Advantage of Regional Algorithms for the Chlorophyll-a Concentration Retrieval from In Situ Optical Measurements in the Kara Sea

Round 1

Reviewer 1 Report

In order to improve the estimate accuracy for chlorophyll a concentration from ocean color data, it’s a critical step to develop a regional algorithm by comprehensive considering the local optical properties. This manuscript focused on the Kara Sea, an area which is strongly influenced by anthropogenic disturbance. The topic is meaningful. Data are much valuable. However, the data number is too small and the discussion is relatively simple. The manuscript needs more improvements.

My suggestions and comments are as follows:

1.     It’s easy for readers to understand the discrepancy of open ocean algorithms (OC4, GIOP) due to the complicated optical properties of water in Kara Sea. Therefore, the meaning to compare the results with OC4 and GIOP is not obvious. 

2.     The reason to select two bands (531 nm and 547 nm) as shown in Equation 6 could be added. These two bands are very close and there is only one band nearby in ocean color sensor. I think here comparison results between the relationships of Rrs at different wavebands with chlorophyll concentration are quite necessary.

3.     The spectral difference of Rrs between in situ measured data and the satellite data could be related with many factors. How about the diurnal variability of sea water? How about the effect of atmospheric correction in this high-latitude area? Are there any results from other studies? Rrs value at blue waveband measured with satellite are quite high, which is not consistent with that measure in situ. How about the relative contribution of CDOM to the total absorption? Is there any reports about CDOM in this area?

4.     Figure captions in this manuscript should be completed to state the full information about the data. 

5.     I think it’s not solid enough to get the conclusion “both regional algorithms can be used to obtain chlorophyll-a concentrations in the Kara Sea in concentration range <1 mg/m3” only based on the low RMSE. For clear waters the relative departure might be very large. The relative errors for the estimation should also be assessed. 

6.     Shortage of in situ data might be the main reason for the low accuracy. Many high quality in situ measurement are strongly suggested for the future work. Deep study about the optical properties of sea water in this area are essential for developing regional algorithms.

7.     The references could be reduced by focusing on the topic of this study.

Author Response

Dear Reviewer!

We are very thankful for the comments provided. In the revised paper, we have attempted to address all your comments and better the quality of the paper. Below are our replies (in bold) to each of the comments.

1.     It’s easy for readers to understand the discrepancy of open ocean algorithms (OC4, GIOP) due to the complicated optical properties of water in Kara Sea. Therefore, the meaning to compare the results with OC4 and GIOP is not obvious.

The reason for using OC4 and GIOP algorithms is that they are included in the standard NASA satellite Level 2 products and used most frequently in the marine bio-optical research. 

2.    The reason to select two bands (531 nm and 547 nm) as shown in Equation 6 could be added. These two bands are very close and there is only one band nearby in ocean color sensor. I think here comparison results between the relationships of Rrs at different wavebands with chlorophyll concentration are quite necessary.

This reasoning was in the text, L264-266, "The channels were chosen so that to exclude short-wave channels (412, 443, 469 nm for MODIS; 410, 412.5, 442.5 for OLCI) where the influence of atmospheric correction errors is known to be maximal".
The algorithm was designed for the future use with MODIS Rrs data, so it uses MODIS 531 and 547 nm channels.
In order to compare the results with different Rrs ratios we have added OC2 and OC3 algorithms, as the other reviewer suggested.

3.    The spectral difference of Rrs between in situ measured data and the satellite data could be related with many factors. How about the diurnal variability of sea water? How about the effect of atmospheric correction in this high-latitude area? Are there any results from other studies? Rrs value at blue waveband measured with satellite are quite high, which is not consistent with that measure in situ. How about the relative contribution of CDOM to the total absorption? Is there any reports about CDOM in this area?

We have mentioned all these factors in the discussion section - difficulties of the atmospheric correction and natural variability. In this work we do not concentrate on the nature of these differences.
We have added the discussion of the CDOM spatial variation influence on Rrs. 

4.     Figure captions in this manuscript should be completed to state the full information about the data. 

Information added to the captions.

5.     I think it’s not solid enough to get the conclusion “both regional algorithms can be used to obtain chlorophyll-a concentrations in the Kara Sea in concentration range <1 mg/m3” only based on the low RMSE. For clear waters the relative departure might be very large. The relative errors for the estimation should also be assessed. 

The relative errors were assessed, MAPE of regression presented in Figure 7. MAPE appears to be rather small compared to other algorithms.

6.     Shortage of in situ data might be the main reason for the low accuracy. Many high quality in situ measurement are strongly suggested for the future work. Deep study about the optical properties of sea water in this area are essential for developing regional algorithms.

