanation of machine learning algorithms should be moved to the methodology section. In the introduction section, please focus on the importance of the study and the major contribution of your research by addressing/forming some questions that readers can easily understand the objective of your research work.
2. What did you use only SVM and ELM? Please provide the reasons behind the selection of these two approaches. There are many ML approaches could be used like relevance vector machine (RVM), Decision Tree, Random Forest and others.
3. I do recommend to add a flowchart that highlights the work methodology.
4. I noticed in section 3.2.1 that you describe the combination of the inputs considered for different models. This should not be part of the results. It has to go the methodology section.
5. I see little explanation and interpretation in the results of section 3.2.2 (Comparison of models between SVM and ELM). It should be explained in a better way.
6. What is the physical interpretation that T_max, T_min and R_s are the best parameters to estimate ET₀?. The authors conducted the comparison from statistical perspective, but there is misinterpretation for the physical meaning of the results. I do recommend to add that.

Detailed notes

1. In line 28, please use the subscript for max , min and s in Tmax, Tmon, and Rs
2. In line 52, use the capital letter for ET₀
3. I guess in equation 5, there is missing parentheses.
4. In line 191, it seems there is an error “Error! Reference source not found”

Author Response

Greetings, 

Kindly see the attached document containing the consolidated point-by-point responses to the reviewer's comments. 

Thank you and keep safe. 

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Still, as have recommanded, more effective way to illustrate the comparison of performances of different models are encouraged.

Reviewer 3 Report

Thank you for the revised ms.