Retrieval and Uncertainty Analysis of Land Surface Reflectance Using a Geostationary Ocean Color Imager
, Eunkyung Lee 1, Donghyun Jin 2, Noh-Hun Seong 2, Daeseong Jung 2, Suyoung Sim 2 and Kyung-Soo Han 2,*Round 1
Reviewer 1 Report
This is a technical paper that demonstrates the use of a lookup-table (LUT) based algorithm to retrieve land reflectance from the Geostationary Ocean Color Imager (GOCI) measurements. The results have been well demonstrated and validated. The work provides the community with a new surface reflectance data set. It can be accepted as a meaningful publication.
However, a major pitfall in this paper is the lack of clear discussions about the methodology and the data used in the paper. For example, readers may be interested in knowing more details about how the atmospheric correction formulas (Equations 2-3) are established. Besides the explanation for the atmospheric correction, authors are expected to provide more details about how to train coefficients ???, ???, and ??? using 6SV simulations. There are several other poorly presented discussions. One example is the lack of clearly quantified colocation information between GOCI and ARONET, ECMWF CAMS datasets. Although the uncertainty estimation for the atmospheric parameters has been presented in the paper, authors still need to quantitatively specify how well the ARONET data and the ECMWF CAMS data are temporally and spatially matched with the GOCI data. And the discussion in the validation section is very confusing due to the phrasing issue and the lack of details. For example, one can only guess that the GOCI LSR results that are validated using the MODIS data are retrieved using TPW and TCO from ECMWF CAMS, but have no idea about the source of the AOD values used for the retrieval. Details about how to construct atmospheric correction using ARONET site measurements without the LUT approach are not presented in the paper. Without those details, the validation for the GOCI LSR retrieval algorithm is not convincingly presented.
And, as a general suggestion, the paper also needs copy editing looking purely for grammar errors. There are cases of awkward phrasing throughout the manuscript. I only note a few grammar corrections in this review, but there are numerous cases throughout the text.
I include several specific comments below for the authors to address.
Line 43-47: The introduction for “atmospheric correction” is really unnecessary.
Line 72: “Conclusion [Shuai]” The reference needs to be fixed.
Line 185: “AER-ONET” or “AERONET”
Line 196: “After spatiotemporal filtering to construct matchups with GOCI and quality control 196 to meet the AERONET criteria” Can you be more specific about the spatial-temporal matching criterion?
Line 214: “contrib-uted”? Why adding a hyphen here?
Line 218: “com-plex” unnecessary hyphen again
Line 229: “High accuracy can be realized when the increment of input parameters is low” Consider rewriting this sentence.
General question: The accuracy of the lookup table (LUT) approach fundamentally depends on how linear the relationship between the observation and the LSR is. It is the step size of the LUT entries, not the multivariate interpolation method, that determines the accuracy. The authors should provide more discussions to convince readers that the current setting of LUT entries is sufficient for the LSR retrieval application.
Line 245: Please define “temp” in Equation (2) and “??“ ” in Equation (3)
Line 259: “Reference datasets of LSR can be constructed using a reliable RTM such as 6SV, which has been found to agree within 0.4%, of Monte Carlo results, with accurate atmospheric properties”
More discussion for the Monte Carlo results and the corresponding references need to be provided.
Line 292: Please provide the full name of BIPM.
Line 330: “According to Merni et al. [72], comparing two products, even those generated from the same sensor, is difficult; therefore, we only performed qualitative comparison.”
Such argument is vague, please elaborate why comparing two products is difficult
Author Response
We thank reviewer #1 for the constructive and insightful comments, which have helped us to substantially improve our manuscript.
Our response to your comments can be found in the attachment.
