Deep-Learning-Based Water Quality Monitoring and Early Warning Methods: A Case Study of Ammonia Nitrogen Prediction in Rivers
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThis paper reports a deep learning-based monitoring and detection method for ammonia nitrogen. Overall, this paper is relevant to the current research trend and provides a better solution. I believe this paper can be published in the Journal of Electronics. The authors need to address the following concerns in the manuscript before getting it accepted.
1. Authors need to discuss why deep learning is a better method than conventional machine learning methods or algorithms like PCA and PLS. This papers might help (https://doi.org/10.3390/w13030343, https://doi.org/10.3390/data3040050, DOI: 10.1149/MA2021-02551613mtgabs)
2. Following papers need to be cited and discussed to give readers a better background research:
https://doi.org/10.1016/j.eng.2020.07.027, DOI: 10.1016/j.scitotenv.2023.165710, DOI: 10.1149/MA2022-01522137mtgabs, https://doi.org/10.1016/j.watres.2022.118908
3. Table 1 needs the unit for the parameters.
4. Use larger fronts for the axis of the figures as they were very difficult to read.
5. Few typos were found in line 632 and 635. Please go through the references to eliminate any errors.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsAccording to the reviewer, the article is far from the scope of an Electronics. Better suited to the magazine: Water, Sustainability
This is a comprehensive summary of the study on a deep learning-based method to predict NH3-N concentrations in river water.
The introduction lacked information why the LSTM network was chosen and not, for example, a regular ANN network.
The research methodology was not described as the input data was marked.
The conclusions do not refer to the research of other scientists.How does the proposed LSTM model's performance compare to other deep learning or traditional methods that might be used for the same prediction task? Was there any benchmarking against simpler models or other deep learning architectures?
Were there any efforts made to address potential biases in the dataset?
Given that the LSTM model is tailored to handle time-series data, how was seasonality or other temporal patterns considered? Were there noticeable differences in prediction accuracy at different times of the year?
LSTMs, being a form of deep learning, can often act as a black box. Were any steps taken to improve the interpretability of the model, so that stakeholders and experts could understand the predictions more intuitively?
Can you quantify the time and economic advantages of the LSTM model over traditional methods?
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsThe article titled: Deep Learning-Based Water Quality Monitoring and Early 2 Warning Method: A Case Study of Ammonia Nitrogen Prediction in Rivers. In this paper, we propose the application of a neural network model based on long-term memory (LSTM) to analyze and model ammonia nitrogen monitoring data, which enables high-precision prediction of ammonia nitrogen indicators.
General comments: The manuscript is well written. The results are as indicated in the objectives. A great job of data processing is appreciated. Although the manuscript contributes to the study of inland water body quality methods, it does not relate to other factors, both climatic and anthropogenic. However, due to the importance of local studies and the increasing use of satellite imagery as valuable data, I believe that, after review, its publication should be considered.
Specific Comments:
1. for international readers it is suggested to add in which part of the continent or within China the Qianshan River is located, figure 1 is very specific.
2. Increase the size of the letters and numbers in Figure 1.
3. in line 116, square the area indicated (338 km2).
4. in figure 2 increase the size of the axes, it does not read clearly.
5. in table 1 add the corresponding units of the water parameters analyzed.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsThe manuscript has been improved significantly by the authors, and I recommend publishing the manuscript in its current form in this journal.
