the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 28 Oct 2025
Linking Permafrost Deformation to Active Layer Freeze-thaw Dynamics in the Qinghai-Tibet Engineering Corridor
Abstract. The Qinghai-Tibet Engineering Corridor (QTEC), traversing hundreds of kilometers of sensitive permafrost, hosts critical lifelines increasingly threatened by climate-induced ground deformation and thermokarst disasters. However, we still lack knowledge about the spatiotemporal characteristics of permafrost states and associated ground deformation at a finer scale. Here we derive ~120-m-resolution surface displacements during 2014–2022 from Sentinel-1 interferometric synthetic aperture radar (InSAR) data processing. We disentangle secular displacement rates from seasonal variations, quantified seasonal amplitudes and timings, and determine displacement directions through complementary ascending and descending observations. Results reveal extensive subsidence throughout the QTEC, exceeding 20 mm/year in areas between Golmud and Nagqu. Seasonal deformation, driven primarily by frost-heave, thaw-settlement cycle of permafrost, can surpass 40–80 mm, with valley floors peaking in spring and hillslopes peaking in autumn due to different hydro-thermal and mechanical responses. Seasonal amplitude of vertical displacements effectively constrains ALT, indicating ALT ≥1 m where seasonal amplitude exceeds 55 mm. With the constraint from in-situ ALT measurements and the assumption of a complete ice melt in the fall, a freeze-thaw density-change model pictures a regional map of ALT at high resolution, where the permafrost to the west of QTEC exhibits a greater ALT compared to the eastern section. Alarming numbers of thermokarst phenomena (18 thaw slumps, 2,812 thaw lakes) within 2 km of critical infrastructure underscore escalating hazards. Our findings emphasize the urgent need for integrated monitoring and adaptive strategies to mitigate intensifying risks from permafrost degradation across the QTEC.
- Preprint
(3418 KB) - Metadata XML
-
Supplement
(1223 KB) - BibTeX
- EndNote
Status: open (until 21 Feb 2026)
-
RC1: 'Comment on egusphere-2025-4184', Anonymous Referee #1, 09 Jan 2026
reply
-
AC1: 'Reply on RC1', Xie Hu, 02 Feb 2026
reply
Reply on RC1:
The authors describe a study that derives surface displacement from InSAR data in an effort to determine changes in active layer thickness in the QTEC. The manuscript addresses an important topic because understanding of changes in permafrost conditions and the implications for ground stability is important for informed engineering design and adaptation to a changing climate. However, I have some concerns with the interpretation of results and their validation. My expertise is related more to permafrost related processes rather than use of InSAR data and my comments are therefore related to interpretation of the surface displacement results.
Thank you for your time in evaluating our manuscript and your insightful comments. Please found our response below. The line numbers correspond to the tracked-change version, all markup mode of our new submission.
The amount of surface displacement including ongoing subsidence of the ground as permafrost thaws depends on the surficial material characteristics, especially the ground ice content. Excess ice (ice that when melted exceeds the water holding capacity of the material) content is especially important for thermokarst processes which have implications for infrastructure. A large change in ALT therefore does not necessarily result in large changes in surface elevation because thaw of ice-poor material will yield less displacement than an equivalent amount of thaw in ice-rich material. No information on surficial geology or the potential for ice-rich material has been presented in the manuscript, and this is important for the interpretation of the displacement results. It is also unclear whether the occurrence of excess ice (and the ice segregation process) has been considered or only the formation and melt of pore ice. See additional comments below.
Thank you for your comments. Accordingly, we generate maps of geology and ground ice content.
Figure. Geological map and ground ice content map.
There might be two major misunderstandings that we would like to clarify. One likely misunderstanding is the approach we use to infer the active layer thickness. Before demonstrating our approach, we would like to recall the definition of two relevant terms: volumetric water content and ground ice content. Although the term “ground ice content” does not appear in our manuscript, the reviewer references it several times in the comments.
- The volumetric water content is the ratio between of the volume of water and the total volume of host materials, i.e., the total volume of water, ice (if any), soils, and air. In this study, we apply the soil moisture derived from remote sensing analysis as the volumetric water content in the thaw season when the ice completely thaws. As the remote sensing analysis fails to capture the soil moisture in the freeze season due to snow/ice land cover (e.g., reported NULL for vast land), we turn to invert the volumetric water content in the freeze season based on the in-situ observations of ALT.
- Ground ice content is usually represented as a percentile of volumetric water content. We download the ground ice content between 2 and 10 m’s depth from the National Tibetan Plateau Data Center / Third Pole Environment Data Center (https://data.tpdc.ac.cn/zh-hans/data/bd7fc972-3707-4c9c-90c4-aae6bfbc3cbf). According to the dataset introduction, the ground ice content was obtained by incorporating 644 in-situ observations and 13 environmental variables using random forest method. The ground ice content is reported at depths of 2-3 m, 3-5 m, and 5-10 m. We average the ground ice content and visualize the results as below.
Figure. Mean ground ice content at depth of 2-10 m (Data source: National Tibetan Plateau Data Center / Third Pole Environment Data Center).
The schematic view below illustrates the frost-heave to thaw-settlement transition in the active layer. Panel a represents the freeze season and panel b represents the thaw season. Water, ice, and soil blocks are separated to facilitate the math derivation. Vertical gray bars represent the frozen parts of the active layer during the freeze time epoch, and vertical brown bars represent the thaw parts of the active layer during the thaw time epoch. h is the active layer thickness, and delta is the seasonal amplitude of vertical deformation due to the transition between ice and water in the active layer. In our approach, we only need the volumetric water content in freeze season (n) to represent the thickness of unfrozen water, and the volumetric water content in thaw season (N) to represent the thickness of unfrozen and melt water assuming the ice completely thaws. In other words, we do not need the ground ice content. If observations about ground ice content in the freeze and thaw seasons are available, we would be able to improve this model by quantifying the portion of unfrozen water; however, we cannot find direct observations on seasonally ground ice content. To avoid confusion, we do not put the new figure about ground ice content in our manuscript.
