Water Quality Classification and Machine Learning Model for Predicting Water Quality Status—A Study on Loa River Located in an Extremely Arid Environment: Atacama Desert

Round 1

Reviewer 1 Report

The manuscript is poorly written, presentation of data and methodology is very poor. The novelty of the work is missing.

 

There are a number of grammatical errors in the manuscript.

 

Author Response

Thank you for your insightful observations and valuable suggestions. We would like to inform that we have revised and modified the manuscript in order to improve the methodology description, the novelty, etc. 
We hope that you will find our update satisfactory.

Reviewer 2 Report

This manuscript reported a study for water quality prediction based on mathematical models, with a river in the Atacama Desert as a case study. The paper title states the combination of deterministic and machine learning model.

Theoretically, deterministic model means the model that incorporates the mechanism of fate and transport of pollutants within the water bodies, which may include both hydrodynamic and water quality model. However, the contents of deterministic model can’t be found. Therefore, the authors should further polish/change the paper tile and re-organize the paper structure, so that the main text is in accordance with the paper title. From my own perspective, I do not think this paper contains deterministic model.

I also have the following comments for this manuscript:

1.      Line 86-89: As you stated, the deterministic model is based on the physics and chemical properties of the indicators that characterize the watershed, such as hydrological models. However, I can not find such type of models in this manuscript. This seems a contradiction with your following study.

2.      Line 168-169: “generated with xx records”, what is the meaning of xx records?

3.      Line 178-179: Is the data normalization able to remove null values?

4.      Materials and methods part: In this part, you introduced predictive models and random forest, but no deterministic models were introduced.

5.      Figure 2: where is deterministic model?

6.      Figure 5: It is only a demonstration of the results with data set L1, which is not universal. For example, only CAoR-low. CAoR-medium and CAoR-high occur in this figure. I suggest you to provide a universal figure for the seven sampling sites.

7.      Conclusions part: there should not be references and figures citation in the conclusion part.

 

 

Author Response

While thanking for your time, I would like to comment that the responses to your kind review are in the attached document.

 

 

Author Response File: Author Response.pdf

Reviewer 3 Report

Before the manuscript can be further considered for a detailed review, I have several major concerns needed to be handled. (1) WQ is a comprehensive indicator of water quality calculated by several water quality indicator. So, why would the authors try to predict the WQ based on some indicators used to calculate it? I would like to suggest the authors to forecast the WQ, which can be important for water quality management. (2) The model illustrated in Figure 5 is not a RF but a DT. The authors should pay more attention on the RF theory. 

English should be carefully checked and improved.

Author Response

While thanking for your time, I would like to comment that the responses to your kind review are in the attached document.

Author Response File: Author Response.pdf

Reviewer 4 Report

1.       Author must constructively change the abstract in terms of adding some more numerical value of the result.

2.       Please add more recent literature (2023) in terms of RF and deterministic approach for better understanding.

3.       Here author consider Loa River basin for research purposes. Is there any specific reason? The author has done any exercise for choosing that particular basin. 

4.       Please modify the objective section for a clear understanding i.e novelty part should be clearly mention.

5.       There are so many optimization techniques in the recent world; why does the author use RF for optimization purposes? Is there any specific reason for this? My suggestion is at least you have to add two or three hybrid models.

6.       Author citation: Water is a Q1 journal and a renowned one; please do not cite any local journal article (Journal of Machine Learning Research)

7.       Study area figure must be changed; in the study area figure, first the country figure, then the state with the river, then the location with the river basin must be there. Request to draw study area figure using GIS

8.       Author must add statistical components/parameters of collected data in the study area section.

9.       Equation for acc, p, r must be provided; also please add citation for reference purposes

10.   Here author must be mentioned where they got the data and what is the span of the used data. Is there any specific reason for that?

11.   Comparison statement (compare with other research articles) must be added in the result and discussion section to better visualize the proposed research.

12.   Author must add future scope in the last portion of the manuscript.

13.   Advantages and limitations of the proposed model must be added.

14.   Author must add Research Gap

15.   Author must mention why the deterministic model provides better performance.

16.   For better analysis of the result author must add a Comparison plot, histogram plot, and Taylor diagram

17.   Author can add Table 4 in supplementary file

Author Response

while thanking for your time, I would like to comment that the responses to your kind review are in the attached document.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The authors have improved the manuscript acoording to my comments, and I think this paper can be accepted for publication.

Author Response

Thank you.

Reviewer 4 Report

Thank you for revising the manuscript. 

Author Response

Thank you.