the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 17 Oct 2023
Permafrost degradation of peatlands in northern Sweden
Samuel Valman
Matthias Siewert
Doreen Boyd
Martha Ledger
David Gee
Betsabe de la Barreda-Bautista
Andrew Sowter
Sofie Sjogersten
Abstract. Climate heating is degrading palsa peatlands across the circumpolar permafrost region. Permafrost degradation may lead to ecosystem collapse and potentially strong climate feedbacks, as this ecosystem is an important carbon store and can transition to being a strong methane emitter. Landscape level measurement of permafrost degradation is needed to monitor this impact of warming. Surface subsidence is a useful metric of change and can be monitored using InSAR satellite technology. We combined InSAR data, processed using the ASPIS algorithm to monitor ground motion between 2017 and 2021, with optical and LiDAR data to investigate the rate of subsidence across palsa peatlands in northern Sweden. We show that 55 % of the area of Sweden’s eight largest palsa peatlands is currently subsiding, which can be attributed to these permafrost landforms and their degradation. The most rapid degradation occurring in the largest palsa complexes in the most northern part of the region of study, also corresponding to the areas with the highest % palsa cover within the overall mapped wetland area. Further, higher degradation rates were found in areas where winter precipitation has increased substantially. The roughness index calculated from a LiDAR-derived DEM, used as a proxy for degradation, increases alongside subsidence rates and may be used as a complementary proxy for palsa degradation. We show that combining datasets captured using remote sensing enables regional-scale estimation of ongoing permafrost degradation, an important step to-wards estimating the future impact of climate change on permafrost-dependent ecosystems.
- Preprint
(1634 KB) - Metadata XML
- BibTeX
- EndNote
Samuel Valman et al.
Status: final response (author comments only)
-
RC1: 'Comment on tc-2023-138', Anonymous Referee #1, 12 Nov 2023
The authors describe ground movements at eight palsa peatlands in northern Sweden derived from satellite InSAR data 2017–2021 and relate detected changes to climatic conditions. The paper addresses an important issue regarding monitoring of subsidence of palsas and has potential to be an important contribution towards advancement of satellite-based methods in (sporadic) permafrost research. In addition to addressing the issues raised below, I believe the paper could substantially improve with thorough proofreading and grammar check.
- Consider adding InSAR in the title, as it is the only dataset on which you base your investigation of permafrost degradation.
- L 40: landscapes -> land area or exposed land surface
- L 41–42: “…of which around 1.7 million km2 is permafrost substantially overlapping with the circumpolar permafrost region” -> perhaps just: “of which around 1.7 million km2 is affected by permafrost” or “.. is within the permafrost region”
- L 43: I suggest specifying that this applies mainly to peatlands in discontinuous and sporadic permafrost zones.
- L 44–46: Please rewrite this part. For instance, palsa peatlands in Sweden as an example in a separate sentence.
- L 50: Unsuitable permafrost conditions where? Everywhere? Fennoscandia? Please specify.
- L 58: “…frozen permafrost core of palsas…” -> e.g. “perennially frozen core of palsas” or just “frozen core of palsas”
- L 68: Abrupt thaw is not limited to peat plateaux but can also occur at the edges of palsas. Also, earlier on L 44 you use ‘plateaus’, not ‘plateaux’. Either form is fine but be consistent.
- L 118: insert ‘data’ after ‘LiDAR’
- L 126–127: Your use of ‘palsa’ and ‘palsa peatland complex’ here is confusing. Could this be rewritten as e.g., “The eight largest palsa peatlands in the region range between 50 …” ? Also, please clarify what do you mean by ‘palsa peatland’, ‘palsa peatland complex’ and ‘palsa complex’ in your paper. If they are interchangeable, then use only one.
- Figure 1: Please export this figure in better resolution, e.g. as PDF. What does ‘ungrouped palsa’ mean in the legend? My guess is that it refers to the scattered black dots representing palsas, which were not part of this study. Could they be indicated with some different colour/symbol to differentiate from the peatlands, which you do focus on?
- L 140–149: I think this could be rewritten to be more concise (e.g. avoid repetition of the fact that the eight selected areas are the largest concentrations of palsas in Sweden) and easier to follow. The last sentence of this paragraph is particularly troublesome. Also, please clarify, did you create the 250 m buffers or was this done already by Backe (2014).
- L: 145: “raised palsa plateaux” -> e.g. “raised palsas and/or peat plateaux”
- Table 1: ‘EU Nature 2000’ -> ‘EU Natura 2000’. I also suggest adding a column showing the respective weather stations, from which the meteorological data were used.