Figure. The schematic view below illustrates the frost-heave to thaw-settlement transition in the active layer.
The other likely misunderstanding is that we do not intend to obtain a change in ALT (added in lines 108 and 219). We disentangle the secular and seasonal ground deformation from December 2014 to September 2022 based on Sentinel-1 InSAR processing. We use the averaged seasonal ground deformation spanning these eight years to infer characteristic ALT, an averaged value spanning the same time and at the same resolution of our derived ground deformation results (120 meters). For large-scale quantification of ALT, we simplify the physical process by presuming an overall water balance of the active layer and do not consider the excess ice (and the ice segregation process) (added in lines 111 and 217, 238).
We are unsure about the reviewer’s concern “A large change in ALT therefore does not necessarily result in large changes in surface elevation because thaw of ice-poor material will yield less displacement than an equivalent amount of thaw in ice-rich material.” First, our study does not examine changes in ALT. We add in line 219 “Note that we do not intend to infer temporal changes of ALT in this study due to short time frame”. Second, the phrase “an equivalent amount of thaw” is ambiguous; it is unclear whether it refers to an equivalent volume of ice/water thawed or to an equivalent temperature forcing that drives thaw. Third, we are unsure what materials mean by “ice-rich” and “ice-poor”—whether it refers to permafrost only or the combined active layer–permafrost column. To clarify, to achieve a regional inversion ALT, we attribute the seasonal ground deformation only to the seasonally interchangeable transition between liquid water and frozen water in the active layer, and assume that the total amount of water, regardless of the states, is overall balanced. To enhance our determination of active layer thickness, we consider the volumetric water content in the freeze and thaw seasons when the ground surface reaches the highest and the lowest, respectively. To put it short, our subject is the transition between liquid water and frozen water in the active layer, rather than the thaw of ice-rich or ice-poor materials. Therefore, the derivation of ALT is based on the intrinsic density difference between water and ice. We further clarify our method in section 2.4.
Validation of the modelled ALT values using field data is also unclear. A lot of data has been collected in the QTEC that facilitates estimates of ALT and also descriptions of subsurface material characteristics, including ice content and thaw sensitivity. The lack of clarity regarding how this field-based information was utilized makes it difficult to evaluate the validity of the analysis used to derive the ALT results.
We elaborate our methods, especially on how the field data of ALT being used in this study. Presuming an overall water balance of the active layer, we can estimate ALT () based on the seasonal amplitude of the vertical displacement from the projection of the radar’s line-of-sight measurements. Note that we do not intend to infer temporal changes of ALT in this study due to short time epoch. We extract the averaged seasonal amplitude of ground deformation from December 2014 to September 2022 to infer the characteristic ALT during the same time frame.
Equation 2
where roh_w and roh_i are the density of water and ice, and here we apply 1,000 and 917 kg/m3 respectively; n and N are the volumetric water content in dry frozen and wet thaw conditions, respectively. Here we assume that the soil moisture in October, when the land surfaces reach the lowest, represents the volumetric water content during the wet thaw condition (N). Launched in 2015, the Soil Moisture Active Passive (SMAP) mission provides a global-scale soil moisture monitoring approach through its L-band radar (operation ceased in early 2015) and L-band radiometer (operational to date). This study employs the SMAP’s Level 4 "root zone" (0-100 cm) soil moisture across the QTEC region. Monthly average soil moisture in QTEC from 2016 to 2022 were acquired via the Google Earth Engine platform, incorporating data at 3-hourly and 9-km resolution. Nonetheless, due to snow cover, the volumetric water content in dry frozen condition (n) can not be observed directly (Du et al., 2025), and it is not viable to be represented by the soil moisture products from SMAP. Previous studies simply applied a constant 5% (e.g., Li et al., 2023; Zhang and Wu, 2012). Here we fully utilize available in-situ ALT measurements in QTEC to constrained at those sites (Eq. 2; Table S1). Thereafter, we use the spline interpolation to generate a spatially continuous map of . Next, we compute the ALT using Eq. 2 at the same resolution of the inputting seasonal displacement amplitude. Although this workflow involves uncertainties originated from parameters themselves in the dry QTP environment, the derivation of ALT is based on simplified physical processes, instead of data-driven machine learning. More complex physical processes, such as ice segregation, can be incorporated in future work for specific sites where detailed information on the subsurface condition is available.
Regarding the in-situ ALT measurements in QTEC, we summarize the statistics in Table S1 in the supplementary materials.
Table S1. In-situ active layer thickness measurements using in this study. We obtain the volumetric water content in thaw season (N) from soil moisture data. We use the in-situ active layer thickness measurements, seasonal amplitude of ground deformation, and the volumetric water content in freeze season altogether to constrain the volumetric water content in dry spring (n) at the location of in-situ active layer thickness measurements.
References for the Supplementary Materials:
Chen, J., Liu, L., Zhang, T., et al. (2018). Using persistent scatterer interferometry to map and quantify permafrost thaw subsidence: A case study of Eboling Mountain on the Qinghai-Tibet Plateau. Journal of Geophysical Research: Earth Surface, 123(10), 2663–2676.
