Extending AVHRR Climate Data Records into the VIIRS Era for Polar Climate Research

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

General Comments:

Not being a remote sensing specialist, I cannot comment on the technical aspects here. As a potential user of these data sets however, I think the manuscript falls short in many ways. Perhaps the issues below are raised elsewhere but this is a major opportunity to address in one place the key aspects below for the benefit of users and to make the case vigorously why these data sets are so important.

First, because you are dealing with visible and thermal imagery, the issue of cloud clearing is a foremost concern to be able to detect surface conditions, especially for the cloudy Arctic in summer. A reasonable summary of the steps employed, and their validity is needed.

Second, I presume that the many cloud variables discussed in Wang and Key (2005) are still being produced. If so, it is important to say this clearly. Only total cloud fraction is listed in Table 4. Certainly, the surface radiative fluxes depend on the inferred cloud conditions.

Third, there is very little context provided for these climate data records. What other data sets, especially from NASA and ESA, are available and how do they compare with yours?  This addresses the importance of these products.

Fourth, some contemporary validation of the geophysical variables is needed: NASA MODIS for skin temperature and surface albedo and NASA CERES for radiation fluxes at the surface and TOA. LUT in Table 4: isn’t this OLR?

For sea ice thickness, I am not aware of anyone using these estimates.  Thickness estimates from Icesat lidar have been emphasized by NASA as a major achievement for both poles using freeboard measurements but with uncertainties in the characteristics of the snow cover on sea ice being an impediment; this would seem to be a key question for these estimates based on the surface energy balance. Cryosat was launched in 2010 and is still operating. Here is a combined sea ice thickness product on monthly time scales from Icesat and Cryosat for the Arctic: https://nsidc.org/data/nsidc-0773/versions/1. Do your data show the major thinning of the Arctic sea ice since the 1980s?

Specific Comments:

1. Table 2: the isolated single numbers look strange. Most obvious for a4 for Channel 2.
2. Your comparison figures are for midsummer conditions. Is this high sun period when your algorithms work best?
3. In general, your results for the Arctic look better than the Antarctic. In Fig. 8, the broadband surface albedo from APP-x and VPP-x appear to differ by 0.1, a large difference. Why is this?

Author Response

Please see the uploaded WORD document for the responses to reviewer's comments. Thank you! 

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The work presented in this manuscript is interesting and potentially valuable, particularly in its attempt to bridge long-term AVHRR data records with newer VIIRS observations. The products described could provide important contributions for climate research and operational monitoring if properly validated and documented. However, in its current form, the manuscript requires substantial revision before it can be considered for publication. My detailed comments are provided below.

Major Concerns:

1. Introduction 
The motivation for this work is not clearly described. From a sensor perspective, VIIRS is significantly more advanced than AVHRR in terms of the number of spectral bands and the accuracy of spectral calibration. Therefore, the rationale for calibrating high-quality VIIRS data to the relatively lower-quality AVHRR data needs to be well justified. Historical consistency may be one of the main reasons, but the authors should provide additional information, references, and supporting evidence to strengthen this point.

2. Suitability of the Regression Model
The regression model (Equation 1) has not been thoroughly validated. The authors assume that one set of coefficients (a0, a1, ..., a4) can be applied to the entire study region (Arctic or Antarctic), with only a basic separation between AM and PM overpasses. However, such large areas consist of diverse surface types, including water, sea ice, land with and without vegetation, and snow-covered surfaces. These different surfaces exhibit distinct BRDF (Bidirectional Reflectance Distribution Function) characteristics. Applying a single set of coefficients mixes these differences, making the quality of the final outputs uncertain. The authors should conduct additional experiments to investigate this impact. For instance, regression coefficients could be derived separately from spatially randomized groups or by surface-type groups, and their stability compared. If significant differences are found, this would indicate that the regression model is surface-type dependent, and appropriate adjustments should be made. Furthermore, the magnitude of the model error or residuals should be quantified and discussed.

3. Validation and Verification
Validation currently considers only one metric: the overall mean bias between the two products (VPP and APP) across the entire domain and study period. This is insufficient. From a statistical perspective, the variability (e.g., standard deviation) of this bias in both space and time should also be presented. Important details or anomalies may be hidden in mean values but revealed in variability measures.

4. Results Sections (Sections 4 and 5)
The results are presented in a mechanical way, with figures and tables listed but no meaningful interpretation or discussion. These sections should be reorganized to provide more analytical content, including explanations of the observed differences and potential causes.

4. Conclusions (Section 6)
Section 6 also requires revision. The first paragraph largely repeats the comparison results, while content in the second and third paragraphs would be better placed in Sections 4 and 5 as part of the results and verification discussion.

Specific Comments (by line number):

L33–37: Provide variability (e.g., error ranges or standard deviations) alongside each mean bias value. The same applies throughout the manuscript.

L43–48: This sentence is overly long; break it into several shorter, clearer sentences.

L54: Indicate the spatial resolution of the grid points (e.g., 5 km for APP and 25 km for APP-x?).

L63: Provide the full name of CLARA upon first use (likely Cloud, Albedo and Surface Radiation).

L90–91: Clarify the feasibility of intercalibration. AVHRR and VIIRS have different spectral bandpasses (widths and response functions). Please discuss potential implications and cite relevant references.

L102: Clarify the definition of composite times (e.g., target times 04:00 and 14:00 for the Arctic). While satellites cross the Equator at fixed LST, this is not the case in polar regions, which may confuse readers. Is this actually based on ascending and descending orbit composites?