- Please clarify, which areas were included in what analyses. I understood that you only focused on the eight palsa peatland complexes in your InSAR ground motion detection and other analyses. However, in some cases (e.g. L 248–251) it sounds that your InSAR analysis covered larger areas than e.g. roughness index.
- Overall, I would like to see more detailed description of data processing. At least add a flowchart indicating different procedures, for which steps 100 m spatial resolution was used etc. Keep in mind that not all readers of the Cryosphere are experts in InSAR. I think with such flowchart you can also easily address my previous comment.
- What is the vertical accuracy of the InSAR data you use? How much of the detected ground motion is within the margins of error? Perhaps this is something that could also be addressed in the discussion.
- L 179: What is SNAPHU algorithm and how did you modify it?
- L 223: “The Naimakka station did not provide” and “the Saarikoski station did not provide” sounds as if the data were there, but the stations did not want to share it, which I doubt. Please rephrase.
- L 244–245: Which time series were incomplete and how short time series were used in those cases?
- I suggest dividing the results section into two or three subsections to make it easier for a reader to navigate through the paper. Same applies to the description of datasets and analyses.
- L 248–251: Does this mean that you derived surface motion from other palsas than the eight focus areas as well? Or where did this 7.7 mm yr-1 uplift occur?
- Figure 2: Could you provide larger versions of these plots alongside respective aerial orthophotos and (estimated) edges of palsas? If not in the main text, at least as supplementary material. In the current form it is impossible for the reader to evaluate, which parts of the plotted areas are actual palsas, and how the InSAR ground motion values are distributed in relation to the permafrost areas.
- L 271: “just under half” is very vague. Also, this seems to contradict with the next sentence, where you say that ca. 55 % of the total palsa complexes’ area was subsiding. Please clarify.
- L 277–278: Do you mean “subsidence rates > 3.5 mm yr-1” or between 3.5 mm and some other value?
- Figure 3: Please change the title of the x-axis to “ground motion”, as you have correctly called it in Figure 2. As you describe in the caption, only the shaded areas represent the subsidence.
- L 306–308 and Figures 4 & 5: Here again, it would be good, if you could add aerial orthophotos as well as estimated palsa edges so that it is easy for the reader to compare different datasets. Regarding the figures, I advise changing the colour assigned to the roughness index or to the lowest elevation values. Same (black) colour is confusing particularly in Figures 4c & 5c. I also suggest removing the shading of the rectangles indicating the spatial extents of the close-ups in Figures 4a & 5a. Some coordinates would also be good. It would have been interesting to see, where in the landscape of Tavvavuoma those positive ground motion values are located, but unfortunately I could not locate, which part of the Tavvavuoma is illustrated in Figure 5.
- Additional comment regarding Figure 4a: It seems that InSAR ground motion data depicts very nicely the subsidence of higher palsas ca. 800-1000 m north-west from the area, which you chose to show in the close up.
- L 319: What do you mean by “many less bands of subsidence and potential palsa”?
- Figure 6. Here again, please change “subsidence” in y-axis title to “ground motion”.
- L 331: What do you mean by “all palsa area” here?
- What is your reasoning to include data from Katterjåkk station, if there are no palsa peatlands and it is not the closest station to any of the eight study sites? Since you do not refer to it anywhere in the discussion, I suggest leaving it out. For example, in the figures 7b & 7c, the values from Katterjåkk draw the attention, while perhaps the records form other stations are more relevant for the study sites that you focus on.
- L 357: This analysis was not mentioned in the data analysis section. Please describe there, which test you used, as well as the level of significance, which you use in all of the statistical analyses.
- L 362: I think it would still be useful to see, how the air temperature, snow depth, and summer precipitation have varied over this long time period. If not in the main text, perhaps in the supplementary materials.
- L 373–378: Unnecessarily long sentence.
- L 379–380: I do agree that permafrost degradation likely affects most of the palsa peatlands of However, to say that the degradation is “significant” and “…in all palsa peatland areas across northern Fennoscandia” based only on your study, in which some of the sites showed mostly uplift rather than subsidence, and one study from Norway, thus completely ignoring Finland and Kola Peninsula, overly simplifies the variety of dynamics in palsa peatlands in this region. I strongly advise to rewrite this statement.
- L 381: subsidence -> ground motion. Same on L 383.
- L 384: Wasn’t the gradient of increasing degradation/subsidence from south to north? Please check and correct if needed.
- You connect the InSAR-based ground motion results solely to the climatic variables. However, you do not provide any close-ups to the climatic conditions at the closest weather stations during the period covered by InSAR data (2017–2021). Were summer and winter air temperatures, summer and winter precipitation, and snow depths all similar at the northern and southern weather stations during this period?