Niu, F. (2022). Demonstration monitoring data for disease treatment of permafrost project in South Asia Channel – long-term monitoring demonstration for subgrade stability of Chumar River of Qinghai Tibet Railway (2003–2021). National Tibetan Plateau Data Center National Tibetan Plateau Data Center. https://doi.org/10.11888/Cryos.tpdc.271933
Qin, Y., Wu, T., Zhao, L., et al. (2017). Numerical modeling of the active layer thickness and permafrost thermal state across Qinghai-Tibetan Plateau. Journal of Geophysical Research: Atmospheres, 122(21), 11,604-11,620.
Wu, Q., Hou, Y., Yun, H. & Liu, Y. (2015). Changes in active-layer thickness and near-surface permafrost between 2002 and 2012 in alpine ecosystems, Qinghai–Xizang (Tibet) Plateau, China. Glob. Planet. Change, 124, 149–155.
Wu, Q., Yu, W. & Jin, H. (2017). No protection of permafrost due to desertification on the Qinghai–Tibet Plateau. Sci. Rep., 7, 1544.
Wu, Q. & Zhang, T. (2008). Recent permafrost warming on the Qinghai-Tibetan Plateau. Journal of Geophysical Research: Atmospheres, 113, D13108.
Wu, Q., Zhang, T. & Liu, Y (2012). Thermal state of the active layer and permafrost along the Qinghai-Xizang (Tibet) Railway from 2006 to 2010. Cryosphere, 6, 607–612.
Xie, C., Zhao, L., Wu, T. & Dong, X. (2012). Changes in the thermal and hydraulic regime within the active layer in the Qinghai-Tibet Plateau. J. Mt. Sci., 9, 483–491.
Xie, C., William, A. G., Zhao, L., Wu, T. & Liu, W. (2015). Temperature-dependent adjustments of the permafrost thermal profiles on the Qinghai-Tibet Plateau, China. Arct., Antarct., Alp. Res., 47, 719–728.
Zhao, L. & Sheng, Y. (2019). Permafrost and Its Change on the Qinghai-Tibet Plateau. (Science Press).
Zhao, L., Zou, D., Hu, G., Wu, T., Du, E., Liu, G., Xiao, Y., Li, R., Pang, Q., Qiao, Y., Wu, X., Sun, Z., Xing, Z., Sheng, Y., Zhao, Y., Shi, J., Xie, C., Wang, L., Wang, C., & Cheng, G. (2021). A synthesis dataset of permafrost thermal state for the Qinghai–Tibet (Xizang) Plateau, China. Earth System Science Data, 13(8), 4207–4218.
The manuscript also requires editing to improve, language, terminology and clarity.
Thank you for letting us know. We have improved the language and clarity through the text.
Additional comments on the manuscript are provided below.
Additional Comments
L31-104 Introduction – This needs much work. Better organization and reduction of text to focus on only the information that is relevant would be beneficial. Some of the statements are unclear or incorrect. Editing is required to improve language.
The introduction includes an overview about Qinghai-Tibet Engineering Corridor (QTEC) (one paragraph), the thermokarst processes threaten the facilities (one paragraph), monitoring tools especially recent progress on ground deformation monitoring using InSAR (two paragraphs), recent progress on the retrieval of active layer thickness over the QTEC (one paragraph), and a summary about this study (one paragraph). We have tried to shorten the Introduction by moving the recent progress on the retrieval of active layer thickness over the QTEC to the methods/results section but found it might be inappropriate and confusing. We retain them in the Introduction to provide background context. We improve our language through the text. We do not paste our revisions here because there are multiple paragraphs. Kindly please refer to our manuscript for revisions.
L31-35 –Note it is thaw of ice-rich material that can result in ground deformation so it is not necessarily caused directly by climate change. Use “thermokarst processes” rather than “thermokarst disasters”. A disaster depends on the consequences of the process and whether they are important with respect to infrastructure, human life for example. Note, the active layer is not part of permafrost. The last sentence in the paragraph about GCOS is not really necessary. It is sufficient to refer to active layer thickness as an indicator of permafrost change.
We update the relevant sentence. Now it reads “The contemporary thaw of ice-rich materials drives ground deformation and thermokarst processes in QTEC”.
We update the term and use “thermokarst processes” throughout the text.
Thank you for pointing out the issue in our description. We update the definition of active layer, and it reads “The active layer is the surface layer that seasonally thaws downward in summer and freezes bidirectionally in winter above the permafrost.”
We agree that the last sentence in the paragraph about GCOS is not necessary and have removed this sentence.
L44-55 – When describing rates of change, the period over which the change has occurred should be given. Here and elsewhere in the paper – refer to permafrost thaw rather than melt. Several statements are unclear or use poor terminology. Repetition needs to be reduced.
We update relevant words and use “permafrost thaw” through the text.
L57 – Delete “dynamic” redundant (“changes” is sufficient)
We have removed “dynamic”.
L57-60 – I assume you are referring to topographic surveys that utilize GNSS, i.e. measuring surface elevation. Surface displacement isn’t directly measured but derived from elevation measurements made over time.
Extensometers, inclinometers, and GNSS can be used to measure displacement at locations where the sensors were installed. We have shortened this sentence. Now it reads “While field observations of temperature, soil moisture, and displacement deliver accurate permafrost status at isolated site, they are labor-intensive and spatially limited across the QTP.”
L63-65 – It is incorrect to say InSAR-derived displacement patterns directly reflect permafrost dynamics. They reflect changes in surface elevation. Surface displacement can be caused by various processes including groundwater withdrawal, sediment compaction, crustal movements etc. Slope movements aren’t necessarily triggered by permafrost change. The attribution of changes requires knowledge of the local conditions including materials and their thermal state, and geologic/geomorphic processes etc. – result of analyses of different types of data.