Table 1: Add a column listing the names of the corresponding VIIRS channels.

L118: The stated angle threshold (“+0.1 to –0.1 degree”) seems unrealistically narrow. Please clarify whether this is truly the selection criterion; it is difficult to imagine all three angles meeting such strict conditions simultaneously for both AVHRR and VIIRS.

Table 2: Adjust formatting to avoid splitting negative signs and final digits across lines.

Figure 2: Coastlines and latitude/longitude lines are hard to see against the dark background. Consider changing them to white. Apply the same adjustment to other figures with similar issues.

Figure 3: Add the brightness temperature unit (K) to the legend. Apply the same for all other similar figures.

Table 3: Include variability values (e.g., standard deviations) for each mean bias.

L186–190: Rewrite for clarity and readability.

Figures 6–9: Consider adding a third column of panels showing the bias (VPP – APP) for clearer visualization.

Author Response

Please see the uploaded WORD document for the responses to reviewer's comments. Thank you! 

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This paper offers a clear and methodologically rigorous approach to extending the AVHRR-based Polar Pathfinder Climate Data Records into the VIIRS period. The intercalibration method is reliable, and validation outcomes show a high level of agreement between products derived from AVHRR and VIIRS. This research is very important for maintaining continuous monitoring of the polar climate. But the methodological robustness, error decomposition, and long-term trend consistency must be clarified before publication.

 

Major Comments

1. The introduction clearly establishes the motivation for the study. However, the literature review seems somewhat limited, relying heavily on the authors' previous work and NOAA program reports. While this focus is understandable given that the APP/-x datasets serve as the foundation of this study, it would enhance the introduction to cite a broader range of international efforts related to intercalibration methodologies and climate data record (CDR) extension between satellite generations. Besides, the challenge of coupling AVHRR and VIIRS products is not stated.
2. Section 2. VIIRS Global Area Coverage: Please include the spectral response functions for both AVHRR and VIIRS.
3. Section3 Intercalibration: The physical basis for band-to-band regression is weak. Please provide MODTRAN-simulated bias values caused by differences in spectral response functions, particularly for the infrared channel 3b.
4. The applicability of method limitations to future JPSS-2/3 missions is not addressed. Additionally, since there are differences between VIIRS sensors, it is advisable to clarify whether this coefficient can be directly applied to NOAA-21 and JPSS-2 or if it requires recalibration for each satellite.
5. The discussion on "sources of error" is overly broad. It is recommended to break down the errors into the following categories: spectral differences, differences in observation times, residuals from spatial matching, discrepancies in cloud masking, and orbital drift of NOAA-19. Methods such as triple-collocation or variance decomposition can be applied to determine the contribution of each factor.
6. Coefficient stability against NOAA-19 orbit drift is missing. Plot yearly boxplots of regression slopes for 2013-2019.

Minor Comments

1. Figure captions end without periods; Table 2 has formatting misalignment.
2. Add units (dimensionless or K) to regression coefficients in Table 2.
3. Combine redundant Sections 4.1 and 5.1 to improve flow.

Author Response

Please see the uploaded WORD document for the responses to reviewer's comments. Thank you! 

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have declined to address the following aspects from my review:

Third, there is very little context provided for these climate data records. What other data sets, especially from NASA and ESA, are available and how do they compare with yours? This addresses the importance of these products.

Fourth, some contemporary validation of the geophysical variables is needed: NASA MODIS for skin temperature and surface albedo and NASA CERES for radiation fluxes at the surface and TOA.

The following reference was quoted in the reply to my comment about sea ice but not included in the manuscript.

Xuanji Wang, Yinghui Liu, Jeffrey R. Key, and Richard Dworak, 2022, A New Perspective on Four Decades of Changes in Arctic Sea Ice from Satellite Observations, Remote Sensing, 2022,14(8), 1846, https://doi.org/10.3390/rs14081846.

From my perspective, it is important to not only describe the methodology but to convince the reader that the results are of high quality.  Regarding point 3, why not discuss what other data sets are available and why yours are the ones to use? Regarding point 4, at least provide contemporary references. Your first version of this manuscript had a dated feel to it with not many recent publications addressing validation. Along these lines why not quote the above sea ice manuscript?

Author Response

Please see the attached WORD document for my responses. Thank you! 

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Lines 149–151: The statement that “the three angular differences between AVHRR and VIIRS are all within the range of +0.1 to –0.1 degrees” still appears unrealistic. Is this criterion applied to observations from the same ground locations on the same date? Even with careful collocation, achieving such precision is rare unless the analysis is limited to nadir-only observations. Are the authors certain that the angular unit is degrees rather than radians? It would be helpful if the authors could provide a few examples (including date, ground location, and the three angular values) from NOAA-19 and NOAA-20 to illustrate the typical matching accuracy.

In the previous review, the reviewer suggested: “Figures 6–9: Consider adding a third column of panels showing the bias (VPP – APP) for clearer visualization.” The authors’ response referred to Table 4. However, Table 4 only presents overall averaged values and cannot illustrate the spatial distribution of the bias. If the authors believe that adding a third column would make the figures too crowded, an alternative would be to include one or two additional figures showing the spatial pattern of the bias. This would be particularly useful for Figure 8, where the albedo bias appears more pronounced (up to 0.1, or 10%?). Such spatially variable bias cannot be adequately represented in Table 4, which summarizes values averaged over space and time (presumably annual means).

Author Response

Please see the attached WORD document for my responses. Thank you! 

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

n\a

Author Response

Please see the attached WORD document for my responses. Thank you! 

Author Response File: Author Response.pdf