Linking Clusters of Micropollutants in Surface Water to Emission Sources, Environmental Conditions, and Substance Properties

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript aimed to find substances clusters to link to possible emission sources, physico-chemical properties and environmental conditions. This is an interesting work focusing on a key point when evaluating of environmental occurrence and fate of pollutants and their association to specific aquatic system.  Although author did not find an application for the two selected rivers, but this work is the starting point to find clusters able to associate groups of pollutants to emission sources environmental conditions and substance physico-chemical properties. The manuscript is well written and discussed. There are only some comments that should be considered by the authors.

Section Methods. Why authors selected these two rivers. Please clarify in the manuscript.

 

Overall, authors failed in linking substances groups to possible emissions sources, environmental conditions and substances physico-chemical properties. Thus, what can be change in the methodological approach. This should be critically discussed in the manuscript. Also, why authors did not apply this methodology to other smaller aquatic systems to tune data selected to develop this methodological approach.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

this is an interesting paper, and should be published but I found the description of the methods and hence the findings quite confusing.  so I would recommend that there should be a major revision to clarify the text.  there are also some statistical errors.  clustering of substances is a powerful approach, but we also appear to be clustering sites.  also to be clear we have time series of observations, so I think in the main text some preliminary plots of the data would be useful.

I would suggest to tighten the language, so eg not use the word sample.  rather we have observations indexed by date and place (station or site), and measurement of multiple substances.  I think that with this clarity of notation, the clustering methodology will then become clearer.  I would suggest a flow chart to demonstrate the pre-processing steps needed.  regarding limits of detection, the method used has been shown to be wrong, the value should not be replaced by a zero.  so I would suggest that should be corrected.   please also explain more clearly the identification of the number of clusters.  I am more familiar with the plotting of a line graph looking for a dog leg.  could there also be more details about the actual associations and relationships with conditions, the boxplots are interesting but would there be any interest in a more regression model.

Comments on the Quality of English Language

this is generally clear.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The amendments are now clearer, I would recommend for limit of detect, that the authors investigate the NADA package in R, this provides a much more statistically sound approach to dealing with limit of detect, this could be used.

Author Response

Please see the attachment

Author Response File: Author Response.pdf