Thank you for your suggestions. We update the text and now it reads “Time-series interferometric SAR (InSAR) further enables detection of ground deformation at regional scales. However, InSAR-derived displacement patterns do not, by themselves, directly indicate permafrost dynamics: they capture changes in surface motion that may arise from multiple processes (e.g., ground ice melt and thaw settlement, slope instability, groundwater changes, sediment compaction). Robust attribution to permafrost-related mechanisms therefore requires integration with local ground conditions and thermal state, as well as supporting geologic and geomorphic context and complementary datasets (e.g., topography, temperature, and land cover).”
L73 – “linear subsidence” is unclear – is this vertical subsidence?
Yes, it is. Now it reads “They documented an annual vertical subsidence rate of up to 20 mm/yr.”
L76-79 – Is this rate for seasonal movement or cumulative over several years?
It is the average rate over several years. Now it reads “the annual subsidence rate faster than 2 mm/yr”.
L98-104 – Is information on surficial geology and ground ice content used. Knowing ALT on its own is sufficient when determining displacement patterns. Is excess ice considered?
We incorporate the volumetric water content in the freeze and thaw seasons to infer the active layer thickness. By assuming a complete thaw in the ground surface reach the lowest when the volumetric water content can be represented by soil moisture, our approach does not require information about the ground ice content and excess ice. But we also note that if the seasonal estimates of ground ice content are available, we can better know the unfrozen water in the freeze and thaw seasons and thus improve this model. Because such datasets are unavailable, we do not introduce ground ice content in our interpretation. Please find our earlier response for more information.
We collect geological maps from the four sources (i, Searle et al., 2011; ii, Online Geological Map of Tibetan Plateauhttps://osgeo.cn/map/m02c2/; iii, 1:1.5 million geological map of Tibetan Plateau and its surrounding areas https://www.tpdc.ac.cn/zh-hans/data/ce047b38-72b4-46a7-8fbe-c9019e6dd5cf/; iv, 1:1.5 million geological map of China). The former three do not have GIS shapefiles. As ALT is essentially inferred from seasonal amplitude of ground deformation in the vertical direction, we superimpose our deformation products on the geological maps for a qualitative analysis. We show below the seasonal amplitude of ground deformation and secular deformation velocity from the descending line-of-sight Sentinel-1 as examples.
We add in line 255 “Overall, the association between displacement and geological conditions is inherently stochastic”. We also add the figure below in the supplementary material.
(Add in the supplementary material) Figure. Geological map (Source: 1:1.5 million geological map of China).
Nonetheless, in our enlarged area of interest around Wudaoliang in Fig. 4, a belt of large displacements overlaps with Quaternary glacial deposits (another source of geological map shows Quaternary alluvial plain; see the second figure below). We add in line 278: “When referring to the geological map, we note such pronounced displacements are well aligned with the extent of Quaternary glacial deposits, suggesting a local impact from the surficial geology”.
(Updated by added a new panel g) Figure 4: Enlarged ground deformation decomposition results around Wudaoliang (the red box in Fig. 3a). a and b show the secular displacement velocity from ascending and descending results, respectively. c and d show the peak-to-trough seasonal displacement amplitude from ascending and descending results, respectively. e and f show seasonal timing (day) when the ground reaches the upmost (peak) in one calendar year ascending and descending results, respectively. g shows the geological map.
Figure. Visual comparison between the deformation fields and the geological map (Source: 1:1.5 million geological map of Tibetan Plateau and its surrounding areas).
L120 – “compiled” is probably more correct than “collected”
Updated.
L176 – What is meant by long-term – what is the period considered.
Now it reads ”the long-term ground deformation trend from December 2014 to September 2022”.
L213 – Replace “can hardly” with “can not”
Updated.
L222 – Is this based on cumulative subsidence over several years or just seasonal movement?
It is based on cumulative subsidence over several years. Now it reads “with rates ranging from a few to over 20 millimeters per annum from December 2014 to September 2022”.
L225 – Avoid words like “drastic”
We update the text to “localized subsidence”.
L226-227 – Permafrost isn’t required to have seasonal heave and subsidence as this can occur where only seasonally frozen ground is present. The change may have more to do with a change in surficial geology and frost susceptibility.
As shown in Fig. 6, the color dot suggesting seasonal ground deformation mainly occur on the dark blue background (permafrost region) instead of the light blue background (seasonally frozen ground). We describe that “seasonal displacements clearly cease at the boundary between permafrost regions to seasonally frozen ground regions”. We add in line 254 “Surficial geology and frost susceptibility may jointly regulate the ground deformation.”
L253 – Are displacements more pronounced here due to the surficial materials?
We believe the large displacements may result from variations in the volume of ground ice and the thermal conditions.
L262-263 – Occurs at/prior to the onset of thaw?
We use the frozen instead of thaw for the description following the temporal sequence. The highest surface position for most of the flat areas occurs in spring when the underlying water/ice-rich layer reaches its most frozen state.
L312-313 – It is unclear whether there is consideration of segregation ice (ice lens formation) which forms as water migrates to the freezing front or just the 9% expansion of pore ice. The segregation process is responsible for excess ice formation which is important for thermokarst processes.
We focus on the regional ground deformation and do not consider ice segregation. It is beyond the scope of this research to quantify the role of ice segregation in ground deformation encompassing such large region.
L315-321 – It takes time for changes in surface temperature to propagate to greater depths. Thaw may still be occurring at depth when onset of freezing occurs at the surface.
Thank you for your comment. We share the same core idea. Heat/temperature signals at the surface take time to propagate downward, so deeper soil temperatures lag behind land surface temperatures. We elaborate the process:
“In-situ measurements from the TGL permafrost station indicate that minimum temperatures at shallow depths occur in January—when surface freezing may already begin—whereas deeper layers continue cooling and do not reach their minimum temperature until March.”
