The Integration of Multivariate Statistical Approaches, Hyperspectral Reflectance, and Data-Driven Modeling for Assessing the Quality and Suitability of Groundwater for Irrigation

Round 1

Reviewer 1 Report

In general has novelty in this research and very quality research. But will be better if Author gives to my those comments:

• Introduction is not strong organized, first paragraph (52-56) are general words, will be more important and soundness if add which type of minerals is dangerous for vegetations in ground water. It will be exploitation importance of finding optimal methods for analysing of elements.
• In abstract SVMR correlation range is from 0,01 to 0,95 why this range non stable. But chosen as optimal variant.
• Line 152-156, what type of soil in both regions and depth of underground water it is important too for mineralisation.
• Conclusion is not strongly clear, will be better if authors will give explaination integration of which approach recommended and/or gave accurate result, how was correlation degree.

Author Response

Reviewer #1

In general has novelty in this research and very quality research. But will be better if Author gives to my those comments:

Response: We greatly appreciate your critical observations as well as your constructive and helpful comments. We hope that we could address your questions/comments by the explanations and revisions made in the manuscript. We believe that the manuscript is substantially improved after making the suggested revisions.

• Introduction is not strong organized, first paragraph (52-56) are general words, will be more important and soundness if add which type of minerals is dangerous for vegetation in ground water. It will be exploitation importance of finding optimal methods for analysing of elements.

Response: Thank you very much for this comment. More information about the negative impacts of excessive concentrations of particular ions in groundwater on growth and productivity of field crops has been mentioned in introduction section as well as in discussion section.

• In abstract SVMR correlation range is from 0.01 to 0.95 why this range non stable. But chosen as optimal variant.

Response: The SVMR model showed a weak estimation for the RSBC and MH in in both the training (0.29 and 0.25) and testing (0.01 and 0.1) datasets, respectively. This may be due to the all nearly samples from both regions met the requirement of water quality for irrigation purpose based on the both indices of irrigation quality as shown in Table 4. Additionally, during the development of the ANFIS models, we applied an optimization technique to optimize the ANFIS parameters. This step was applied to all models that developed for the investigated variables. The main advantage of this technique is that it gives the model abilities to minimize the RMSE and consequently increases the value of R². However, the case was not the same with the SVM therefore R² values are relatively lower compared with those of the ANFIS model for RSBC and MH parameters.    

• Line 152-156, what type of soil in both regions and depth of underground water it is important too for mineralization.

Response: Many thanks for this comment. All information related to soil type and depth of groundwater surface for both regions has been added in M&M section.  

• Conclusion is not strongly clear, will be better if authors will give explanation integration of which approach recommended and/or gave accurate result, how was correlation degree.

Response: The conclusion section has been improved.   

Author Response File: Author Response.pdf

Reviewer 2 Report

Review report of “The integration of multivariate statistical approaches, hyperspectral reflectance, and data-driven modeling for assessing the quality and suitability of groundwater for irrigation”

This study analyses water quality using several statistical methods. Using several irrigation water quality indexes and data-driven modeling it was possible to simulate water quality results based on the hyperspectral reflectance of the samples from two locations in Egypt. 

Results are interesting and important for the express field measurements. I find methods appropriate and results robust. However, there are some comments to address.

Many methods used in the study, while some are well explained, others doesn't provide much needed connection. For example what FA and PCA accomplishes here? Integration of these methods doesn't provide additional information to the main focus of the study - data driven modeling of the water quality indexes. It does explain some water characteristics in different sampling sites though. Author's intent in using these methods for the study as a whole is not well explained. Factors, and correlation is not seen to have impact to other places of the study. It can be easily removed without impacts to the main points of the study.

Another comment on IWQI, the system establishes set of classification, however it doesn't solve the problem of "better interpretation". There are still much variations in interpretation of the water quality, even in this study. Where WD is "suitable for irrigation with some restrictions" based on authors recommendation, while some results are negative in many indexes. Authors, should include a scheme for interpretation or wight parameters by main importance to minor.

