Relative Merits of Optimal Estimation and Non-Linear Retrievals of Sea-Surface Temperature from MODIS

Round 1

Reviewer 1 Report

This manuscript present a comparison of two widely used SST retrieval methods: NLSST and OE. The algorithms are applied to a set of data acquired by MODIS sensor and collocated with drifting buoys, moorings and ship mounted radiometers. SST retrieved by NLSST and OE are compared to in situ measurements and analysed to determine the strength and weaknesses of both retrievals.

This topic is perfectly in line with the scope of the journal.

 

General comments:

The manuscript is clear, well written and easy to follow. This type of work is very much needed because it is often difficult to see clear in the various retrieval methodologies.

While this study is very thorough I have three major concerns:

1- Application of OE: Merchant et al. (2009 for instance,[31] in the manuscript) highlight the necessity to adjust simulated BTs before applying OE. This is done to account for some systematic errors. I believe the authors have not performed that step and I think this really compromise the usefulness of the results. Do the authors have justifications for not doing it? Can they think of a way of doing it? It is usually performed on satellite data (scenes) rather than matchups, and this is extremely heavy to do because it entail performing simulations for all pixels in all scenes.

2- Selection of data points: If I understood correctly, the authors have selected the matchup data according to MODIS SST best quality level (0 and 1) and then evaluate the performance of the OE based on this selection only. To me this necessarily bias the analysis. Can the authors comment on that?

3- The comparison of the results is performed by looking mostly at the bias (and the RMS). I would have liked to see the standard deviation of the difference between in-situ and satellite SST for instance. It is not enough to focus on the bias especially as with NLSST the bias can easily be tuned (regionally for instance) to be 0.

 

Below is a list of more specific comments:

L 131: did you mean “prior SST” instead of “retrieved OESST”? Or perhaps you meant “source” instead of “consequences” in the line above? Do you actually show results of OE retrieval with prior from in situ measurements?

L 210: please explain condition number.

L286: why talking about subskin here?

L301-303: diurnal warming does not happen everywhere so it is not surprising that on a global dataset there is no clear relationship.

L366: “Good et al” (not “all”)

Section 6.1: This section seems a little out of place here because there is no clear link with the results presented and no transition.

L499-500: You do not show SST0-SSTb!

L562-563: It is mentioned earlier in the paper that the prior SST has a weak contribution to the retrieved SST. Can you comment?

L571-583: I disagree, in my opinion part of what you see is due to the fact that the BTs have not been adjusted.

Table 1a,b and 2a,b are difficult to read. Could they be represented as plots instead? STD is needed, see comment #3 above.

L638-641: In my opinion it is unlikely to be a reliable diagnosis of the potential of OE to improve on NLSST because this analysis . Can you comment?

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The work is aimed at comparing  of methods for

 retrieving  the sea surface temperature from satellite data - using only satellite data NLSST and additional information OESST. The data are compared with buoy measurements in the North Atlantic Ocean.

In the general case, the difference between the temperature values ​​measured (estimated) from satellite data is determined not only by the methodological errors of the temperature estimation algorithms, but also by factors that determine the physical difference between the temperature measured by contact data at a point and the satellite data averaged over the area. In this case, spatial and temporal variability of the SST field, including vertical, especially during the period of daytime warming.

Some recommendation

Different fonts in the text and formulas make it difficult to read.

Large tables do not facilitate data comparison. Would it be better to present the information in a graphical form?

More info needs for ERA5 SST0 description.

I would like to see the temperature fields, and their difference.

After the remarks have been corrected, the work can be published.

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Dear Authors,

thank you for answering and taking inton account my questions and comments.

I do not fully agree with all your answers but it is a matter of opinion and I consider that your manuscript is now ready for publication.