Assessment of the Radiometric Calibration Consistency of Reflective Solar Bands between Terra and Aqua MODIS in Upcoming Collection-7 L1B

Round 1

Reviewer 1 Report

In this paper authors present some of the major improvements to the MODIS RSB calibration algorithms in the upcoming release of C7 L1B algorithms. Authors present the performance of the C7 L1B product through intercomparison to the VIIRS. Different intercomparison methods and the observations used in these intercomparisons are described.  Authors show that with the improvements to the calibration algorithm the Terra and Aqua MODIS C7 L1B products are stable within 1% over the mission period, and that two L1B  products are in agreement exhibiting good RVS characterization across the scan. 

Paper presents an important research work, i.e. generating a stable calibrated L1B product from two aging MODIS instruments. Authors identifies the product performance issues in the earlier version of the product, root cause of the problem and mitigation approach adopted in the upcoming collection. Analysis methods and data used in this work are clearly described. 

Author Response

See attached file

Author Response File: Author Response.docx

Reviewer 2 Report

This is an expertly written article that I enjoyed reading. The results are well-presented and relevant. Also, they are supported by background studies. 

I will only propose one change which is regarding the formatting of Equation 1 which is clumsy as it is. I propose that the authors should work on the alignment of this equation, relative to the rest of the text. Also,  \frac{}{} should be used in LaTex to have better formatting of the numerator/denominator.

Also, on line 92, after "expressed as", there should be a ":" that leads into the equation on the next line.

Author Response

See attached file

Author Response File: Author Response.docx

Reviewer 3 Report

This paper uses four independent vicrious methods to assess the calibration stability of Terra and Aqua MODIS RSB and their radiometric consistency using what to be produced by the Collection 7 reprocessed L1B.  It is a very meaningful reference for readers or users to understand the quality of modis Collection 7 products.  It can be acceptable for publication, but several minor issues need to be addressed to enhance the reliability of the article.

 

1.The bi-directional reflective properties of the surface greatly affect the uncertainty of the product results assessment, this article, although all four methods of bi-directional reflective correction methods are briefly stated, in order to better enhance the readability of the article, it is recommended to refine a correction method.

2.Chapter 3 of this paper describes the relevant methodology used to assess the data, but it is too general, and it is suggested that the reliability of the assessment methodology be increased by listing mathematical formulas.

3.Is it possible to add the results of the collect 6 version to the trend comparison of the assessment results? Plotting the trend of collect 6 vs. collect 7 on a graph would provide a better visualization of the enhancements of collect 7.

4.line296-298,says 'Trending results indicate that the desert site is superior to the Dome C site in tracking satellite sensor calibration performance due to its excellent long-term stability with relatively low uncertainties." , why ? can explain it?

5.line316-317 says 'Results from the DCC approach show better stability than the other approaches.' why? 

Author Response

See attached file

Author Response File: Author Response.docx

Reviewer 4 Report

The intent of the paper is to provide users of the NASA MODIS L1B product, the anticipated RSB stability and radiometric scaling factors and uncertainties.  They provide a Table of the Terra and Aqua MODIS stability and a table of Terra and Aqua radiometric scaling factors as a function of scan position. Although excellent RVS performance was claimed in the abstract, the individual MODIS sensor RVS was not evaluated. The radiometric scaling factors are not consistent, and many differ by more than 1% amongst the Libya-4, Dome-C and DCC approaches at nadir and more than that by BOS and EOS in their table 2. It is difficult to validate the EOS and BOS methodology if the approaches provide inconsistent Terra/Aqua ratios. If the authors goal is to demonstrate the C7 had improved RVS factors, than the authors would need to show that the MODIS C6 RVS Terra/Aqua ratios are more disparate than the C7.

 

This paper will be used as a definitive paper to cite the C7 calibration accuracy and the authors must show due diligence of their statements. It almost seems that adding BOS and EOS scaling factors was an afterthought. I believe the authors need to demonstrate that the nadir view approach is consistent amongst Libya-4, Dome-C and Dome-C and then continue the study by scan angle. The overall goal of this paper is admirable, however, the lack of detail, lack of discussion of approach result discrepancy, and hand waving of conclusions require a major revision addressing my comments.

