Spatial Mapping and Prediction of Groundwater Quality Using Ensemble Learning Models and SHapley Additive exPlanations with Spatial Uncertainty Analysis
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsTo improve the article, the following terms need to be considered:
1. The research title should be summarized.
2. The numerical results of the research should be mentioned in the abstract.
3. What is the previous research gap? Research innovation should be well noted.
4. What is the reason for dividing 80:20? Don't other divisions such as 70:30 affect the accuracy of modeling?
5. Given that the pixel size of the factors is different, how are these factors considered in the modeling?
6. The quality of all figures should be improved.
7. The discussion section needs to be strengthened. Investigate factors, models, limitations and future proposals.
8. All code must be uploaded to GitHub.
Minor editing of English language required
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThank you for undertaking this work. It shows a great deal of thought and effort and is leading us towards the way these studies need to be undertaken.
I do have a significant number of questions and suggestions, however, that I would appreciate consideration prior to acceptance. These may be summarised:
1. The methodology is well-described, but the interactions and synergies between the different models and strengths and weaknesses are poorly illustrated. This could be improved.
2. Results, interpretations and discussion are mixed together and should be clearly separated. Particularly separate results from any interpretation.
3. In the results, clearly include how the results might be judged, both on their own and against the other metrics. That is, what is the magnitude of response for each metric and how do these values compare across the different metrics. This is includes description of normalisation and how this might be biased by the data or weighting process.
4. The interpretations and discussion are a little light and jumbled. I had to read the second half of the paper several times to appreciate what was being suggested. This will be helped by expanding the points I make above. In particular, discussion on the aggregation of indicators and features (tested using PCC) should be expanded and explored.
5. The use of an 80:20 training:validation set could be explored. You note the variability even with this limited change in random elements. At the least, a suggestion for future work to explore different quantiles of training:validation would be useful.
6. For me, the separation and then linking of the GWQ and the spatial assessment was not clear until I read through several time
Overall, I think this is a worthwhile study and should be published. I would like a little more clarity on the items above and consideration of points made in the manuscript.
Comments for author File: Comments.pdf
English is generally very good. Just a few editorial comments to be considered.
Main criticism is the need to separate some of the sections and clarity over results and interpretations.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
