Assessment of BRDF Impact on VIIRS DNB from Observed Top-of-Atmosphere Reflectance over Dome C in Nighttime

Round 1

Reviewer 1 Report

Great article and well written. Only three minor comments.

1: Too many acronyms in abstract

2: Wavelength of DNB only on page three (line 108), suggest this should be earlier in the document.

3: Too many very similar graphs. Is it possible to tabulate some of the graphs instead?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper addresses the important and challenging problem of calibrating the VIIRS/DNB high-gain stage.  Dome C in Antarctica is used as a calibration target and three models are tested to represent reflected lunar illumination at night.  The illumination and viewing geometries exhibit great diversity and modeling them is deemed crucial.

Specific comments:

1. “Nighttime imagery” could be IR. “Nighttime visible imagery” or “low-light imagery” would be better terms.
2. It is not clear that daytime and nighttime BRDF models should differ as a matter of first principles. The scattering physics should be the same, only the illumination differs (sun vs. moon, anthropogenic sources, air glow, aurora).  The authors make a point that “no nighttime models exist”.  I am not sure that is important.  Maybe this should be expanded upon.
3. The authors make the point that the atmospheric conditions at Dome C make it an ideal ground target site. The DCC method uses targets nearly at the TOA.  Can atmospheric transmission be quantified and compared to the DCC case?
4. RossLi is selected as one of the three BRDF models. This makes sense given its use in the MODIS community, but it might be best suited for generally different types of targets.  As I understand it, there are Ross volume scattering thick and thin variants, and the Li geometric part can be parameterized.  The model described may be more appropriate for canopies than a uniform snowpack.  So, why did the authors adopt it?  Some motivation and expansion of the discussion in Sec. 2.3.1 would be good.  Maybe it is included because of its MODIS legacy only.  It seems somewhat of a strawman against to compare the other two.
5. The Warren model is ultimately shown to be the best one. This is hardly a surprise given that it was developed for exactly this target and it has an ample number of parameters (12) to use for fitting.  The Hudson model should also have been expected to do well for the similar reason.
6. Sec 2.3 could benefit from a good drawing identifying the angles.
7. Equation 4 uses pi-phi as the angle operated on by cosine and cosine of the double angle. What is the benefit to this notation when a plain cosine would do?  Equation 4 is not a Fourier series in the general sense since the sine terms are missing.  The latter could be present with appropriately textured surfaces but presumably are not useful in this context.  Is it worth a sentence to address this?
8. Table 3 has a lot of unnecessary precision in the decimal numbers.
9. The “V” distribution identified in Sec 3.3 is due to “opposition” effects at high illumination and low SNR at low illumination. A brighter moon should not necessarily be a “bad effect”.  This would be a deficiency with the MT2009 model.  The authors show an interesting signature and perhaps the analysis here could point towards an improvement?  It seems a missed opportunity.
10. Sec 4.2 and conclusion cite RMSE that I think should have units.
11. The authors miss an opportunity to compare with DCC results.
12. The paper would be of more general use to the community if the actual calibration results over time were discussed in detail for each VIIRS.  The paper is mostly an exposition of a method.  The regression lines in Figure 27 seem to indicate that systematic adjustments could be justified to the operational calibration.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Dear authors,

your work is really interesting although surprise me that you have not heard of Román, M. O., Wang, Z., Sun, Q., Kalb, V., Miller, S. D., Molthan, A., ... & Seto, K. C. (2018). NASA's Black Marble nighttime lights product suite. Remote Sensing of Environment, 210, 113-143.

Please, put in context your results with the Roman et. al. 2018.

Also, I find that the scatter of many of your plots is too large and the number of points too small to assume normality, and too few points. You should make a nonparametric linear fit and give errors of the fits. Bootstrap can be of great help in this kind of situation.

 

 

 

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

Thanks for addressing the changes.