- I think you are missing the possible effect of palsas’ shape and geometry on their degradation rates. This has been pointed out by, e.g. Borge et al. (2017) and Mamet et al. (2017), which you cite, as well as more recently by Wang et al. (2023; https://iopscience.iop.org/article/10.1088/1748-9326/ad0138). Your DEM and roughness index data in addition to aerial orthophotos seem sufficient to assess the differences in palsas’ geometries between the sites. See also comment #28. I do not expect you to quantify all palsa geometries for this case, but at least some visual assessment would be beneficial for the discussion.
- L 403: Here you use ‘Scandinavia’, as well as later in the discussion. Please clarify, whether you aim to discuss palsa peatland changes only in Scandinavia or in the whole northern Fennoscandia. These terms are not interchangeable.
- L 419–421: I suggest adding the importance of thin snow cover, as is also mentioned by Vorren (2017), which you cite here. I also want to note that the palsas investigated by Vorren are located in climatically very different area.
- L 424–429: I think the high number of pixels with positive ground motion values in your study areas deserves a bit more attention. As I mentioned earlier, showing aerial orthophotos and estimated palsa edges with your InSAR ground motion values could help evaluation of what could be the reason(s) behind the “uplift”. Are positive values mostly in the fen areas? Or areas with tall vegetation on mineral ground?
- L 439–441: I suggest editing this to, e.g.: “ However, our large-scale assessment of permafrost subsidence provides a baseline for future assessment of subsidence in northern Sweden”. In the current form the sentence is very difficult to read. And what “would be advantageous”? Your assessment for the future field monitoring or the field monitoring itself? I believe both.
- L 518: Add line change before Ballantyne
Citation: https://doi.org/10.5194/tc-2023-138-RC1 -
RC2: 'Comment on tc-2023-138', Anonymous Referee #2, 14 Nov 2023
The paper “Permafrost degradation of peatlands in northern Sweden“ demonstrates the use of the ASPIS-InSAR technique to identify subsidence over permafrost peatlands in Northern Sweden which covers a relatively large region. They have used primarily Sentinel-1 satellite data from 2017-2021. This technique can be applied over the pan-Arctic as well.
Comments on substance
In the Introduction it would be interesting to mention what the current annual degradation rates are, if that information is available. Probably in the paragraph that ends around Line 95.
The Introduction could use more info on what InSAR for subsidence studies have found and the accuracy they obtained.
Line 130 – why not use Naimakka instead of Karesuando weather statistics? Most of the palsas that earlier existed in Karesuando have thawed and disappeared, so it doesn’t represent the climate as well as Naimakka.
Line 142 – I would call it a grid cell or raster cell instead of a pixel, since a pixel refers to a picture element, usually from an image, while with a processed raster data set, it should be a grid cell or raster cell.
Under Section 2.2 Datasets, please put the processed cell size of the data, which should be 5 m in range and 20 m by azimuth if I’m not mistaken. Is this the resolution of the input data you worked with?
Line 166 – What program or code was used to run APSIS? Please describe more about APSIS, for example, I believe it needs to resample the result to be larger than the input pixel size, so I wonder how it is that your output is 20 x 20 m. This can help us interpret the result better.
Line 184 – what kind of accuracy? This sentence is a bit vague. Of identifying that permafrost degradation has occurred? Or accuracy of measured vertical subsidence?
Line 189 – Are you sure you used panchromatic orthophotos? The ones available from 2016 are either RGB or False-color IR, so I would think you used one or the other of these, not panchromatic.
Line 192 – The Swedish national lidar scanning consists of several different dates of data from ca 2009-2015, so it is likely not from 2016. You need to download the shapefile to determine the dates of the lidar data, which were acquired in 25 x 50 km blocks. See for example Nilsson et al, (https://doi.org/10.1016/j.rse.2016.10.022) for a description of the national lidar data, or else go directly to Lantmäteriets website for a description, as well as the shapefile.
Line 201 – For the Discussion: How do you think taking the mean value over a 100 m x 100 m area may have affected your results? I am wondering why you didn’t maintain the 20 x 20 m resolution from ASPIS and simply use the values from the grid cells that were showing subsidence. You could even determine the minimum and maximum subsidence per 100 x 100 m grid cell using the data from the 20 x 20 m cells. By taking a mean value, I wonder if you might be smoothing out results where the palsas are not filling the 100 m x 100 m grid cell. If I misunderstand your process, it needs to be more clearly described. When you resample, you create a new value. Perhaps you used the 20 x 20 m product for some processes, and 100 x 100 m in others? It needs to be clear what you used and when – it can have a large effect on your results.
Line 203 says that the frequency numbers are derived from the 100 x 100 m resampled raster. But Fig 3 showing the frequency says it is based on the 20 x 20 m grid cells.