Which band correlation were finally selected for each indexes, couldn't be found. Did authors use all spectrum or have removed low correlation wave length from the analysis. Questions of interchange and applicability of trained model, should also be mentioned. For example, can model be trained on one basin simulate reflectance values of other?

 

Minor comments

Abstract: 34-36 Is not based on any of analysis but more on the hypothesis and should be discussed but not stated as a conclusion of the study.

35 “may are the main” should be “may be the main ”

 

Include study area map and sampling points locations

186    Provide weights of each parameter and show why it is so, and how they were decided  

214 – should be - very hard

Distinguish figures by WD and CND areas in supplementary. Currently it is only separated by number of samples

Please, provide results of the SVMR model

Author Response

Reviour#2

Review report of “The integration of multivariate statistical approaches, hyperspectral reflectance, and data-driven modeling for assessing the quality and suitability of groundwater for irrigation”

This study analyses water quality using several statistical methods. Using several irrigation water quality indices and data-driven modeling it was possible to simulate water quality results based on the hyperspectral reflectance of the samples from two locations in Egypt. 

Results are interesting and important for the express field measurements. I find methods appropriate and results robust. However, there are some comments to address.

Response: We greatly appreciate your critical observations as well as your constructive and helpful comments. We hope that we could address your questions/comments by the explanations and revisions made in the manuscript. We believe that the manuscript is substantially improved after making the suggested revisions.

• Many methods used in the study, while some are well explained, others doesn't provide much needed connection. For example what FA and PCA accomplishes here? Integration of these methods doesn't provide additional information to the main focus of the study - data driven modeling of the water quality indexes. It does explain some water characteristics in different sampling sites though. Author's intent in using these methods for the study as a whole is not well explained. Factors, and correlation is not seen to have impact to other places of the study. It can be easily removed without impacts to the main points of the study.

Response: Many thanks for this comment. The one objective of this study is to incorporate the different physiochemical parameters (ions and cations analysis) into a multivariate analysis to better understand and interpret the major contributing parameters that influence the groundwater quality of two distinct regions (WD and CND). Therefore, in this study, FA is used in conjunction with PCA to investigate the influence of different physicochemical parameters on groundwater quality of both regions and to determine the quantitative relationship between them.

The integration of multivariate statistical analysis approaches such as FA and PCA and a large set of physiochemical parameters has been widely applied as an effective strategy for providing meaningful information regarding water quality and sustainably managing water resources. For instance, FA is a powerful tool that is used to reduce the complexity of several physiochemical parameters without losing a great deal of information from the original measurements. This appreciable reduction in physiochemical parameters is highly necessary for the superior interpretation of the original measurements, as well as for reduced costs of laboratory measurements of these physiochemical parameters. Additionally, the PCA was applied for the 11 physiochemical parameters to confirm the results of the FA. The primary objective of the PCA is to classify the different physiochemical parameters into major components such that the parameters that are their vectors, adjacent or parallel to each other and with a small angle between them, signify the strength of their reciprocal association. By contrast, the divergence of the vectors of the parameters expresses weak relationships between them

In summary, FA can determine the key parameters that explain original data to the greatest extent, whereas PCA can provide a comprehensive picture of the interrelationships between all parameters and confirm their relative degrees of influence.

  

 

 

 

• Another comment on IWQI, the system establishes set of classification, however it doesn't solve the problem of "better interpretation". There are still much variations in interpretation of the water quality, even in this study. Where WD is "suitable for irrigation with some restrictions" based on authors recommendation, while some results are negative in many indexes. Authors, should include a scheme for interpretation or weight parameters by main importance to minor.

Response: Many thanks for this reminder. A scheme for interpretation of WQI has been provided based on Meireles et al. (2010).

