This paper compares remotely sensed seasonal ground surface displacement, measured using the InSAR technique with Sentinel-1 (C-band) data, to soil texture and ground ice profiles obtained from 12 core samples in contrasting landform locations in Adventdalen, Svalbard. While there is a rapid increase in studies dealing with permafrost InSAR (particularly SBAS-type), field validation studies remain limited. This research offers significant progress in explaining InSAR spatial variations using detailed frozen-ground core analysis on a watershed scale. I am glad to see the conclusion drawn from your InSAR and in-situ investigations. Your work goes beyond merely addressing the oversimplification of AL-subsidence models; it highlights a critical oversight—the neglect of classic frost heave studies and the role of excess ice in such models. Although I identified several weaknesses and limitations, I support the publication of this paper following the necessary revisions outlined in the attached file.

Go Iwahana

The submitted article by Wendt et al. primarily compares seasonal ground surface displacement using InSAR remote sensing for 2023 to expected subsidence derived from ice contents from core drilling at 12 sites in Adventdalen, Svalbard. They establish a reasonable correlation between the InSAR-derived subsidence and that expected from the ground ice content in the cores, which includes determinations for pore ice, excess ice, and water drained upon the melt of excess ice. The authors determine that excess ice melt is the key contributor to observed subsidence at many of the sites, with pore ice typically being of secondary importance. The authors further demonstrate that without detailed knowledge of excess ice conditions, active layer thickness cannot be reliably estimated from InSAR in ice-rich terrain.

The strength of this paper lies in the fact that the InSAR subsidence trends can be partly, and fairly strongly, supported by the in situ ground ice determinations from immediately before the remote sensing record, which are commonly not available in similar remote sensing studies. The sampling scheme was well thought out and captured a significant range in ground ice conditions due to the selection of sites from different landforms and substrate conditions. Overall, I recommend this paper for publication however I have many minor comments and a couple more substantive ones that, if addressed, I think will strengthen the manuscript.

Main comments/suggestions:

Ice wedges

The ground resolution is stated as 18.2 x 28.2 m. Some of the sites include ice wedge polygons. In years of very deep thaw, presumably thaw would extent into the tops of ice wedges, and this could materially contribute to subsidence (e.g., https://doi.org/10.1002/ppp.2113). Based on the size of the polygons on Svalbard, I assume some pixels that included a core sample may have also included an ice wedge trough (or more than one). If this is the case, it should be discussed. Could this help explain why the 2023 InSAR derived subsidence is commonly higher than expected subsidence (Figure 6)?

Date of snowmelt

The InSAR record includes scenes following the melt of snowpack at the ADV met station. Did you examine whether snow had melted by this date in ice wedge troughs (or more generally in different topographic settings at different sites)? I presume snow may have persisted later, particularly in deeper troughs, as the snow depths are greater there. I observed this when I was on Svalbard. If this is likely to have occurred also in 2023, you may wish to consider what effect this may have had on the InSAR results at sites with ice wedges, or other settings where deeper snow could have accumulated, and include it in the discussion of limitations.  

Thaw penetration and subsidence rates

The role of pore ice in the nature of the subsidence curves over the summer could be better presented and discussed in relation to established theory and observations. During the thawing season, some of the sites follow a characteristic exponential decay curve in subsidence in layers where excess ice is not present. This generally follows the Stefan equation that described expected progression of active layer thawing. The pattern has been examined in relation to subsidence previously, for example in this paper that you cite in your discussion: https://tc.copernicus.org/articles/14/1437/2020/, and in other applications involving permafrost thaw. So, when the authors indicate that pore ice contributes to subsidence in a more “continuous manner” l. 346, I don’t think this is the best way to describe it. Though it is continuous, the rate is not. Furthermore, indicating that Schuh et al. 2017 confirmed the inverse relationship with ice content, while not inaccurate, is perhaps not the best option to support the relation observed. An inverse relation exists in the absence of excess ice, and governing equations that relate the thaw rate to the square root of time significantly predate the cited study. So, I suggest familiarization with the Stefan equation and the expected exponential decline of thaw progression with time and edits to associated text, and reference to pertinent literature.

Figures

Line 242 indicates that “Due to the dominant contribution of excess ice melt and drainage to the total subsidence…”. However, this isn’t presented or established until Figure 6, so the statement is confusing to the reader because this result hasn’t yet been shown. Figure 6 showing this partition should come earlier. I understand this would entail showing the InSAR max earlier than the InSAR section, which is not ideal, but I think it is perhaps better overall because at least then the readers will be familiar with the expected subsidence in section 3.1.

It would be useful to have a figure showing photos of some of the site types: e.g., an Eolian one in ice wedge polygon fields, and alluvial example, slope/solifluction example, etc.

Minor comments/suggestions:

Line 16 : change “allowing to estimate” to “allowing estimation of”

Line 18: delete “thickness” after “active layer”.

Line 31. I presume here you mean increases in ALT are “largely influenced by the presence of ground ice” but the link is not explicit nor explained. Suggest restructuring this sentence.

Line 34-35 “Long-term ground ice loss is associated….” This sentence should be supported by appropriate reference(s). This recent one covers the topics described: https://doi.org/10.1002/ppp.2261

Line 37. Consider indicating specifically which traditional methods you mean (e.g., probing, thaw tubes, dGPS surveys, etc).

