the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 18 May 2026
Alaska-COLD: Linking Surface Temperatures and Subsurface Thermal Dynamics in a Multi-Year Hourly Dataset From Interior and Northern Alaska
Abstract. We present the Alaska Coupled Observations of Land-atmosphere Dynamics (Alaska-COLD) dataset, comprising hourly air and soil temperature measurements from 12 sites across Alaska's north-central regions (64°–70°N). Each site measures air temperature and soil temperatures at four depths (0–0.7 m). Seven sites provide 714–726 days of continuous records (Summer 2023–2025) capturing multiple freeze-thaw (FT) cycles. We classify FT phenology into thawing, thawed, freezing, and frozen phases using surface temperature observations, enabling phase-specific analysis of thermal dynamics. We further derive FT metrics, including n-factors, degree-day totals, and damping ratios, to characterize surface energy exchange and subsurface thermal attenuation. The records reveal strong thermal damping with depth: surface-soil diurnal amplitudes range from 0.7–4.3°C, while the deepest sensors show only 0.004–0.047°C variation across the mean daily cycle. At the annual scale, ground-temperature ranges decrease from 11–25°C near the surface to 3–5°C at 45–75 cm depth, compared with air temperature ranges of 36–45°C. Thawed periods range from 39 to 92 days annually, and air-ground coupling vary strongly by FT phase, with R2 increasing from 0.02 during thawing to 0.69 during thawed conditions. A concise comparison with colocated ERA5-Land estimates shows stronger agreement for air temperature than for soil temperatures, highlighting the value of Alaska-COLD for resolving site-level ground thermal dynamics in permafrost environments. Alaska-COLD is accessible via: https://doi.org/10.5281/zenodo.17980271.
- Preprint
(10118 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 26 Jul 2026)
- RC1: 'Comment on essd-2026-213', Inge Grünberg, 06 Jul 2026 reply
-
RC2: 'Comment on essd-2026-213', Anonymous Referee #2, 13 Jul 2026
reply
Comments to the Authors
The manuscript describes a database of shallow soil temperatures measured to depths of 70 cm from 12 sites in Alaska. Additional data includes onsite meteorological data from two sites as well as air temperature for all sites. Statistical analysis has been used to characterize the data sets and to describe freezing, frozen, thawing and thawed periods. Some analysis is also presented on relationships between air and ground temperature and influence of snow cover.
The data set itself is useful, although most records are fairly short (max 2 years). However, with continued data collection and dissemination its value will increase, and it will continue to be of interest to permafrost scientists, ecologists and the modelling community. It also enhances and complements existing data available for Alaska.
The analysis presented however, is probably less useful given the short data records for some of the sites (< 1 year). The mismatch in data collection periods makes it difficult to combine data sets in the type of analysis that is presented. It is also not clear why global scale data sets are being relied upon for parameters such as bulk density, soil type, snow depth etc. Soil samples could have been collected during installation of instrumentation to determine organic layer thickness, soil type, moisture content and bulk density. A better description of subsurface conditions based on data collected at the sites would enhance the database and provided useful information for those using the database. On-site observations of vegetation cover could have been done rather than using regional products. Snow depth has been measured by the meteorological stations at two sites and snow depth and density were acquired during seasonal site visits. It is unclear why these data were not used as they would better reflect the local conditions.
Much of what is presented in section 4.2, which includes interpretations of results, has been discussed by others so there is nothing new being presented. No literature has been cited in this section, and it probably should be given there are some well known concepts being presented. Some suggestions have been provided.
The Introduction includes much background information on permafrost including a description of international monitoring networks. However, no data are collected from permafrost as the measurement depths are all above the permafrost table. The measurements made are not really relevant to the main indicator variables covered by the GTN-P. Much of the background information is unnecessary and can be removed. The authors should focus on existing data collection in Alaska and place their work in the context of these other studies.
The title refers to a multi-year dataset. However, that is a bit misleading given 5 of the sites have records of one year or less and the rest are slightly less than two years long. The title also refers to subsurface thermal dynamics, but it is more correct to refer to shallow soil temperatures given the maximum measurement depth is <0.5m for the majority of sites.
