the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 16 Apr 2026
Kilometer-scale distributed temperature sensing reveals heterogeneous permafrost warming near Arctic infrastructure
Abstract. Accelerated Arctic warming is destabilizing permafrost, threatening ecosystems, infrastructure, and northern communities, yet permafrost thermal dynamics remain poorly characterized and highly uncertain due to sparse observations and limited representation of fine-scale heterogeneity. Here we combine kilometer-scale distributed temperature sensing (DTS) with data-driven hysteresis modeling to resolve spatiotemporal variability in permafrost temperatures across disturbed and undisturbed Arctic landscapes near Utqiaġvik, Alaska. A 2-km fiber-optic array recorded continuous ground temperatures from 2021 to 2024, revealing persistent warming associated with civil infrastructure and pronounced thermal heterogeneity in patterned tundra. Ice-wedge polygon troughs consistently exhibit lower temperatures than polygon centers, highlighting the role of subsurface ice distribution in controlling ground thermal regimes. Using these observations, we develop a multivariate hysteresis model that captures lagged ground–air temperature responses and incorporates snow, precipitation, wind, atmospheric pressure, and ground surface conditions. The model accurately reproduces observed permafrost temperatures, fills observational gaps, and enables projections under future climate scenarios. Projections indicate continued warming through 2075, with enhanced temperature increases in infrastructure areas and ice-rich terrains. Our results demonstrate the power of DTS to resolve permafrost thermal heterogeneity at unprecedented scales and provide a transferable framework for predicting permafrost temperature dynamics in a rapidly warming Arctic.
- Preprint
(1990 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-1809', Anonymous Referee #1, 05 May 2026
-
RC2: 'Comment on egusphere-2026-1809', Anonymous Referee #2, 17 Jun 2026
This nicely done study represents one of the few long term applications of fiber-optic distributed temperature (FO-DTS) sensing to the monitoring of fine scale soil temperature patterns in cold regions. While the manuscript is generally well-written and organized, I do not believe the authors make a strong case as to defining compelling and transferrable research objectives framed around process understanding. Instead, too much of the focus seems to be in gap-filling missing (some substantial) FO-DTS data chunks, which is a hyper-local application and bends the paper toward a case study application. The transferrable value of the study comes out in the latter results and discussion sections, but needs some better framing up front to help keep the reader engaged. I have also listed a range of suggestions, more major or minor, below for consideration in a revision.
More major points:
-Clear research questions or hypotheses are needed at the end of the introduction section. As stated, it seems like a lot of work was done to fill gaps in soil temperature monitoring data for one case study that might have limited transferability to other sites. Section 3.3 feels very much like the aim of an isolated case study or site characterization, not broadly useful research on soil temp controls in permafrost areas. In section 4.4, it would be good to reinforce for the reader why this gap filling of measured soil temperatures is necessary and how it adds to the transferrable research aspects of this study. Is the more important use as a statical model to test projected warming scenarios on soil temperature under varied land cover/use? Other studies have already broadly shown soil warming in transitional permafrost areas where infrastructure is added and natural vegetation disturbed (including local changes in ground shading). Perhaps a more novel and is based on section “4.3 Meter-scale thermal heterogeneity associated with ice‑wedge polygon microtopography” as monitored with FODTS. That is the most scientifically interesting section of the manuscript, IMO. Can you dive deeper into the potential drivers of the non-patterned vegetation areas having the coolest relative soil temperatures? Ie Consider speculating as to whether this is attributable to lower surface water, vegetation type, or shallower active layer. Do you have any additional active layer thickness data distributed along the DTS line at point(s) in time?
- Section 3.2 states: “To model permafrost temperature dynamics and reconstruct missing DTS observations, we developed a non-linear, multivariate hysteresis-based framework that explicitly represents the asymmetric, lagged relationship between air temperature and ground temperature, while considering meteorological factors and ground surface conditions. (FODTS cable)”. However, earlier, it is stated that “…burial depths of approximately 20 to 40 cm” . Buried FODTS deployments are tricky as the exact sediment depth is often a strong control on the measured temperature, especially in evaluating time lags from air or the land surface. Given the depth of burial here varied by up to 100%, I would expect large uncertainty to propagate into the statistical modeling, unless data were selected from sections of known common depth when assessing air to ground temp signal time lags. Please explain your approach in more detail related to this concern.
-I would expect soil moisture to be one of the strongest local controls on heat exchange at a given depth. It does not seem that was measured here, which would have been quite difficult to accomplish along the cable without active heating FODTS methods. I do not believe this is a fundamental flaw of the study, but the topic deserves some discussion and summary of related literature. Beyond advective heat exchange due to percolation of precipitation, soil moisture is a primary control on unsaturated soil thermal diffusivity.