• Which band correlation were finally selected for each indexes, couldn't be found. Did authors use all spectrum or have removed low correlation wave length from the analysis. Questions of interchange and applicability of trained model, should also be mentioned. For example, can model be trained on one basin simulate reflectance values of other?

 Response: For each IWQIs, the correlation coefficient (r) between the input parameters (original canopy spectral reflectance) and the desired output (different irrigation water quality indices (IWQIs)) was calculated to select the wavebands that have high value of r and exclude those with low r value based on the Figure 6. Finally, the wavebands selected in ANFIS and SVMR models are those with r values higher than 0.70, 0.50, 0.35, 0.49, 0.52, and 0.57 for WQI, RSC, RSBC, PS, TH, and MR, respectively, as mentioned in lines 638-639. 

Minor comments

• Abstract: 34-36 Is not based on any of analysis but more on the hypothesis and should be discussed but not stated as a conclusion of the study.

Response: The sentence has been changed into “From the multivariate analysis, it was concluded that the combination of factor analysis and principal component analysis was found to be advantageous to examining and interpreting the behavior of groundwater quality in both regions, as well as to predicting the variables that may impact groundwater quality by illuminating the relationship between physiochemical parameters and the factors or components of both analyses”.

• Include study area map and sampling points locations

Response: The study area map and sampling point locations has been added as supplementary figure (Supplementary Figure 1).

• 186    Provide weights of each parameter and show why it is so, and how they were decided  

Response: all these information have been provided in M&M section.

• 214 – should be - very hard

Response: Done

• Distinguish figures by WD and CND areas in supplementary. Currently it is only separated by number of samples

Response: In this study, the process of the model development started with combining the measured data from WD and CND as well. The data set was then split randomly into two parts for training and testing of the model for each desired output, and the area of data was not criteria in this case, since the study aimed to develop a model that can be applied in the two areas.  Moreover, it was not practically easy to distinguish WD and CND in each model, i.e for IWQI, RSBC, TH, PS, RSC and MR models.

• Please, provide results of the SVMR model

Response: The results of the SVMR model have been provided in the Supplementary Figures 5-7.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The manuscript was improved but I still have some unanswered comments.

 

Map of study area should be incorporated into the paper, instead of

supplementary material, as it is important information.

 

502-504 “The strong loadings of the salt ions in the groundwater of the WD may be attributed to the incursion of seawater from the underlying fractured limestone of the Eocene aquifer”

I am not sure about the seawater in WD region. It is located really far from both seas. According to the most publication, I found about El Fayoum groundwater’s main sources are indicated as a runoff of the Nile River and the agricultural runoff. Could you back it up with other reference?

For example: https://www.sciencedirect.com/science/article/abs/pii/S1464343X19302079

https://www.sciencedirect.com/science/article/pii/S2090997718300683

 

507 Sea water? From where?

545-548 Repetition, and see comment above

 

561-562 Not all groundwater. In previous paragraphs, WD samples were characterized as suitable with limitations. Then few paragraphs later, they are classified as “injurious to unsatisfactory”.

 

Again, you haven’t addressed my comment about IWQI interpretation. Your response is about WQI -

“Response: Many thanks for this reminder. A scheme for interpretation of WQI has been provided based on Meireles et al. (2010).”

However, you didn’t include interpretation of the IWQI. Which of the indexes in the list are most important? Should we just consider WQI? 

For example in WD: RSC is good, RSBC - Satisfactory, but PS is Injurious to Unsatisfactory. So how you interpret this?

 

Small comments

29 “interpretation groundwater quality” should be  “interpretation of groundwater quality”

SVMR model doesn’t show good results compared to ANFIS and this should be mentioned in conclusion and abstract.

Author Response

Reviour#2

Response: We greatly appreciate your critical observations as well as your constructive and helpful comments, which were helped us to further improve of manuscript. We hope that we could address your questions/comments by the explanations and revisions made in the manuscript. We believe that the manuscript is substantially improved after making the suggested revisions.

