Expanding the Application of Sentinel-2 Chlorophyll Monitoring across United States Lakes

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript successfully demonstrates the feasibility of estimating chl a concentration in freshwater lakes and reservoirs at a broad scale by applying the MCI or NDCI to S2 data. Both algorithms can estimate chl a concentration over a wide range of conditions, enabling transformation of those estimates into trophic state. Generally, MCI outperformed NDCI by a small margin. This work parameterized, assessed error, and demonstrated application of these algorithms, previously only validated at select few individual lakes, at a broad spatial (103 lakes) and temporal (4 years) scale. This study contributes toward algorithm maturity, transitioning from accuracy assessment using a small number of measurements to employing a larger, more spatiotemporally expansive dataset.

It is a very interesting approach, attractive to many scientists.

 

In the following lines you will read my overall comments on the manuscript:

The subject of the manuscript falls within the scope of the journal.

This is a very interesting contribution for the specific area.

The results are of sufficient impact for publication in an international journal.

The interpretations and the conclusions are sound, justified by the data and consistent with the objectives.

The organization of the article is very satisfactory.

The title of the manuscript clearly reflects its contents.

The abstract is sufficiently informative, especially when read in isolation.

The keywords provided are appropriate.

The introduction sets the manuscript in an international context and shows how it builds on previous work on the subject.

The statement of objectives of the manuscript is adequate and appropriate in view of the subject matter.

The methods are correctly described and sufficiently informative to allow replication of the research.

The results are clearly presented.

The references are adequate for the subject.

 Overall, the manuscript is very interesting and useful to the scientific community, and I suggest being accepted as is.

 

Author Response

We thank the reviewer for their warm comments and positive reception of the manuscript. We are glad they find it valuable.

Reviewer 2 Report

Comments and Suggestions for Authors

Line 140: consider adding reference (if any) and/or website link

Line 190: Do you mean within 12 hours around satellite overpass (i.e., 6 before and 6 after) or 12 hours before and 12 hours after the overpass time? It is not very clear in this sentence. Please clarify and modify the text accordingly. 

In any case, it seems to be a large time window around the satellite overpass. 

Line 192: If data are sampled within 24 hours (+/- 12 around overpass) and the overpass time is close to the local noon, why not filtering using only the date and not taking into account the time at all?

Again, I think that 24 hours is a very large temporal window for data pairing. There should be diurnal variations in Chl-a concentrations.

Line 203: consider adding reference (if any) and/or website link

Line 210: However, Acolite showed lower error at 664 and 705 nm when compared to radiometric data. 

Why the error at 740 nm is more important, considering that the Chl-a peak is at 705 nm?

Line 221: +/- 30 minutes around overpass or +/- 15 minutes? Why not applying a strict filtering like this (maybe 1 or 2 hours around overpass) for all water bodies?

Line 304: Did you check the data for outliers before applying RMA? RMA is sensitive to the presence of outliers and they should be removed

Line 330: This refers to the 20% of the data used for testing or to both train and test subsets?

Line 331: Can these negative values be explained somehow? Maybe attributed to MSPM contamination or uncertainties in atmospheric correction? Or they are just low Chl-a cases?

Line 338: Consider calculating also the respective median metrics, which are not sensitive to outliers.

Line 448: According to which criteria the linear model was the best fit? It is not clear from Figure 4 (also Figure A1)

Line 455: Not sure about that. The ρt and ρs validation results are comparable, especially for MCI

Table 4: The derived validation metric values are quite high, i.e., the modeled Chl-a is 2x the in situ Chl-a on average based on the MAE and MAPE values. Consider using the respective median metrics, which are not influenced by high differences if the latter are few. 

Line 508: As mentioned before, consider removing extreme values before applying RMA regression, which is sensitive to the presence of outliers

Line 513: From Figure A1d, it seems that the problem is some quite high values of NDCI. This means that either the nominator is quite large (atmospheric correction not effectively removed using those bands) or the denominator is quite small. What are the typical values of Rrs and what are the values related to those extreme cases of NDCI?

Line 594: If I am not mistaken, this would lead to Chl-a underestimation, which is not the case here.

Line 718: The selection of the colormap may be misleading. Red is usually used for higher values and blue for the lower

Author Response

Thank you for your review. Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors Although the manuscript is not innovative, it has certain application value. From the experimental results, it still needs a great improvement.There some comments that I hope will help you improving this manuscript.

(1)    In Introduction, the authors have reviewed what is comprehensive enough. I suggest the author to highlight the current problems of this topic and the contribution of the author's research.

(2)    Line 61, “This is information is often provided ….” this sentence needs to be revised.

(3)    In in situ data section, the in situ data is very important for the training of the model. It is suggested that the author should strengthen the description of the quantity and rationality of the in situ data.

(4)    In algorithms section, the authors reviewed the MCI model. Here suggested that the authors clarify their application or improvement of the model.

(5)    From Figure 5 a and c, there is no consistency between in situ values and derived values, and this result is worrying. What is the reason? Can it be improved?

(6) From the Introduction section (“At high concentrations, chl a displays a distinct reflectance peak in the red edge spectral range near 700 nm”) and the Trophic state section (“Though both are measuring the reflectance peak at 705 nm, the maxi mum reflectance of the chl a spectral signal tends to be at shorter wavelengths, particularly at low concentrations”), Is there really a peak drift between different concentrations? What is the boundary between high and low concentrations, and what is the drift of spectral peaks corresponding to different concentrations?

Author Response

Thank you for your review. Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The comments have been addressed one by one in detail.