Line 39. Consider examining use of “utilized” throughout the text and replace with “used”, which is more concise and generally has the same meaning.

Line 43. “stronger consolidation” suggest changing this to “larger magnitude thaw subsidence”.

Line 44. “likely drains” suggest changing to “may drain away.”

Figure 1. The stream and lake (reservoir) colour is the same as InSAR heave; suggest changing all waterbodies to a colour not in the InSAR legend. The inset map of Svalbard is very hard to discern, and the contrast between land and water is poor. Consider enlarging and colour changes.

Line 105. Add “anticipated” before “InSAR subsidence magnitudes” since 2023 magnitudes were not known when sites were selected.

Line 111. Clarification on sampling – “soil moisture conditions were considered to include dry and wet locations”. Explain specifically how soil moisture conditions were considered. I presume there were not soil moisture instruments at each site, and that this was done based on some visual or field interpretation?

Line 117. “0.5 m core” is ambiguous, indicate this is the core length.

Line 119. “freezing container” is unclear. Do you mean a cooler? Or something that actively freezes contents?

Line 127. Add “area” after “surrounding”.

Table 1.  Header for column 5 does not indicate that the information at the top of the cell is the ALT measurement date. Row E8 – “Drill location  in center” of what (ice wedge polygon)?

Line 139. What classification was used to classify cryostratigraphy? A reference should be provided.

Line 151. “the factor 1.09 represents the density of ice relative to water”. The equation deals with volumes, so it is better to say that the 1.09 is to “estimate the equivalent volume of ice” from the water volume as Kokelj and Burn did.

Line 156. This should be 9% shouldn’t it? This is why the factor in Eq. 2 is 1.09. This would also affect your derivation of Eq. 4. You have 0.92 in Eq. 4 but this should be 0.912 I think, so rounded to 0.91. So, you will likely have to redo your calculations though they won’t differ much. I think the confusion/error lies in the fact that the percent difference is 9.2% (e.g., see percent difference equation at https://www.calculatorsoup.com/calculators/algebra/percent-difference-calculator.php).  

Line 164. Change “which” to “that”.

Line 176. Can you clarify to the reader whether “temporal baseline” is synonymous with the return frequency of the satellite?

Line 227. Indicate it’s expected for two-sided freezing specifically.

Line 295. Delete “coring” and “located”, these words are not required. Also, suggest changing “rather” to “mostly” in second sentence.

Line 316. Add “sand” after “dry”.

Line 317. Add “early in the thawing season” after “quickly”

Figure 8. “Grain type” figure title should perhaps be changed to “Material type” because organic is not a grain type, and neither is ice lens, or disturbed.

Line 341. Add “from the active layer and upper permafrost” after “in situ-ice contents”.

Line 357. Remove “excess” because ice-rich permafrost, by definition, includes excess ice. Check this throughout.

Line 365. I think it would be good to give a few examples of sites from Figure 6 where it dominates (e.g., the A sites).

Line 371. This part could be strengthened by giving examples of the magnitudes/proportions accounted for by excess ice in the late thawing season.

Line 393. “drainage variations can control the presence or absence of excess ice”. While I don’t disagree, because fundamentally moisture is required for ice formation, based on detailed coring I conducted in the eolian sediments (a GSC Open File is now in press), excess ice was mainly controlled by grain size of the eolian materials, regardless of present-day moisture conditions in the polygons (the polygon with standing water and wet active layer had on average half the excess ice content in the top 1 m of permafrost). Siltier layers, which imply slower rates of loess aggradation and different climatic, eolian source conditions, and likely microtopography in different time periods, were associated with higher ice contents. You have not measured “lateral drainage” (l. 395) in this study, though you may have observed it at the surface. Also, it is hard to know whether drainage conditions at the surface today reflect those when the ground ice aggraded in the past, as the syngenetic polygons fields are dynamic. Therefore, you cannot confidently say that the drainage conditions are controlling the ice contents at much depth beyond the current permafrost table; this text should be modified.

Line 412 last word: Should be “basis”.

Line 435. This sentence suggests that InSAR time series could be used in conjunction with models incorporating ice segregation processes/excess ice content. The reader is left confused what the objective of such an exercise is. Is it to use the excess ice content from the model to validate an InSAR signal? If so, this would surely not be appropriate given that such models cannot accurately capture conditions that lead to excess ice formation over hundreds or thousands of years, and thus cannot produce accurate estimations of ground ice at the site scale. This is discussed in, e.g.,: https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2023JF007262. If this is not the intent, then can you please clarify specifically what you mean in terms of “combining” InSAR with such models?

Line 444. At this time of the year, the whole active layer is frozen, so the meaning of this sentence is unclear. Please clarify what you mean in terms of where/how water is moving at this time in relation to the anticipated ground temperature gradient(s) from the ground surface to the upper permafrost at the time of drilling.

Line 462. Remove “excess”.

Line 463. Why might probing be less precise than borehole measurements? This is highly dependent on the spacing of thermistors, and the material being probed in. This should either be explained further or amended.

H. Brendan O’Neill