Next Article in Journal
Fast and Effective Techniques for LWIR Radiative Transfer Modeling: A Dimension-Reduction Approach
Next Article in Special Issue
Assessing Spatiotemporal Variations of Landsat Land Surface Temperature and Multispectral Indices in the Arctic Mackenzie Delta Region between 1985 and 2018
Previous Article in Journal
Detection of Soil Pipes Using Ground Penetrating Radar
Open AccessArticle
Peer-Review Record

Seasonal Progression of Ground Displacement Identified with Satellite Radar Interferometry and the Impact of Unusually Warm Conditions on Permafrost at the Yamal Peninsula in 2016

Remote Sens. 2019, 11(16), 1865; https://doi.org/10.3390/rs11161865
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Reviewer 3: Anonymous
Reviewer 4: Anonymous
Remote Sens. 2019, 11(16), 1865; https://doi.org/10.3390/rs11161865
Received: 24 June 2019 / Revised: 1 August 2019 / Accepted: 6 August 2019 / Published: 9 August 2019
(This article belongs to the Special Issue Remote Sensing of Permafrost Environment Dynamics)

Round 1

Reviewer 1 Report

This paper introduced the usage of InSAR to measure seasonal displacement at the Yamal peninsula. The paper is fairly well written with clear logic and structure. The results generally support the conclusions.

I have one concern: As shown in Figure, the land cover of the studied area contains various types and arbitrary shapes. Unwrapping for such area should be very tricky. I would ask the authors to provide more datails of how they implement the unwrapping procedure. Unless correct unwrapping procedure is performed, the provided InSAR deformation measurements are convincing. 

Author Response

Reviewer comment: I have one concern: As shown in Figure, the land cover of the studied area contains various types and arbitrary shapes. Unwrapping for such area should be very tricky. I would ask the authors to provide more datails of how they implement the unwrapping procedure. Unless correct unwrapping procedure is performed, the provided InSAR deformation measurements are convincing.

Reply: As stated in the manuscript, “Our InSAR processing sequence … includes … phase unwrapping using a minimum cost-flow algorithm [Costantini, M. A novel phase unwrapping method based on network programming. IEEE Transactions on Geoscience and Remote Sensing 1998, 36, 813–821. doi:10.1109/36.673674].” We also remarked that “phase unwrapping errors could be not completely avoided and were observed in particular at the scale of a few tens of meters. They are also depending on the acquisition time interval and season (i.e. better for very short time intervals and at the end of the season when displacements are smaller).” One of the consequences we already indicated in the manuscript is that in many cases lower InSAR derived displacements than in situ observations are observed, which can be also caused by “under-sampling of the SAR data in relationship to large local variations of the rates of movements, which can cause phase unwrapping errors.”

In order to provide more information on the implementation of the phase unwrapping procedure we now complemented at line 253 the sentences “phase unwrapping using a minimum cost-flow algorithm [50]” with “implemented in the Gamma Software 

[50] [Werner C., U. Wegmüller, T. Strozzi, and A. Wiesmann, Processing strategies for phase unwrapping for INSAR applications, Proceedings of the European Conference on Synthetic Aperture Radar EUSAR 2002, Cologne, Germany, 4-6 June 2002].”

 

Reviewer 2 Report

The manuscript titled “Seasonal progression of ground displacement identified with satellite radar interferometry and the impact of unusually warm conditions on permafrost at the Yamal peninsula in 2016” present a comprehensive analysis of the seasonal evolution of subsidence over central Yamal, Russia using C- and L-band InSAR observations, in-situ subsidence records, active layer thickness measurements, borehole temperature records and land-cover classification maps. Overall, I think this paper is in good structure and writing, but the statement on result is limited, so there are spaces for improvement. In the following I provide comments and suggestions for this manuscript.

 

1) The authors defined too much abbreviations that make it’s hard for the readers to understand the article. Using too many abbreviations will make your manuscripts confusing to the reader, forcing the readers keep on thinking about what the abbreviation is. Please utilize the common-sense abbreviations, and define your own abbreviations as few as possible in a paper, e.g., degree-days of thawing is usually written as ‘thaw degree days (TDD) instead of DDT’.

 

2) Figure 1:

The authors mentioned this figure in the main text after Figure 2 and 3. The figures in a paper would be serving as the most visual information for interpreting your results and discussion. You’d better link your text with each figure and mine useful information from them.

I would suggest the authors enlarge (b), and reduce the size of (a). Maybe just have (a) as an inset of (b). 