Line 233 – you say you used modelled permafrost probability distribution and make a reference to Obu 2018, however I didn’t find it clear that you were using their data set. Please give a little more information – also, did you use the 1km grid cell data set and in that case what was your resampling method to 100 m cell size?
Line 248 – Is this first number across all of the palsas, or all of the grid cells containing a certain percentage of palsas (in this case, what is the %?) or all of the grid cells within a certain area? Also, is this based on the 100 x 100 m grid cells that you have resampled? Because that might explain a lot when you are looking at areas with small isolated palsas, or even narrow ridge palsas.
Fig 6 – It seems the roughness index gave a higher value at the edge of the palsa, so if the roughness is indicating palsa edges only (and not the body of the palsa where there isn’t a high roughness value), then you wouldn’t expect to see the best correspondence with subsidence. It would be a bit noisy, as it is here. The result would be clearer if you had a clear geospatial layer delineating palsa from fen, perhaps by segmenting the palsas (based on roughness and elevation). Then this relationship would look better. That should be in the Discussion. I think the roughness index helps you identify some of the edges of palsas, but doesn’t go as far as you need it to to give you a good map of palsa locations.
Table 3 - Ok, so I have these data, and I looked up the Min daily temperature for Naimakka and found that it was -41.20 on 2021-12-06 (and the single Minimum temperature occurred on this day at -43.40). This doesn’t match what you have in Table 3, which I am reading as being the lowest daily (averaged hourly) temperature that occurred between 01-01-2000 to 12-31-2021. Is there a misunderstanding of the terms or years given in the Table text? I did not check any other numbers.
In the Discussion I think you should discuss the error levels of ASPIS InSAR as found by other studies. You weren’t able to measure elevations and subsidence in this study to compare to what you found with InSAR, but other studies in other ecosystems have looked at the accuracy of InSAR and assessed the potential error. How reliable are your estimates, and how do you judge that? For example, Line 435 and 436 states “However, the high precision of the change in vertical position means that InSAR is an important tool to employ to detect the initial stages of large-scale permafrost degradation.” – I think you need a reference here since you don’t yourself do an accuracy assessment.
Line 424 – In this part of the Discussion, do you have any idea whether the dates of the SAR images in the stack can influence that there may be uplift? Or do all dates weigh in evenly in the InSAR stack? If you use images from April, May and June, and it has been a cold spring, for example?
Grammatical comments
Line 24 – something to fix with the grammar of the sentence
Line 30 – towards
Line 38 – remove comma, and check through your use of commas
Line 111 – What is meant by this sentence? It’s unclear to me. Pertinent to this is that any InSAR-detected changes can be associated with known and delineated targets in the wider landscape.
Line 145 and 149 – consistency with the term plateaux or plateau
Line 179 – reference the SNAPHU algorithm (did you use the SNAP plugin?).
Line 189 - It’s a mouthful, but Lantmäteriet is officially called “The Swedish Mapping, Cadastral and Land Registration Authority” in English.
Line 198 – I would re-write as “… Kiruna, Karesuando, Saarikoski and Naimakka.” Since the last two are two separate weather stations.
Line 237 – resampleto … resampled to
Line 248 – make sure the – sign before 9.9 stays with the number. It took me a couple of reads before I saw it.
Line 249, 251 – palsa should be palsas (if plural)
Figure 2 – the figure text should mention that this is based on data for the years 2017-22. Also, the images are so small, it is hard to see the values of the grid cells. Can you make them a bit larger?
Line 313 – “… subsidence proximal to …”
Line 334 – meteorological (spelling)
Line 338 – Naimakka (spelling)
Fig 7 text - Is that daily snow depth measurements?
Line 359 – “… the northern-most of the three weather stations with long-term records …” Since Naimakka is further north, good to make the distinction that Karesuando is furthest north of the three….
Line 362 – add “for any of the sites” to make this clear. Also, what temperature? Mean annual temperature? Specify.
Line 370 – remove the comma “on-going subsidence”
Line 441 – something missing from a sentence.
Line 446 – Why Arctic DEM and not Copernicus DEM (see for example Karlson et al., 2021. https://www.diva-portal.org/smash/get/diva2:1617869/FULLTEXT01.pdf Or Similar.
Line 495 – replace the ??
Line 685 – terrain using
Line 699 – thermal not theral
Citation: https://doi.org/10.5194/tc-2023-138-RC2
Samuel Valman et al.
Samuel Valman et al.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 211 | 46 | 5 | 262 | 2 | 3 |
- HTML: 211
- PDF: 46
- XML: 5
- Total: 262
- BibTeX: 2
- EndNote: 3
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1