• Map of study area should be incorporated into the paper, instead of supplementary material, as it is important information.

 Response: Many thanks for this suggestion. The map of study area has been incorporated into the paper as Figure 1.

• 502-504 “The strong loadings of the salt ions in the groundwater of the WD may be attributed to the incursion of seawater from the underlying fractured limestone of the Eocene aquifer”. I am not sure about the seawater in WD region. It is located really far from both seas. According to the most publication, I found about El Fayoum groundwater’s main sources are indicated as a runoff of the Nile River and the agricultural runoff. Could you back it up with other reference?

 Response: The strong loadings of the salt ions in the groundwater of the WD may be attributed to the recharge from the underlying fractured limestone of the Eocene aquifer through hydraulic connection (El Sheikh 2004, refrence NO. 69). The deterioration of the groundwater quality could be resulting from high mineralization processes through rock-water interaction such as, reverse ion exchange, leaching, dissolution and precipitation processes (Gad and El Osta 2020, refrence NO. 71). In addition, contaminations by irrigation return flow through high application of agrochemical pesticides and the seepage from irrigation drainage canals (Gad and El-Hattab 2019, refrence NO. 70). This information has been incorporated into the text.  

• 507 Sea water? From where?

Response: This information has been revised and changed accordingly.  

• 545-548 Repetition, and see comment above

 Response: Repetition has been deleted.  

• 561-562 Not all groundwater. In previous paragraphs, WD samples were characterized as suitable with limitations. Then few paragraphs later, they are classified as “injurious to unsatisfactory”.

  Response: in this sentence, the quality of groundwater was classified based on residual sodium carbonate (RSC) variable, which all groundwater samples (100%) from the both regions were classified as good water based on this variable (Table 4). However, in the other paragraph, the quality of groundwater was classified based on other variables, which showed other classification. The classification of groundwater quality based on different irrigation water quality indices (IWQIs) was addressed based on individual IWOI variable as shown in Table 4.

• Again, you haven’t addressed my comment about IWQI interpretation. Your response is about WQI -

“Response: Many thanks for this reminder. A scheme for interpretation of WQI has been provided based on Meireles et al. (2010).”

However, you didn’t include interpretation of the IWQI. Which of the indexes in the list are most important? Should we just consider WQI? 

For example in WD: RSC is good, RSBC - Satisfactory, but PS is Injurious to Unsatisfactory. So how you interpret this?

  Response: The important of different variables of IWQI (RSBC, TH, PS, RS, and MH) for assessment of groundwater quality for irrigation purposes is already mentioned in introduction section (Lines 67-76) as well as in M&M section (Lines 227-248). In addition, as we explain in the subtitle “Interpretation of groundwater quality through physiochemical parameters using a multivariate analysis” The strong loadings of the salt ions in the groundwater of the WD may be attributed to the recharge from the underlying fractured limestone of the Eocene aquifer through hydraulic connection, as well as agricultural runoff into the quaternary aquifer system, which is considered as the most important aquifer in the WD region

Small comments

• 29 “interpretation groundwater quality” should be  “interpretation of groundwater quality”

Response: interpretation groundwater quality” has been changed into “interpretation of groundwater quality”

 

• SVMR model doesn’t show good results compared to ANFIS and this should be mentioned in conclusion and abstract.

Response: This sentence is already mentioned in abstract and conclusion as following:

In abstract “but the ANFIS model (R2 = 0.74–1.0) was superior to the SVMR (R2 = 0.01–0.88) in both the training and testing series”.

In conclusion “The ANFIS models offer a more accurate estimation of the different IWQIs in both the training and testing datasets (R2 was 1.00 in the training datasets and from 0.74 to 0.98 in the testing ones) than those from the SVMR models (R2 ranged from 0.29 to 0.88 in the training datasets and from 0.01 to 0.70 in the testing ones)”.