We agree, and we certainly would like to have a possibility to get more in situ data, but the study area has not only complex water conditions, but also difficult weather conditions as well. We are lucky to have what we got.

7.     The references could be reduced by focusing on the topic of this study.

Some references were deleted.

thank you again for the review!

Reviewer 2 Report

This paper validated remotely sensed chlorophyll estimation algorithm in the Kara Sea using in situ and satellite data. Validation based on many Rrs/Chl-a datasets in the field is very valuable. However, due to the following matters that need to be corrected, the manuscript will be judged for re-peer review.

1. L545-692: It is very difficult to interpret the relevance of the whole sentence because the references are not numbered.

2. Title: Since this journal specializes in the marine field, it is not possible to judge whether it is a remote sensing research based on the "local algorithm" alone. Therefore, it would be better to add words like "remote sensed" or "for the satellite ocean color sensor" to the title.

3. Abstract L13, etc.: In this paper, there are many dates notation such as "10.08.2021", but it should be changed to notation such as "8 Oct. 2021".

4. L134-135: What kind of product is the "dark glass" in the "Lsky" measurement method? And, please write the approximate dimensions of the Cuvette (also in Figure 3(b)).

5. Figure 1: You should draw river shapes. Please include the name of the river.

6. The English in the caption of Figure 4 should be corrected. For example, “Measured remote sensing reflectance in ~”.

7. Table 3: Why not compare OC2 and OC3?　OC4 is not necessarily good for coastal areas. 8. Figure 9 and discussion: Isn't it natural that the error is high at high concentrations of chlorophyll because the original algorithm is log10(Chl)? It is self-evident that the error for high Chla concentrations is high, except for algorithms such as "Rrs(700 nm) / Rrs(670 nm)" that are often used along the coast. If you go to the trouble of discussing GIOP, you should add a little more bio-optical reason why K17 is more accurate than OC4. For example, in this case, if CDOM is one of the reasons, K17, an algorithm that avoids the CDOM absorption region, would be considered more effective.

That’s all

Author Response

Dear Reviewer!

We are very thankful for the comments provided. In the revised paper, we have attempted to address all your comments and better the quality of the paper. Below are our replies (in bold) to each of the comments.

1. L545-692: It is very difficult to interpret the relevance of the whole sentence because the references are not numbered.

We are very sorry, there must have been some glitch in the file. The numbering is restored.

2. Title: Since this journal specializes in the marine field, it is not possible to judge whether it is a remote sensing research based on the "local algorithm" alone. Therefore, it would be better to add words like "remote sensed" or "for the satellite ocean color sensor" to the title.

The algorithms we propose are for the in situ measured Rrs. Title changed, 
"Advantage of regional algorithms for the chlorophyll-a concentration retrieval from in situ optical measurements in the Kara Sea".

3. Abstract L13, etc.: In this paper, there are many dates notation such as "10.08.2021", but it should be changed to notation such as "8 Oct. 2021".

Date notation has been changed as proposed.

4. L134-135: What kind of product is the "dark glass" in the "Lsky" measurement method? And, please write the approximate dimensions of the Cuvette (also in Figure 3(b)).

Dimensions were added, and glass specifications were also added as a reference.

5. Figure 1: You should draw river shapes. Please include the name of the river.

Done.

6+. The English in the caption of Figure 4 should be corrected. For example, “Measured remote sensing reflectance in ~”.

The caption was corrected.

7. Table 3: Why not compare OC2 and OC3?　OC4 is not necessarily good for coastal areas. 

We have used OC4 beacuse it is a standard product provided with MODIS, OLCI and VIIRS data.
We have added the comparison with OC2 and OC3 for in situ Rrs data.

8. Figure 9 and discussion: Isn't it natural that the error is high at high concentrations of chlorophyll because the original algorithm is log10(Chl)? It is self-evident that the error for high Chla concentrations is high, except for algorithms such as "Rrs(700 nm) / Rrs(670 nm)" that are often used along the coast. If you go to the trouble of discussing GIOP, you should add a little more bio-optical reason why K17 is more accurate than OC4. For example, in this case, if CDOM is one of the reasons, K17, an algorithm that avoids the CDOM absorption region, would be considered more effective.

We have added the explanation of K17 advantages compared to other empirical algorithms, namely, it avoids the channels with atmospheric correction errors and channels with high influence of CDOM.

Thank you again for the feedback!

Round 2

Reviewer 1 Report

My suggestions have been considered in the revised version.I agree with the acceptation of this manuscript for publish.

Reviewer 2 Report

The revised manuscript has adequately responded to the review comments. Therefore, I recommend acceptance.