L333-335 – ALT can be derived from ground temperature measurements, frost probing and thaw tubes. I’m not sure that most would call extraction of stratigraphic profiles the most fundamental method of determining ALT.
Thank you for your suggestion. We update our description:
“The general method for determining active layer thickness involves directly measuring maximum seasonal thaw depth in the field (e.g., probing, pits/cores, thaw tubes), or indirectly inferring it from subsurface temperature profiles, calibrated heat-transfer models, or geophysical surveys (e.g., electrical resistivity tomography, ground-penetrating radar, seismic).”
L350 – Is excess ice considered?
No, we do not consider excess ice. We clarify the sentence:
“InSAR-derived time-series displacement provides a solution for quantifying large-scale ALT by leveraging the basic principle of density differences between ice and water assuming an overall water balance, regardless of the water states, in the active layer through our study period of eight years”.
L361-362 – ALT is not directly measured using GPR surveys but interpreted from the data – requires additional site knowledge etc.
We update this sentence to “In the QTP, ALT is primarily measured using soil sensors, boreholes, or inferred from ground-penetrating radar imaging”.
L380-388 – How do these derived values of ALT compare to measured values?
We use the in-situ measurements of ALT to constrain the corresponding volumetric water content in the frozen season. Therefore, the consequent spatially continuous map of our derived values of ALT is the same to the measured values. We attempt to effectively use the measured values of ALT in our model.
Relevant information is available in the method section:
“Here we fully utilize a limited number of in-situ ALT measurements in QTEC, constrain at those sites where all other parameters are available (Eq. 2; Table S1). We use the spline interpolation to generate a map of after removing anomalies. Thereafter, we compute the ALT at the same resolution of the input seasonal displacement amplitude (Eq. 2). Although this workflow involves uncertainties originated from parameters themselves in the dry QTP environment, the derivation of ALT is based on physical processes, instead of data-driven machine learning.”
L413-415 – Refer to thermokarst processes rather than disasters.
Updated through the text.
L412 – Section 4.3 – The rest of the paper focussed on vertical displacements derived from InSAR. It isn’t clear whether displacements derived from InSAR are used to generate the results presented here regarding thaw slumps etc. or some other technique given other satellite products are mentioned.
Because this study focuses on the Qinghai–Tibet engineering corridor, we report thermokarst statistics within buffer zones along the corridor to emphasize the need for sustained monitoring of ongoing deformation in this critical region.
L493-496 – Note that these studies do not consider infrastructure design in their analysis. This is important because infrastructure design will influence the impact of permafrost thaw on the infrastructure. These studies are often done at scales that are insufficient for more regional to local scale assessments which are required when considering infrastructure impacts.
Thank you for your careful thought. We add in the end of this section “Furthermore, the present-day studies do not usually consider how infrastructure is designed, built, or maintained, which may also impact the soil’s thermo-hydro-mechanical properties in permafrost regions and greatly change how permafrost thaw affects damage. To support decisions for a specific corridor or community, finer-scale studies are needed that include local conditions and detailed engineering information about the infrastructure.”
-
AC1: 'Reply on RC1', Xie Hu, 02 Feb 2026
reply
-
RC2: 'Comment on egusphere-2025-4184', Anonymous Referee #2, 03 Feb 2026
reply
The authors focus on the ground deformation along the Qinghai–Tibet Engineering Corridor, a key infrastructure zone that crosses extensive, climate-sensitive permafrost. Using multi-year Sentinel-1 InSAR observations, the authors separate long-term displacement trends from seasonal freeze–thaw cycles and infers deformation directions by combining ascending and descending satellite tracks. The results indicate widespread subsidence and pronounced seasonal deformation linked to frost heave and thaw settlement, with different timing patterns between valley floors and hillslopes that reflect contrasting hydro-thermal and mechanical controls. The seasonal deformation signal is then used to infer active layer thickness with constraints from in-situ measurements of active layer thickness, revealing broad regional differences along the corridor. The work also documents substantial thermokarst activities including thaw slumps and thaw lakes near major infrastructure. The paper is highly informative, which is a strength, but it can also be distracting. With all comments well addressed, the paper can be considered for publication.
Detailed comments are outlined below:
The authors may consider adding InSAR in the title to highlight the technique and spatially continuous mapping.
Line 54: “This situation presents substantial risks to the transportation safety of QTEC (Hjort et al., 2022; Yu et al., 2013).” Do both references talk about to the condition in QTEC?
Table 1: The authors can add an illustrative figure about the geometry of satellite InSAR.
Line 180: “we identify and tally the peaks and troughs of the seasonal deformation. Typically, one calendar year holds one peak and one trough. This allows us to compile a collection of pixels with an anticipated number of peaks and troughs.” The authors should emphasize that this approach only favors for multiple years of observations.
Figure 2: add label “not to scale”.
Figures 5 and 6: The authors used displacement in the figure captions and legends but deformation in the text. The authors should use consistent wording to avoid confusion. The authors should rewrite the captions. What are the white polygons in the figures?
Line 252: change “permafrost dynamics” to “subsurface environment”.
Line 260: “On consolidated hillslopes, elastic unloading peaks in late summer when substantial meltwater drainage occurs, thereby leading to the peak surface position.” This is only a plausible reason. Can it be related to different arrival time of the lowest ground temperature at different altitude?
Line 272: what is the meaning of “spaceborne line-of-sight measurements represent real movements”?
Line 275: what are the traditional and non-traditional interpretations of secular and seasonal displacement directions?
What is the purpose of section 3.2 Disentanglement of ground deformation? It doesn’t seem to show any regulations in Figures 5 and 6.
Line 315: elaborate “hydro-thermal alterations”
Line 321: “coupling” is too vague.