-It is not clear what data were used to create the statistical models described in section 3.2. Is that all based on local MET obs and FODTS data from times when the system was fully functional?
-The FODTS instrument type/manufacturer needs to be listed in the methods section. Also, the text states that double ended measurements increase data accuracy. That is not necessarily the case. In double ended mode in the current setup, data collected 2 km further out along a given fiber as a given location, but at the same spatial location due to the fiber loop, are folded onto the data collected nearer to the control unit. Data collected multiple km further out are inherently nosier due to signal loss with distance. So that near and far data averaging process actually decreases the precision (and possibly accuracy) of the near field data. Double ended mode is most important in accounting for step changes in signal loss around fiber splices, which is not accounted for, will decrease measurement accuracy. Also, more detail is needed on how the reference baths were setup.
-L178 and elsewhere: Why was it necessary to estimate missing FODTS data chunks using the model? I am not sure that improved the scientific understanding of the heat exchange processes at hand.
-FODTS data precision increases with time, and data (in this case 5-min) can be stacked after data collection to improve precision. Why was that not done here to achieve better resolution? It seems odd that all data are reported to the nearest whole number, which is atypical of FODTS data, and seems to under-utilize the capability of the technology, especially in nuanced, complex terrain.
- L342 and following paragraph: Why do the FODTS data need to be ‘validated’? FODTS instruments are advanced, high tech methods for measuring temperature. Do you have reason to believe your calibration of the data for offset over time or the dual ended calculations were flawed? As borehole and FODTS data are never truly co-located, the fact your FODTS instrument integrated soil temperature over 1 m linear spatial scales, and the fact temperature is often highly variable in the subsurface, it can be difficult to ‘validate’ in this way. FO-DTS has been used for environmental monitoring for at least 20 years, so there seems little need to broadly ‘validate’ the reliability of the technology, so you may want to reword the final sentence in this paragraph. Also, what do you mean (quantitatively) by “closely align”?
-*I am wondering how much of this story is local vegetation shading related vs infrastructure per se. Section 4.2 is titled “Infrastructure-related permafrost warming” (though it seems we are talking about active layer warming), though I am reminded of this study ( doi: 10.1002/2014GL059251) and others that showed a strong local control on permafrost persistence based on the shading of individual clusters of bush vegetation in transitional permafrost areas.
Minor comments:
- The first sentence of the abstract is too long and complex and should be broken into two
- L27: the more precise acronym is ‘FO-DTS’ to indicate sensing via fiber-optics. Distributed temperature sensing can occur via a variety of technologies including thermal infrared.
- L31: it would be helpful to indicate what type of civil infrastructure you are referring to (ie roads, buildings, etc)
- L36: please add a quantitative metric to support the ‘accuracy’ of simulated vs observed temperatures (ie within 0.5 deg C, or similar)
- L38: why is the year 2075 relevant to state here? I am not aware that is a particular inflection point for arctic warming. If it is simply based on the timescale of the climate forecast you reference, then I suggest removing the year
- L58: Consider adding “and microtopography” since later sections rely heavily on polygonal terrain.
- L74: FO-DTS is the more precise acronym as mentioned above
- L77: ‘temperature dependent backscattering of light’
- L99-105: This does not feel like a complete paragraph. Combined with the one above?
- L107: please state the FODTS measurement spatial resolution, as that is at least as important as the km-scale cable length. You should also mention that the cable was buried in the soil.
- Why is DAS capability labeled in Fig 1 but not mentioned in the introduction? If you do not show or discuss DAS data the label should be removed from figure 1 as it is not directly relevant to this study
- L124: “in the Arctic coastal plain”
- L130: you could clarify why the infrastructure you monitored with FODTS might be expected to impact local soil temperatures, ie “due to altered surface energy balance and soil thermal properties”
- L131: this sentence is quite long, please break it in two
- L136: the air temp at time of deployment is not directly relevant to this goals of this study
- L134: fiber-optic is typically hyphenated
- L136: ‘near the base’ would be better than ‘close to the bottom’
- The DTS deployment section mentions multimode fiber, but DAS is labeled along with FODTS in Fig1, and I believe DAS uses single mode fiber. Please explain.
- L159: the linear measurement spatial resolution (1 m) should be stated up front in this section. On a re-read I see it is listed in the second line, but not in a way that makes it clear to the reader the scale is 1-m, rather than referring to the meter scale more generally
- L186: Consider clarifying whether the model is site‑specific or generalizable to other permafrost locations
- L188: Is “Hysteresis implemented” or allowed to arise based on the physics included in the model. ie is the model deterministic
- L332: It is a stretch to say the 2-km FODTS cable surveyed “three distinct landscapes”. I believe it would be more accurate to indicate that distinct patches were monitored across one permafrost landscape impacted by some basic infrastructure.
