A Multi-Pixel Split-Window Approach to Sea Surface Temperature Retrieval from Thermal Imagers with Relatively High Radiometric Noise: Preliminary Studies
Round 1
Reviewer 1 Report
The study was performed with SLSTR data at 1000m spatial resolution.
It would be interesting to perfom the same with a higher resoluted sensor as data are available (ECOSTRESS, ASTER). Actually, I am not totally conviced by the extension of the conclusion obtained for SLSTR to high resoluted sensors with higher NeDT. How clouds will affect results on these sensors ? How the moving window of 3x3 pixels should be optimized for high resoluted sensors ?
Results about the average and StD of the SST retrieved for one pixel with the method are not shown and are missing. These results should be a relevant information on the robustness of the method.
Moreover, the study would benefit to use some validation in situ data in order to evaluate the performances of the method in terms of SST retrieval in comparison to classical SW approach. What would be the precision of the SST retrievals in K with the use of the multi pixel SW in comparison with a classical approach ?
Author Response
See attached pdf detailing the answers and the revised version of the manuscript.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The article presents a thorough examination of the effects of noise and spatial window averaging on atmospheric correction. Specifically, the split window algorithm with a 3x3 window averaging is recommended for a 1km resolution infrared sensor. However, some questions remain such as the significance of the empty parentheses in the abstract and the presence of an additional label (b) in Figure 3.
This paper provides a noise reduction solution for the data processing of SST products of high-resolution infrared remote sensors with relatively high equivalent noise, but the substantial contribution to SST products will not be obvious. Authors should compare true high-resolution infrared data with general-purpose data, such as landsat9 and S3 SLSTR, to show the impact of resolution and equivalent noise on the conclusions. The conclusions of the article are consistent with the arguments, and the references are adequate and reasonable. The figures and units of the chart in the article are relatively good, basically do not need to be modified, except for Figure 3. The definition of high resolution in this article is not clear, and the spatial resolution of 1km as a representative of high resolution is difficult to be convincing.
Author Response
See attached pdf detailing the answers and the revised version of the manuscript.
Author Response File: Author Response.pdf
Reviewer 3 Report
The paper describes a methodology to optimize the process of See Surface Temperature analysis through Temperature Infrared satellite Imagery.
The paper is well organized and well written and deserves publication. However, I would suggest the following changes.
The abstract should be amended. Between all, there are two couples of brackets instead of two words in the first sentence (words are both acronyms and are visible on the web site). In my opinion, Acronyms should be explained at their first use even within the abstract but I do not know the editor’s policy about.
The statistical variables and indices could be better explained, even the well-known ones.
Figure regarding results and discussion should be enlarged and better explained.
In Par 2.1 I would better explain the characteristics and the meaning of “data flag” and “quality flag”.
Author Response
See attached pdf detailing the answers and the revised version of the manuscript.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
I thank the authors for addressing the question of the extension of the results to high resouted sensors.