Line 336: “at a limited number of isolated locations” and “the monitoring stations are clustered”. It sounds contrasting.
Line 338: “assumption-based numerical models and limited-data-driven machine learning approaches”. This should be rewritten.
Line 358: what is the meaning of “soil texture” here?
Figure 8: What is the meaning of “SeaAmpVel” in the right panels?
Section 4.2 Implications for quantifying the active layer thickness. This section is a critical part of the manuscript, but it is too crowded. The author can provide subtitles to make the context flow better.
The authors can move the last but two paragraph on ALT in the Introduction to Section 4.2.
The “wiggle” in ALT peaks in the frequency histograms in the past through the future can be used as indicators for evolution of permafrost degradation. Please elaborate.
Sections 4.3 Thermokarst disasters over the QTEC and 4.4 Risks mitigation due to permafrost thaw seem to be disconnected with the previous sections. Some connecting context should be added.
Citation: https://doi.org/10.5194/egusphere-2025-4184-RC2 -
AC2: 'Reply on RC2', Xie Hu, 06 Feb 2026
reply
The authors focus on the ground deformation along the Qinghai–Tibet Engineering Corridor, a key infrastructure zone that crosses extensive, climate-sensitive permafrost. Using multi-year Sentinel-1 InSAR observations, the authors separate long-term displacement trends from seasonal freeze–thaw cycles and infers deformation directions by combining ascending and descending satellite tracks. The results indicate widespread subsidence and pronounced seasonal deformation linked to frost heave and thaw settlement, with different timing patterns between valley floors and hillslopes that reflect contrasting hydro-thermal and mechanical controls. The seasonal deformation signal is then used to infer active layer thickness with constraints from in-situ measurements of active layer thickness, revealing broad regional differences along the corridor. The work also documents substantial thermokarst activities including thaw slumps and thaw lakes near major infrastructure. The paper is highly informative, which is a strength, but it can also be distracting. With all comments well addressed, the paper can be considered for publication.
We appreciate for your careful read and a thorough summary of our manuscript and constructive suggestions. We add subtitles for section 4.3, an illustrative figure about the geometry of ascending and descending InSAR satellite trajectories and imaging. We clarify that, theoretically, this approach can be used to infer ALT changes, but need to use with caution because the determination of seasonal deformation amplitude can be biased when the time window is less than a couple of years (Lines 228 and 416; track-changes version of our new submission). We also read through our text and updated some descriptions for clarity.
Please find our point-to-point response as below.
Detailed comments are outlined below:
The authors may consider adding InSAR in the title to highlight the technique and spatially continuous mapping.
Thank you for your suggestion. We update the title to be “Linking InSAR-derived Permafrost Deformation to Active Layer Freeze-thaw Dynamics in the Qinghai-Tibet Engineering Corridor”.
Line 54: “This situation presents substantial risks to the transportation safety of QTEC (Hjort et al., 2022; Yu et al., 2013).” Do both references talk about to the condition in QTEC?
One of the references (Hjort et al., 2024 <it should be 2024 not 2022 in our original text>) is about the risks to transportation safety in permafrost environment across the Earth. We update the text as “This situation presents substantial risks to the transportation safety in permafrost environment (Hjort et al., 2024; Yu et al., 2013).”
Table 1: The authors can add an illustrative figure about the geometry of satellite InSAR.
Thank you for your suggestion. We add an illustrative figure about the geometry of satellites InSAR trajectories and imaging.
Line 180: “we identify and tally the peaks and troughs of the seasonal deformation. Typically, one calendar year holds one peak and one trough. This allows us to compile a collection of pixels with an anticipated number of peaks and troughs.” The authors should emphasize that this approach only favors for multiple years of observations.
We add the limitations. “This approach favors for multiple years of observations. The more the years the less fraction of the uncertain peaks and troughs among the total number.”
Figure 2: add label “not to scale”.
Added in the figure.
Figure 2: Schematic view of frost-heave to thaw-settlement transition in the active layer (not to scale). a, freeze season. b, thaw season. Water, ice, and soil blocks are separated to facilitate the math derivation. Vertical gray bars represent the frozen parts of the active layer during the freeze time epoch, and vertical brown bars represent the thaw parts of the active layer during the thaw time epoch. We refer to Liu et al. (2012) and Li et al. (2023) on the design of this schematic view with modifications.
Figures 5 and 6: The authors used displacement in the figure captions and legends but deformation in the text. The authors should use consistent wording to avoid confusion. The authors should rewrite the captions. What are the white polygons in the figures?
Thank you for noting this. We update and use “deformation” consistently through the text, including those in the figures.
The white polygons in the figures are glaciers. We add in the figure caption “The hollow (white) polygons represent glaciers.”
Line 252: change “permafrost dynamics” to “subsurface environment”.
Changed.
Line 260: “On consolidated hillslopes, elastic unloading peaks in late summer when substantial meltwater drainage occurs, thereby leading to the peak surface position.” This is only a plausible reason. Can it be related to different arrival time of the lowest ground temperature at different altitude?
We cannot conclude that the timing differences in ground deformation are driven by altitude. As the other reviewer noted, surficial geology, ground-ice content, and vegetation conditions may also contribute. Given the lack of high-resolution data and because this is beyond the scope of this study, we prefer not to overinterpret altitude here.
Line 272: what is the meaning of “spaceborne line-of-sight measurements represent real movements”?
It is important to note that spaceborne InSAR measures only the projection of true ground deformation onto the satellite’s line of sight; therefore, line-of-sight observations alone cannot determine the actual deformation direction. We update this sentence for clarity.
Line 275: what are the traditional and non-traditional interpretations of secular and seasonal displacement directions?