- L364: Please remind the reader what specific infrastructure you are discussing here. Is 1.4 deg C the mean difference for when you had measured data?
- L372: “these dynamics”
- L374: this sentence about fast transition in temp during thaw could use a reword, I don’t follow the reasoning as stated
- L411: The continued referral to some minor dirt roads and a small structure as ‘civil infrastructure’ seems unnecessarily ambiguous for the reader. Consider just stating road/building, or whatever, more specifically. The concept of civil infrastructure brings a much more developed setting to my mind.
- Section 4.4 discussion of the accuracy of simulated soil temperatures could be reinforced by RMSE, correlation, or other quantitative metrics rather than relying on qualitative language
- L584: you have already defined the DTS acronym
- L592-97: Please add some references to previous studies here, and do some exploration as to whether the observed warming might primarily be do to shading and moisture changes, not the infrastructure, per se
Citation: https://doi.org/10.5194/egusphere-2026-1809-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 209 | 98 | 21 | 328 | 21 | 29 |
- HTML: 209
- PDF: 98
- XML: 21
- Total: 328
- BibTeX: 21
- EndNote: 29
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Great paper! Minor feedback.
Abstract: Concise, well written. Excellent to see such a long term study. Ready to read the paper.
86-88 – Well set up to describe the novel data set and analysis presented in this paper.
99-100 – “A defining feature…” citation?
102-105 – “Although hysteresis has been recognized in observational and modeling studies, it is often treated implicitly rather than incorporated explicitly…” – Any more recent citations on this than 2003? Modeling has changed a lot.
109-111 – “A 2-km fiber-optic DTS array recorded continuous ground temperatures from 2021 to 2024 across civil infrastructure, non patterned tundra, and ice-wedge polygon terrain.” – AWESOME
117-119 – “The model reproduces observed permafrost temperatures, fills observational gaps, and enables projections of future ground temperatures under climate-change scenarios both along the DTS transect.” This sentence is a little confusing. Suggest rewording.
140-144 – Confirming that 2 km of cable was installed one shovel width at a time?
144-145 – So 1km of cable was installed, but the fiber was spliced upon itself to make 2km?
160-164 – Sentence is a bit long and mixes present and past tense. Might read more easily as two sentences.
169-170 – Did you see drift? At 1-2km?
173 – How were the in situ reference baths performed?
176-177 – How often did the power interruptions and unstable sensors occur? What percentage of the data was removed? How much of the data was anomalous?
199-210 – so pressure only has a weak to moderate correlation and relative humidity negligibly correlates to wind speed.
216-228 – did you come up with these equations?
258-259 – “The effects of precipitation…” is there a citation, or was this your own analysis?
308-310 – How many points are represented here? How many are in a, b, and c for Oct-Dec 22/23, Jan-May 24, and May-Sept23. Why are the dates broken up like that? 3 months, 5 months, 5 months?
332-340 – definitely helpful to have clarity in 144-145 to interpret this section.
342-360 – In Figure 5, would be helpful to add thermometer readings / other validation to this somehow. In Figure 6, though a PDF can be zoomed in, it’s still challenging to see the time series, particularly the yellow civil thermometer and the gray/purple tundra thermometer. Relatively, it does not look like the tundra thermometer agrees with the DTS? The legend is a tad confusing, are there many tundra thermometers? I see a lot of gray lines and a dark line that doesn’t look like it’s gray for purple/blue. Suggest updating this figure. Would be great to understand it better!!!
364-369 – would be neat if you had a figure showing this geospatially.
371-382 – Different time periods selected than 308-310. These periods are explained in this section (excellent!). Leads back to the 308-310 question of why that time breakout? Figure 7 – not sure I understand the arrows. Freezing is happening below -5 and thawing above -2.5? Could something be thawing from -5 to 15? Could something be freezing from 0 down to -25?
397-400 – would be neat to know what the warmer spots (particularly in b) are attributed to.. could add a text box in the fig depending on the reason.
415-419 – nice!
421-423 – Excellent Fig. clearly shows impact of infrastructure.
447-450 – Figure 10 is important. Suggest overlaying horizontal lines / zones where the model has filled in the missing information. It would be easier for the reader to see how challenging it might have been for the model to match up with measured data.
467-476 – Figure 10, great fig
501-513 – Interesting. Is there any engineering work to point to that would make linear infrastructure less influential?
530-544 – would be nice to highlight a more temperature volatile period the model successful produced.
605-610 – Does this work continue? Can you project to 2030 and then cross check your predictions in a few years?
743-745 – Figure B2 is GREAT. Would be help to do something like this for 397-400
752-756 – Again – curious about why those time periods. Confusing figs