Review Reports

Round 1

Reviewer 1 Report

Summary

The paper compared surface reflectance of corresponding bands from Landsat 8 and Sentinel-2 at three sites in eastern Siberia, and adjusted data from Sentinel-2 to match those from Landsat-8 using least square regressions. The adjusted data for all corresponding bands show a near perfect alignment from the two satellites, in contrast to no improvement in the global harmonized product. This is potentially useful to obtain a dense time series of clear sky observations from combined Landsat-8 and Senteniel-2 data, and benefit applications in the high latitude where cloud cover often persists.

Overall the paper reads well. Outlined in the comments below are a number of suggestions that will hopefully improve the final version of the paper.

Comments

Given the paper does not include any specific applications of permafrost, I found it somewhat confusing to mention it in the title. I’d suggest the authors modify the title.

L26-27, what does “stronger” mean in  “precipitation regimes changed in the Arctic stronger than the global average”?

L30-31, Considering that about a quarter of the Northern Hemisphere is underlain by permafrost, and even 65% of the land north of 60 degrees

This doesn’t read well, suggest change “and even” to “which is”?

L32, suggest change “Continuation of warming” to “Continued warming”

In Section 2.3.4, the images and site masks are both resampled to 60m resolution, however, in Section 2.3.2, the site masks are created at 30m resolution. I wonder why not just creating the masks at 60m?

L298, Both the cloud and especially cloud shadow masking were not very reliable, neither for the Landsat-8 nor for the Sentinel-2 images [58], [51].

Can you select some reliable clear sky pixels at each site through visual inspection? Then compare the reflectance of these pixels with the remaining pixels to test the impact of cloud and cloud shadow?

Can you explain why clouds affect band 2-4 (low correlations) more than the other bands in Table3 at Batagay? I wonder if heterogeneous terrain is a factor.

Suggest merge Table 3 and 6.

The discussion Section listed a few factors that may have contributed to the differences in the reflectance of two sensors, including poor cloud/cloud shadow detection, mixed pixels, and heterogeneous and complex terrains. In addition, the images have to be downscaled to 60m. These are important elements that potential users should be aware of, and should be mentioned in both the abstract and conclusions.

L496-497, We showed that it is possible to reduce the persistent error

 I don’t think these are errors, just different spectral response.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

Summary

The aims of the paper are to compare and assess the compatibility of spectral characteristics of Landsat-8 and Sentinel-2 data across different landscapes and ecosystems in Arctic-Boreal permafrost regions. Although the methods used are not novel, their application for Eastern Siberia improves significantly the applicability of combined time series of L8 and S2 data sets in this region, where various forms of permafrost disturbance are taking place. Basing the assessment only on the freely available data sets and open source software adds to the overall value of presented work. Below are few comments and suggestions, which can help to improve the paper further.

Broad comments

1.       Overall, the paper is well structured, all objectives stated in the introduction are addressed and conclusions supported by the results. The introduction provides good background for why we need regional regression model for combining Landsat-8 and Sentinel-2 data.

2.       Aside of some long sentences (e.g. L59 – 63), the text is well-written and easy to read.

3.       Figures and tables are clear and easy to understand. Adding the explanations of abbreviations to the captions of some tables would make them more self-explanatory.

4.       The data processing and analysis are well described.

5.       The study sites descriptions lack a bit of consistency. The climate type and mean annual precipitation are given for Batagay and Yakutsk, but not for Lena Delta. The mention of permafrost extension / distribution in Lena Delta could also be good, as it might not be obvious from presence of ice-wedge polygons and thermokarst for readers not familiar with these landforms.

6.       It is not quite clear for how large area the regional (ES) adjustment coefficients are applicable. Is it only the region shown in the Figure 2 or Arctic-Boreal regions of Eurasia and N America in general?

7.       As you mention the long record of the Landsat data, one might get an impression that Landsat-8 and adjusted Sentinel-2 data sets are directly applicable with the long time series of Landsat imagery. Although the differences in spectral bands of Landsat-8 and previous missions (e.g. Flood, 2014: https://doi.org/10.3390/rs6097952) are out of the scope of this paper, it could be good to add that additional spectral adjustments might be needed if one wants to prolong their time series with L8 and S2 data.  

 

Specific comments

8.       L42 – 46: The description of press and pulse disturbances could be clearer.

9.       L73: What do you mean by “full monitoring”?

10.   L79: One ‘addition’ too many

11.   Figure 1: Is the scale correct? The average number of clear pixels seems to be extremely low. 

12.   L223: Not necessary to repeat Landsat 8

13.   L246-247 and table 2: Did you find only one pair of same day images with max. cloud cover of 80 % per study site?

14.   L290 – 291: unnecessary brackets

15.   L342: second ‘NIR’ -> ‘Near-infrared’

16.   L386: ‘East’ -> ‘Eastern’ for the consistency

17.   Figure 5: Missing preposition in the first sentence of the caption.

18.   Table 6: Why the Pearson correlation coefficients are the same after both adjustments?

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

General comments

The article entitled "Comparing spectral characteristics of Landsat-8 and Sentinel-2 same-day data for Arctic-Boreal permafrost regions" presents in a very clear way the methodology used to combine Landsat 8 and Sentinel 2 images to be used in time series. This study has been conducted in a boreal arctic permafrost region, where one of the main drawbacks to the observation of the earth through remote sensing is the high percentage of cloud cover. This situation is repeated in other regions of the planet, which is why this study would be of interest in other places. In addition, three sectors with different land cover have been considered within the study area to analyze their spectral response, which makes the work more rigorous. I believe that the objective of analyzing the possibility of combining both images has been successfully developed. But, in addition, there is another objective linked to the need to analyze in more detail the processes, at different time scales, of degradation of permafrost and its consequences in the landscape.

While the manuscript has not described how many images Landsat 8 and Sentinel 2 have been used for the period 2016-2018. From the temporal resolution of the satellites and the data provided in Figure 1, on the amount of clear pixels in each of the images, it would seem that the number of usable Sentinel 2 images would be much higher to the amount of Landsat images. At this point, it would be crucial to ask whether the available Landsat 8 images would be necessary to include them, according to the objectives of the work, considering the time and effort involved in making the spectral adjustment. Possibly, for the period starting from the placing in orbit of the Sentinel satellites 2 A and B, it is possible to study the changes in the landscape through the images provided by these platforms only. It would be important that the authors consider adding in the article a section on the analysis of the evolution of some landform of the landscape through the combination of both images. In this way, the applicability of the study carried out according to the described objectives would be demonstrated.

Minor comments can be found in the attached file.

Comments for author File: Comments.pdf

Author Response

Please see the attachement

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

I consider that the manuscript entitled ¨Comparing spectral characteristics of Landsat-8 and Sentinel-2 same-day data for Arctic-Boreal permafrost regions¨ has been significantly 
improved and now warrants publication in Remote Sensing. In addition, I consider that the methodological proposal developed in this work will allow to improve the development of more detailed studies in the near future in arctic permafrost regions.

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.