White dots represent three Vaskiny Dachy research stations in Figure 1b, but it seems that there is another white dot at the left-bottom corner. Is this also a Vaskiny Dachy research station?

 

3) My main concerned problem of this manuscript is the observed InSAR subsidence result. I could not be convinced by the results listed in Figure 3 and 4. InSAR presents a subsidence map with resolution of about 13 m (as you described in the text), while the CALM grid covers an area of about 75*75m2. There are about 5*5 pixels InSAR data in this area, but how many InSAR measurements have been actually detected over this area? I fully recommend the authors provide an InSAR map to show the spatial distribution of ground displacement. So we could clearly see where are deforming and which annual subsidence measurement point (in Figure 3) should the InSAR measurement pixels corresponding to.

             

Figure 3: The authors should show all the measurement points including the new sites after 2018 by using different color for two districts of measurements.

 

Figure 4: I am a little bit confused of this figure. Did you use the seven sites in Figure 3 (red dots) to conduct the comparison between CALM grid and InSAR? If so, how do you separate InSAR measurements of point 2, 3, 4 and 7? They are very close to each other, even maybe in the same InSAR pixel. In addition, there is a high risk to use only one InSAR measurement pixel to compare its value with the ground observations. We usually select few InSAR pixels around the other ground stations, average the InSAR values, and then do comparisons.

 

In addition, you stated at lines 302-305 “For 304 comparison with satellite observations, all pixel values overlapping with the CALM grid (in total 22 305 pixels) have been also averaged.”

So how do you have that many points on Figure 4?

 

4) The authors mentioned generating vertical deformation method in lines 279-282, but without applying to your analysis. All the figures are suggested to be used the line of sight InSAR measurements (e.g., Figures 4, 5, 6, and 7). It’s necessary to compare the vertical displacement between different InSAR sensors with different observing directions.                   

 

5) Given InSAR is measuring the relative deformation, where is the ground reference pixel for your processing and how does it been selected?  

 

6) Did you pay special attentions to the height errors during the PSI processing?

                                                                                                                                                 

7) P. 27, Lines 3-8, 43-52: Could you please state more about how do you conduct the ” offset correction”, which have been mentioned in, e.g., lines 437, 460, Table 3 caption.

8) Line 266: ionospheric or tropospheric disturbances?

 

Minors:

9) Line 35: [9, also includes the impact of precipitation].

I would suggest you modify it like the following: “[9] (also includes the impact of precipitation).”

There's plenty more like this, please modify.

10) Lines 69-70: So far only few studies are available for comparisons between X and C band [4, 19].

I would suggest you modify these kinds of description like the following: “So far only few studies are available for comparisons between X and C band [e.g., 4, 19].”

We usually write something based on our knowledge, but Knowledge is limited of any one of us.

11) Line 137: What is GTN-P short for?

12) Line 185: (see Figure 3)

 


Author Response

1) The authors defined too much abbreviations that make it’s hard for the readers to understand the article. Using too many abbreviations will make your manuscripts confusing to the reader, forcing the readers keep on thinking about what the abbreviation is. Please utilize the common-sense abbreviations, and define your own abbreviations as few as possible in a paper, e.g., degree-days of thawing is usually written as ‘thaw degree days (TDD) instead of DDT’.

Reply: We have revised the manuscript for abbreviations and spelled out where missing.

In permafrost research the convention is to use DDT (degree-days of thawing). We therefore prefer to use this abbreviation, also in line with in situ studies on permafrost subsidence (e.g. Streletsky et al. 2016). We have extended the paragraph where we first mention DDT to stress its importance for permafrost research. In addition we revised the manuscript for consistent use of naming where the abbreviation is not used (degree-days of thawing).

2) Figure 1:

The authors mentioned this figure in the main text after Figure 2 and 3. The figures in a paper would be serving as the most visual information for interpreting your results and discussion. You’d better link your text with each figure and mine useful information from them.

Reply: adjusted

I would suggest the authors enlarge (b), and reduce the size of (a). Maybe just have (a) as an inset of (b). 

Reply: adjusted

White dots represent three Vaskiny Dachy research stations in Figure 1b, but it seems that there is another white dot at the left-bottom corner. Is this also a Vaskiny Dachy research station?

Reply: the bottom left corner white spot corresponds to no data in the landcover classification. This comes from the Sentinel-2 cloud mask. We have now increased the map, so it is now visible that its not a circle with a black outline. In addition we changed the colour to grey. We have added also a comment in the caption.