Thank you for your comment. We realize that the term “non-traditional” can lead to confusion.
Traditionally, determining the true direction of ground motion requires at least three independent InSAR line-of-sight viewing geometries. This is difficult to achieve with SAR satellites because they operate in near-polar orbits and image to the side, producing look directions that are predominantly downward to the east–west with only a limited north–south component. As a result, sensitivity to north–south motion is intrinsically constrained by the orbital geometry.
We therefore describe our approach as “non-traditional.” Rather than estimating motion direction as an azimuth (degrees clockwise from north)—given the fundamental limitations of InSAR viewing geometry—we use the consistency between ascending and descending line-of-sight measurements to qualitatively infer the dominant deformation mode (e.g., primarily vertical or primarily horizontal). When motion is primarily horizontal, we can sometimes further distinguish whether the east–west or north–south component is dominant.
We simplified our writing by removing “non-traditional” to avoid misleading the readers.
What is the purpose of section 3.2 Disentanglement of ground deformation? It doesn’t seem to show any regulations in Figures 5 and 6.
Thank you for your notes. We intend to look for regulations in the direction of ground deformation in section 3.2. We agree that Figures 5 and 6 do not show a strong regulatory signal. As an overview, for secular deformation, 31% of pixels show similar ascending/descending magnitudes (vertical), 19% show similar magnitudes with opposite signs (east–west; mainly near lakes and seasonally frozen ground around Lhasa and north of Golmud), and 50% are indeterminate. For seasonal deformation, 40% show similar amplitudes and in-phase timing (vertical; common in permafrost near Chumaerhe and south of Fenghuoshan), <1% indicate east–west motion (similar amplitudes, 180° out of phase), and the remaining 60% are indeterminate (9% similar amplitudes with other phase offsets; 51% different amplitudes). We anticipate that our movement-direction inference approach can be applied across broader permafrost regions to produce more robust statistics.
Line 315: elaborate “hydro-thermal alterations”
The temperature variation at greater soil depths typically responds to changes in air and shallow ground temperatures with a time delay, depending on the thermal diffusivity. In-situ measurements from the TGL permafrost station indicate that minimum temperatures at shallow depths occur in January, when surface freezing may already begin, whereas deeper layers continue cooling and do not reach their minimum temperature until March (Zhao et al., 2022).
Line 321: “coupling” is too vague.
We change it to “correlation”.
Line 336: “at a limited number of isolated locations” and “the monitoring stations are clustered”. It sounds contrasting.
Thank you for your notes. Now it reads:
These methods measure frozen-soil deformation and active-layer thickness at only a limited number of sites and are often limited by accessibility. On the QTP, monitoring stations are concentrated along the highway because access is easier there, whereas the vast lands with low population density to the west of QTEC have a notable scarcity of stations (Ni et al., 2021).
Line 338: “assumption-based numerical models and limited-data-driven machine learning approaches”. This should be rewritten.
Thank you for pointing it out. Now it reads:
To estimate large-scale, continuous ALT, numerical models and machine-learning approaches are widely used, but both have clear limitations. Models such as GIPL2 rely on simplifying assumptions (e.g., no water movement and no heat sources or sinks; Debolskiy et al., 2020), while machine-learning methods are data-driven and lack explicit physical derivation, particularly, the data in QTP are often limited as well.
Line 358: what is the meaning of “soil texture” here?
Sorry for the confusion. We update it to “vegetated”. Now it reads:
The relationship between seasonal deformation and ALT varies with the vegetated and moisture conditions.
Figure 8: What is the meaning of “SeaAmpVel” in the right panels?
It means the seasonal amplitude of vertical deformation. We update the abbreviation to be “SeaAmpVer” and added the description in the figure caption.
Section 4.2 Implications for quantifying the active layer thickness. This section is a critical part of the manuscript, but it is too crowded. The author can provide subtitles to make the context flow better.
Thanks for your suggestion. We provide subtitles:
4.2.1 ALT estimates on the QTP
4.2.2 Relationship between ALT and seasonal deformation
4.2.3 Derivation of ALT from InSAR-derived ground deformation
4.2.4 Comparison of ALT estimates
The authors can move the last but two paragraph on ALT in the Introduction to Section 4.2.
Thank you for your suggestion. We move this paragraph to section 4.2.2.
The “wiggle” in ALT peaks in the frequency histograms in the past through the future can be used as indicators for evolution of permafrost degradation. Please elaborate.
Sorry for the confusion. We mean the ALT peaks in the frequency histograms over QTEC may change through time. For example, the ALT is mostly between 4 and 6 meters for one decade but may change to 3-5 or 5-7 meters in the next decade. The overall thickening or thinning of ALT may infer the stage of permafrost degradation or recovery. It might not be appropriate to demonstrate this casual example. We update the text and now it reads:
The “wiggle” pattern of ALT modes (peaks) in the frequency histograms from past to future can be interpreted as an indicator of evolving permafrost conditions. In other words, the modal ALT over QTEC is expected to shift through time, reflecting changes in the most common active-layer thickness. A systematic shift of the distribution toward thicker ALT, or a broadening and migration of its dominant peak(s), may indicate progression in permafrost degradation, whereas a shift toward thinner ALT may be consistent with stabilization or recovery.
Sections 4.3 Thermokarst disasters over the QTEC and 4.4 Risks mitigation due to permafrost thaw seem to be disconnected with the previous sections. Some connecting context should be added.
Thank you for your comment. We update the beginning sentences of these two sections to make it flow better.
4.3 Thermokarst processes over the QTEC
Thermokarst processes poses direct risks to the engineering integrity of infrastructures constructed on permafrost along QTEC (Niu et al., 2011; Xia et al., 2022; Zhou et al., 2024).