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The paper “Retrieval and uncertainty analysis of land surface reflectance using Geostationary Ocean Color Imager” by Kyeong-Sang Lee et al., describes the calculation of land surface reflectance (LSR) from top-of-atmosphere (TOA) images of the Geostationary Ocean Color Imager (GOCI) by using the land atmospheric correction algorithm basing to the 6S vector (6SV) radiative transfer code. For that, the lookup table (LUT) approach and simple correction formula has been used. In order to reduce the errors between the 6SV LUT values, additionally the 6D interpolation has been used. However, in reference [57] (by Kyeong-Sang Lee et al., Improvements of 6S Look-Up-Table Based Surface Reflectance Employing Minimum Curvature Surface Method), this approach is already thoroughly described by at least partly the same group of authors. Explanations presented in [57], in Fig. 1, equation (3) and symbols used there are clearer than paragraph 3.1 and equations (2) and (3) in present paper; I suggest revision by authors of that part.
For validation of estimated GOCI LSR data, comparison with LSR data obtained at the same time and in similar spectral conditions from Moderate Resolution Imaging Spectroradiometer (MODIS) products, and also with ground measurements of LSR is applied. However, the uncertainty of compared data has not been incorporated into the interpretation of suitability and efficiency of the method used by authors for derivation of GOCI LSR data.
Uncertainty evaluation of the estimated GOCI LSR data is a novel part of article. Uncertainty evaluation includes three input factors for the uncertainty: total precipitable water (TPW), total column ozone (TCO), and aerosol optical depth (AOD). Validation of the obtained uncertainty estimates is still not sufficient.
I suggest that authors should consider to use En numbers for the evaluation of the agreement between the LSR results obtained from different sources (measurement systems satellite/satellite or satellite/in situ, two different calculation models, etc.) En numbers can be calculated following [ISO 13528:2015(en) Statistical methods for use in proficiency testing by interlaboratory comparison] as follows
Equation for En, see the attached document,
where x1 and x2 are the independent results subject to comparison; U1 and U2 are the expanded uncertainties of these results with k = 2, respectively. The agreement between the compared values is considered satisfactory if |En|≤1 and non-satisfactory if |En|>1. When uncertainties are estimated in a way consistent with the Guide to the expression of uncertainty in measurements (GUM), En numbers express the validity of expanded uncertainty estimate associated with each result. A value |En|<1 provides objective evidence that uncertainty estimate is realistic and consistent with GUM definition.
I recommend that after revision by authors the article may be published in Remote Sensing.
Numerous small technical mistakes in the text of the article should be corrected. Some suggestions for authors are listed below.
- Row 134: where ?? is the TOA radiance (W⋅m−2⋅sr−1⋅μm−1) of GOCI image in band ?; ??? is the digital number from the GOCI image in band ?; ?? and ?? are calibration coefficients gain and offset, respectively.
- Row 227: (Table 3)
- Equation (2): Instead of symbol LSRi, more standardized symbol, for example ρs,i is advisable; explanation in text for tempi and better symbol for it is advisable.
- Table 3: More compact table is advisable: abbreviations for LUT entries and SRF; instead of column "values" should be two columns: "range" and " increment".
- Equation (4) to (6): Symbols that are more compact are advisable. The same index i as in equation (1), better to have different indices here and in (1).
- Figure 6: Quite likely, difference between the satellite and in situ data can be better revealed if presented by ratios of these values.
- Table 4: Excessive number of significant digits in numerical values shall be deleted (at least in Percent values).
- Figure 8: Instead of “total combined uncertainty” depending on content, should be “standard combined uncertainty” or “expanded combined uncertainty”.
Comments for author File: Comments.pdf
Author Response
We thank reviewer #1 for the constructive and insightful comments, which have helped us to substantially improve our manuscript.
Our response to your comments can be found in the attachment.
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
After the revision efforts by the authors, I would recommend the acceptance of the paper for publication. However, I still see grammar errors and awkward phrasing throughout the manuscript even though the authors claimed that "English in the document has been checked by at least two native professional editors." For example. the way of listing the radiative transfer models with both full name and abbreviations in Line 226 are not correct. And the references for those listed models should be provided. There are also typo errors like "performance static".
Anyway, I guess it will be the job of the editor's office to work with the authors to further ensure the language quality of the publication.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