 

General comments

 

Line 18 abstract, “Comparison results show the VIS/NIR bands are in good agreement around nadir and at beginning of scan for all the  approaches. For cases at the end of scan, the agreement varies depending on the approach but is typically within ±2%, indicating an excellent on-orbit RVS characterization.” The case for this statement has not been made. Desert, Dome-C and DCC results in Table 2 are not consistent.

 

The individual band figures 7, 8, and 9 have no corresponding trend slope and uncertainty and corresponding Terra and Aqua calibration ratio.

 

Table 1 is missing the associated standard errors or some kind of significance statistic. Table 1 is the combined desert, dome-C and DCC stability approaches. The text does not describe how the approaches were combined. Table 1 should have the same format as Table 2, where the BOS, NAD and EOS are stratified, the radiometric scaling factor for each stability approach and associated uncertainty. One table for Terra and another table for Aqua.

 

The Terra and Aqua BOS and EOS radiometric scaling could be aliased with unresolved BRDF issues. Here are two tables to illustrate this

West	Terra	East
	Descending node during day	Location of Sun
End of Scan	Nadir	Beginning of Scan
Backscatter		Forward Scatter

 

West	Aqua	East
Location of Sun	Ascending node during day
Beginning of Scan	Nadir	End of Scan
Forward Scatter		Backscatter

Based on Table 2 visible channel results, I am concerned that the DCC BOS and EOS both have a greater ratio than for nadir scans, whereas for Libya4 The BOS has a positive ratio and EOS has a negative ratio. For Dome-C no generalizations can be made.

 

There are no other papers that specifically quantify the BOS and EOS Terra and Aqua radiometric scaling. This makes it difficult to validate the methodology and to quantifiy the actual instrument scaling difference. To validate the methodology consistent scaling factors between the desert, dome-C and DCC approaches would need to be demonstrated. The current status of providing BOS, NAD, and EOS Terra and Aqua scaling factors is that it is within the noise of the approaches and is therefore not conclusive. If the purpose of this paper is that C7 is an improvement over C6, then Table 2 can be generated with C6 results, and show that the ratio ranges have been reduced.

 

Specific comments

 

line 18, of the Terra and Aqua calibration consistency, replace “their” with “Terra and Aqua”

 

line 229, There is not enough information on the off nadir DCC approach in this paragraph. What is the VZA threshold for DCC in this study? What would be convincing are the BOS and EOS DCC PDFs as part of Fig. 4. Is there a distinct seasonal cycle of mean DCC reflectances? What is the main reason for the lack of sampling at large VZAs, is the increased pixel size?

 

Fig. 5. Some geographical information would be helpful. Is this a full MODIS and VIIRS granule.

 

Fig. 7 and 8. Maybe I am missing something but the Libya-4 and Dome-C BRDFs are based on the first 5-year of Aqua-MODIS observations. The Libya-4 Aqua normalized reflectances are close to 1.0, however, The Dome-C seem to be slightly higher than 1.0, especially the first 5-years.

 

Fig. 7 and 8. The Terra-MODIS/NPP-VIIRS SNOs occur over the poles. These intersects occur at local noon near the north pole limiting the sunlit observations mainly during summer. However, the SNOs in the figure occur for more than half a year. What is the SZA threshold utilized.

 

Fig. 7a,b and 8a,b, Fig. 9a, 9b, and Fig. 10a and 10b. I would highly suggest plotting the Terra and Aqua linear trends (shown as lines) and the associated slope in percent and the standard error, that can be tied to Table 1.

 

Table 1 is missing the associated standard errors or some kind of significance statistic.

 

Table 1. The Terr-MODIS stability for bands 1 through 4 are slightly negative for the Libya-4 and DCC approach, whereas for Dome-C the trend is greater than 1%. The Dome-C methodology does not seem to be the issue because the Aqua-MODIS Dome-C stability is close to 0. Is there an explanation for this?

 

 

Author Response

See attached file.

Author Response File: Author Response.docx

Round 2

Reviewer 3 Report

 These minor issues has been fixed in the  revised manuscript, thanks for your work.

Reviewer 4 Report

I have reviewed the manuscript and found it acceptable for publishing