4.4 Risks mitigation due to permafrost thaw
Mitigating permafrost-thaw–induced damage has become a key priority for the government.
-
AC2: 'Reply on RC2', Xie Hu, 06 Feb 2026
reply
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 333 | 89 | 34 | 456 | 48 | 26 | 27 |
- HTML: 333
- PDF: 89
- XML: 34
- Total: 456
- Supplement: 48
- BibTeX: 26
- EndNote: 27
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Comments to the Authors
The authors describe a study that derives surface displacement from InSAR data in an effort to determine changes in active layer thickness in the QTEC. The manuscript addresses an important topic because understanding of changes in permafrost conditions and the implications for ground stability is important for informed engineering design and adaptation to a changing climate. However, I have some concerns with the interpretation of results and their validation. My expertise is related more to permafrost related processes rather than use of InSAR data and my comments are therefore related to interpretation of the surface displacement results.
The amount of surface displacement including ongoing subsidence of the ground as permafrost thaws depends on the surficial material characteristics, especially the ground ice content. Excess ice (ice that when melted exceeds the water holding capacity of the material) content is especially important for thermokarst processes which have implications for infrastructure. A large change in ALT therefore does not necessarily result in large changes in surface elevation because thaw of ice-poor material will yield less displacement than an equivalent amount of thaw in ice-rich material. No information on surficial geology or the potential for ice-rich material has been presented in the manuscript, and this is important for the interpretation of the displacement results. It is also unclear whether the occurrence of excess ice (and the ice segregation process) has been considered or only the formation and melt of pore ice. See additional comments below.
Validation of the modelled ALT values using field data is also unclear. A lot of data has been collected in the QTEC that facilitates estimates of ALT and also descriptions of subsurface material characteristics, including ice content and thaw sensitivity. The lack of clarity regarding how this field-based information was utilized makes it difficult to evaluate the validity of the analysis used to derive the ALT results.
The manuscript also requires editing to improve, language, terminology and clarity.
Additional comments on the manuscript are provided below.
Additional Comments
L31-104 Introduction– This needs much work. Better organization and reduction of text to focus on only the information that is relevant would be beneficial. Some of the statements are unclear or incorrect. Editing is required to improve language.
L31-35 –Note it is thaw of ice-rich material that can result in ground deformation so it is not necessarily caused directly by climate change. Use “thermokarst processes” rather than “thermokarst disasters”. A disaster depends on the consequences of the process and whether they are important with respect to infrastructure, human life for example. Note, the active layer is not part of permafrost. The last sentence in the paragraph about GCOS is not really necessary. It is sufficient to refer to active layer thickness as an indicator of permafrost change.
L44-55 – When describing rates of change, the period over which the change has occurred should be given. Here and elsewhere in the paper – refer to permafrost thaw rather than melt. Several statements are unclear or use poor terminology. Repetition needs to be reduced.
L57 – Delete “dynamic” redundant (“changes” is sufficient)
L57-60 – I assume you are referring to topographic surveys that utilize GNSS, i.e. measuring surface elevation. Surface displacement isn’t directly measured but derived from elevation measurements made over time.
L63-65 – It is incorrect to say InSAR-derived displacement patterns directly reflect permafrost dynamics. They reflect changes in surface elevation. Surface displacement can be caused by various processes including groundwater withdrawal, sediment compaction, crustal movements etc. Slope movements aren’t necessarily triggered by permafrost change. The attribution of changes requires knowledge of the local conditions including materials and their thermal state, and geologic/geomorphic processes etc. – result of analyses of different types of data.
L73 – “linear subsidence” is unclear – is this vertical subsidence?
L76-79 – Is this rate for seasonal movement or cumulative over several years?
L98-104 – Is information on surficial geology and ground ice content used. Knowing ALT on its own is sufficient when determining displacement patterns. Is excess ice considered?
L120 – “compiled” is probably more correct than “collected”
L176 – What is meant by long-term – what is the period considered.
L213 – Replace “can hardly” with “can not”
L222 – Is this based on cumulative subsidence over several years or just seasonal movement?
L225 – Avoid words like “drastic”
L226-227 – Permafrost isn’t required to have seasonal heave and subsidence as this can occur where only seasonally frozen ground is present. The change may have more to do with a change in surficial geology and frost susceptibility.
L253 – Are displacements more pronounced here due to the surficial materials?
L262-263 – Occurs at/prior to the onset of thaw?
L312-313 – It is unclear whether there is consideration of segregation ice (ice lens formation) which forms as water migrates to the freezing front or just the 9% expansion of pore ice. The segregation process is responsible for excess ice formation which is important for thermokarst processes.
L315-321 – It takes time for changes in surface temperature to propagate to greater depths. Thaw may still be occurring at depth when onset of freezing occurs at the surface.
L333-335 – ALT can be derived from ground temperature measurements, frost probing and thaw tubes. I’m not sure that most would call extraction of stratigraphic profiles the most fundamental method of determining ALT.
L350 – Is excess ice considered?
L361-362 – ALT is not directly measured using GPR surveys but interpreted from the data – requires additional site knowledge etc.
L380-388 – How do these derived values of ALT compare to measured values?
L413-415 – Refer to thermokarst processes rather than disasters.
L412 – Section 4.3 – The rest of the paper focussed on vertical displacements derived from InSAR. It isn’t clear whether displacements derived from InSAR are used to generate the results presented here regarding thaw slumps etc. or some other technique given other satellite products are mentioned.
L493-496 – Note that these studies do not consider infrastructure design in their analysis. This is important because infrastructure design will influence the impact of permafrost thaw on the infrastructure. These studies are often done at scales that are insufficient for more regional to local scale assessments which are required when considering infrastructure impacts.