 

3) My main concerned problem of this manuscript is the observed InSAR subsidence result. I could not be convinced by the results listed in Figure 3 and 4. InSAR presents a subsidence map with resolution of about 13 m (as you described in the text), while the CALM grid covers an area of about 75*75m2. There are about 5*5 pixels InSAR data in this area, but how many InSAR measurements have been actually detected over this area? I fully recommend the authors provide an InSAR map to show the spatial distribution of ground displacement. So we could clearly see where are deforming and which annual subsidence measurement point (in Figure 3) should the InSAR measurement pixels corresponding to.

Reply: We have now added an example for 2016 (end of season) including the in situ values.

Figure 3: The authors should show all the measurement points including the new sites after 2018 by using different colour for two districts of measurements.

Reply: we have now revised Figure 3 accordingly

Figure 4: I am a little bit confused of this figure. Did you use the seven sites in Figure 3 (red dots) to conduct the comparison between CALM grid and InSAR? If so, how do you separate InSAR measurements of point 2, 3, 4 and 7? They are very close to each other, even maybe in the same InSAR pixel. In addition, there is a high risk to use only one InSAR measurement pixel to compare its value with the ground observations. We usually select few InSAR pixels around the other ground stations, average the InSAR values, and then do comparisons.

Reply: This issue is the reason why we average over the entire CALM grid for the main result of the in situ comparison. The results are listed in Table 3 of the original manuscript (line of sight as well as scaled; together with the average for the in situ data). The subsidence on the CALM grid is not uniform due to the heterogeneity of landcover and organic layer. Therefore, we decided to provide both, the averaged and none-averaged values. Figure 3 is meant to give insight into the heterogeneity and that also variations over the season are in the same order of magnitude. We have now restructured the first part of the results section to clarify this (new order: first the averages, then the single measurements).

The values 2 and 3 are in one pixel and 4 and 7 are in one pixel. One of the reasons that we kept all values was to represent the heterogeneity within an InSAR pixel. A second reason is that in several years, in situ was only available at one of the sites each. Point 3 only had a measurement in 2018. And point 7 had no measurement in 2018. We have now included this information in the section describing the dataset.

In addition, you stated at lines 302-305 “For 304 comparison with satellite observations, all pixel values overlapping with the CALM grid (in total 22 305 pixels) have been also averaged.”

So how do you have that many points on Figure 4?

Reply: The comment on averaging refers to the results of Table 3 which lists one average per year. We have now restructured and extended the first part of the results section to clarify this (first the averages, then the single measurements).

4) The authors mentioned generating vertical deformation method in lines 279-282, but without applying to your analysis. All the figures are suggested to be used the line of sight InSAR measurements (e.g., Figures 4, 5, 6, and 7). It’s necessary to compare the vertical displacement between different InSAR sensors with different observing directions.                   

Reply: The correction for vertical deformation has been applied in Table 3 and well as Figure 4 (left) with regard to the comparison with the in-situ data over the CALM grid. We referred to that as ‘scaled’. We have now added some text to the captions for clarification and also clarify this in the methods section. Added: We refer to the results as 'scaled' in the following.

We now provide also the other figures with displacement in vertical direction.  

 

5) Given InSAR is measuring the relative deformation, where is the ground reference pixel for your processing and how does it been selected?  

Reply: we have now added: The InSAR reference point was selected on an airstrip at 70.317 68.324

6) Did you pay special attentions to the height errors during the PSI processing?

Reply: Given the good quality of the TanDEM-X 30 m considered for the removal of the topographic-related phase, we did not estimate an interferometric height correction in the short-baseline InSAR inversion.                                                                                                                                                 

7) P. 27, Lines 3-8, 43-52: Could you please state more about how do you conduct the ” offset correction”, which have been mentioned in, e.g., lines 437, 460, Table 3 caption.

Reply: The offset correction is described in the first subsection of the methods section. We have now extended the description of the procedure and we have also added the following sentence for clarification: ‘This modification is referred to as ‘offset correction’ in the following.’

8) Line 266: ionospheric or tropospheric disturbances?

Reply: Do you refer to line 256? Then it is indeed ionospheric disturbances.

Minors:

9) Line 35: [9, also includes the impact of precipitation].

I would suggest you modify it like the following: “[9] (also includes the impact of precipitation).”

There's plenty more like this, please modify.

revised

10) Lines 69-70: So far only few studies are available for comparisons between X and C band [4, 19].

I would suggest you modify these kinds of description like the following: “So far only few studies are available for comparisons between X and C band [e.g., 4, 19].”

