Evaluation and Modelling of Reference Evapotranspiration Using Different Machine Learning Techniques for a Brazilian Tropical Savanna
, Rafael S. Palácios 3, Leone F. A. Curado 4, Widinei A. Fernandes 5, Vinicius B. Capistrano 5, Clóvis L. Fritzen 5, Hamilton G. Pavão 5 and Thiago R. Rodrigues 5,*
Round 1
Reviewer 1 Report
General comments
This study is based on three different machine learning methods (Artificial Neural Networks (ANN), Random Forests (RF) and Support Vector Machines (SVM)) to improve the calculation accuracy of reference evapotranspiration. As a whole, the manuscript is a methodological article. However, there is no specific innovation content, just a simple application of the method, and the relevance of the research is not very relevant to the journal.
Special comments
1. Line18-28, key results or conclusions of this study are not identified in the abstract section.
2. The introduction does not point out the advantages and disadvantages of relevant research methods, as well as the main problems and innovation solved by this study.
3. Line80, “five °C” should be “5 °C”.
4. Line87, In Fig. 1, the different classifications of the right subgraph need to be explained.
5. Line133, Why do you choose these three methods of machine learning? What are the advantages and disadvantages of these three methods?
6. Line163, In Fig. 4, the figure is irregular and the dotted lines are too long.
Some sentences have grammatical errors and need to be corrected.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This is very nice paper about machine learning techniques for evaluation evapotranspiration. The description of the method is very clear. However, the authors did not provide the detail of how to use the machine learning. Please give out what kind of tool or software is used so that the readers can reproduce the same results.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
This article investigates a new method for reference evapotranspiration using machine learning techniques. Generally, I found it interesting for the journal’s audience and easy to follow by the readers. So it is worth publication. The main points for revision and the reviewer suggestions are listed below:
In the introduction add some comments to highlight the importance of accurate estimation of reference evapotranspiration in climate research and hydrology. For instance, a recent study concludes that the selection of the reference evapotranspiration method highly affects the climate zone classification based on the aridity index and the drought magnitude and duration (https://0-doi-org.brum.beds.ac.uk/10.3390/hydrology10030064).
Additionally add some brief description of the various machine learning techniques and their use in several scientific domains (eg. https://0-doi-org.brum.beds.ac.uk/10.3390/w10111536, https://0-doi-org.brum.beds.ac.uk/10.1016/j.ecolind.2021.107443, https://0-doi-org.brum.beds.ac.uk/10.1002/qj.3410 ) Why they are important, which are their advantage and disadvantages?
Please improve the resolution of Figure 5.
The article is not prepared according to the journal’s instructions for authors. Please check and revise respectively.
Some grammatical and syntax errors need correction.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The authors has solved my concerns and suggested receiving for publication.
Reviewer 3 Report
Some of my comments are not addressed. None of the proposed references to enhance the added comments are added. Please add the following in the appropriate locatinon in the text as suggetsed in the previous review form.
Also the reference both in the text and in the references list are not as the journal instructions.
Author Response
We greatly appreciate the presented considerations. The changes made were of great value and enhanced the quality of our Manuscript. We apologize for the oversight, as the revised version submitted after the initial revisions did not include the proper references. An issue with the organization of the review files was the cause of the inconvenience. Once again, we thank you for your consideration. In this review round, we could not send a new revised version of the Manuscript through the journal's system. As a result, I emailed the corrected version, with the appropriate references requested, to Editor Dr. Chapman Zhang. We apologize for the inconvenience. I am attaching to these corrections the reformulated version I sent to the Editor.
Author Response File: 
Author Response.docx