We usually write something based on our knowledge, but Knowledge is limited of any one of us.

Changed accordingly

11) Line 137: What is GTN-P short for?

Global Terrestrial Network on Permafrost, added

12) Line 185: (see Figure 3)

Changed accordingly

Reviewer 3 Report

This paper presents a case study of the displacements monitoring permafrost area over central Yamal, Russia with COSMO Skymed and Sentinel-1 datasets. The authors specially emphasize the impact of unusually warm conditions in 2016. The study fit the topic of the journal. This paper can be published while following questions are clarified. 
1.     Line 40, ‘……valuable data in in this context’.2.     As the authors mentioned in this paper the movements of the permafrost are mainly in the vertical direction. The reviewer think it’s better to project all the LOS displacement to the vertical direction. Thus, the comparison between the CSK and Sentinel-1 results might be more rigorous in Fig.8. Moreover, the measure displacement from different years from CSK or Sentinel-1 are also comparable in Fig.9 and 10.3.     Fig.2, in my opinion, I can recognize the unusually warm conditions in 2016. The authors have to find a better way to show the reader the change of temperature and the distribution of the data.


Author Response

1.     Line 40, ‘……valuable data in in this context’.

changed

2.     As the authors mentioned in this paper the movements of the permafrost are mainly in the vertical direction. The reviewer think it’s better to project all the LOS displacement to the vertical direction. Thus, the comparison between the CSK and Sentinel1 results might be more rigorous in Fig.8. Moreover, the measure displacement from different years from CSK or Sentinel1 are also comparable in Fig.9 and 10.3.   

Reply: we now provide figures with displacement in vertical direction.  

Fig.2, in my opinion, I can recognize the unusually warm conditions in 2016. The authors have to find a better way to show the reader the change of temperature and the distribution of the data.

Reply: We have now added a timeseries of cumulative degree days of thawing to this figure and used colours to distinguish the years.

Reviewer 4 Report

Please consider the attached file

Comments for author File: Comments.pdf

Author Response

I suggest the Authors to enumerate the figures in the progression they are called.

Revised

 

Line 129 : I would really appreciate if the Authors could explain how DDT is evaluated.

 

Reply: DDT is a different representation/ a derivative of measured air temperature. It serves as an index for the warmth of the unfrozen period. Figure 2 now also shows the development of DDT to demonstrate the differences between the years better. We have also modified the text for better understanding. See new lines 131 ff

 

Line 181 : I think can be very helpful for the Reader if the Authors would explain in Fig. 2 the difference between the light-green line and the black line.

 

Reply: we have now revised the figure as well as caption. A different colour is now used for the temperature records for each year.

 

Lines 127-129: “Summer 2016 has been distinct regarding air temperature [22]. Mid July (day of year 190-200) air temperature has been above average what was followed by a second warm period in August (Figure 2).” At least to me, it is not really immediately clear to read that from the data. I suggest the Authors the possibility to highlight this aspect with a zoom on the interested time intervals.

 

Reply: Instead of the zoom, we have tried to improve the readability in general. We have revised the figure, the caption, and introduced subfigure numbering (a, b, …) for reference in the text. We have now also included the day of year for start of August in the text (start of second warm period) to aid the identification of this period in the figure.

 

Lines 271-275: In Table 1 are enumerated the Sentinel-1 acquisition dates. There isn’t the

information about the pairs of SAR images to obtain the interferograms. Maybe the Authors can explain better how they obtain the comparability of the years.

Reply: Pairs are always the acquisitions following each other within a year. This is described at the beginning of Section 3.1: “Our InSAR processing sequence … includes … the computation of interferograms in series … [and the] computation of summer cumulative displacement maps and time series of movement via short-baseline InSAR” with reference to a previous work (Strozzi et al. 2018).

Decorrelation does not allow the connection between the years. The comparability between the years is achieved by offset correction with DDT. We have now revised this part of the methodology for better understanding.See new lines 281 ff

 

Lines 277-278: I suggest the Authors to explain better the sentence : “A rate of subsidence per thawing degree is derived and applied to the time before the first data pair in 2016.”

 

Reply: We have now extended this part of the methodology to clarify this. See new lines 281 ff

Round 2

Reviewer 2 Report

This is just an advise for your future manuscript. If you make some changes in text or add something new when you reply reviewer's comments, I would suggest you also provide the line numbers in your revised paper. It's difficult for me to find the changes associated with each comment some time.   

Back to TopTop