Additional Comments
L15 – General comment on the Introduction – This could be much shorter as it contains much background information that is not necessary. A few references could replace Fig. 1 and Table 1. The long discussion of international monitoring programs (GTN-P, CALM) is unnecessary given the information collected by the authors does not fit with the defined ECV indicator variables under these programs. Since the study focusses on sites in Alaska it would be better to limit the discussion of other monitoring efforts to Alaska and how any regional gaps are addressed by the new data collected.
L24-25 – This is general background and Fig. 1 is not necessary – a few references are all that is required. (the study only focusses on very shallow temperatures and does not consider all the aspects shown in the figure)
L30 – The relationship between air and surface temperature is required. The direct observations are used to characterize this relationship
L32-73 – This discussion is on both international and regional (Alaska networks). There are other regional networks in Canada, Scandinavia etc. that form the GTN-P. It would be better to focus on the data collection efforts in Alaska and how your data collection improves this (addresses gaps etc.). If Table 1 stays it should be revised to only include the work in Alaska.
L34 – CALM primarily compiles data on ALT. Near surface temperatures are used at some sites to determine ALT but it is not a primary product of CALM.
L43 – What IPCC and SWIPA reports are being cited?
L46 – “detect thermal trends across…” would be better
L53 – MAAT has not been defined yet.
L55-57 – Other studies have used daily temperatures to capture the interannual and seasonal variations. Sub-daily temperatures capture diurnal cycles.
L57-59 – Data collected at greater depths from these boreholes (i.e. below level of seasonal variation) are better for determining long-term trends.
L73-84 – The detail could be reduced here – just mention various types of data are collected remotely and in the field to assess environmental conditions and then describe the contribution of your study to this program.
L86 – The ground temperatures would be better described as near-surface or shallow temperatures.
L99-101 – Fig. 3 shows the distribution of sites and should be cited in the first sentence. Fig. 2 only shows 8 sites and it is unclear what the photos are meant to show – the instrumentation or examples of site environment/vegetation. The caption provides no useful information – if the goal is to show examples of instrumentation, then the type of instrumentation shown in each photo should be provided. The 2nd sentence is redundant if the figure reference is in 1st paragraph.
L102 – Isn’t it more correct to say that the sites capture different climates?
L103 – Ground ice content depends on the material types and also depositional environment and history. Latitude and elevation aren’t the main influence. “Vary sharply” does not describe how these things vary as there is more of a gradual transition that occurs. Snow cover is also influenced by vegetation
L108 – MAAT was used earlier but not defined – it should be defined the first time it is used.
L112-113 – Snow accumulation and vegetation height are related. Snow cover and vegetation modulate the heat transfer between the air and the ground surface.
L115-177 – You could refer to the photos in Fig. 2 to show the different vegetation types.
L118-119 – It would be better to integrate the reference to the tables into the preceding text (and probably the following text) where appropriate.
Table 2 – The information is more a description of location and the temperature measurements made rather than characteristics of the sites themselves.
Table 3 – The data source for soil type indicates that this is based on 250 m resolution map which isn’t that appropriate for site specific work. Were samples collected from the sites when they were instrumented? What about the organic layer thickness?
L121 – Is “meteorologic” a better term than “atmospheric”
L125 – Refer to figure (photo) that shows the equipment
L126 – What is the height of the air temperature measurement?
Table 4 – The excess ice contents given are for permafrost not for the active layer which is where the sensors are. Moisture content would be good to include (based on measurements at the site). The soil bulk density appears to be based on a reference that according to the title, is on predictive mapping so the values given are not site specific. The depth given is 100 cm which is below the depth of measurement for soil temperature (and may or may not be in permafrost which will differ from active layer). There were snow measurements made at two sites with the weather stations – why aren’t those values give? What is timing of the snow value provided (month, year) – is it the seasonal maximum?
L141 – At what height is the snow sensor mounted – this will determine the maximum depth that can be measured.
L159 – This is unlikely to be at the time of maximum thaw for many sites.
Table 5 – Both accuracy and resolution/precision should be given
L166 – High frequency/short-term variations occur at these depths, i.e. diurnal to seasonal variations occur. The longer-term damping occurs at greater depths and seasonal variation will eventually be damped out at the depth of zero annual amplitude.
Figure 4 – This isn’t that useful and could be deleted. The explanation in the text should be sufficient. Temperatures aren’t necessarily “stable” and they may still be changing during the frozen and thawed periods.
L170-189 – It would have been useful to determine the timing of complete freezing/thawing of the active layer rather than just the surface conditions as the timing will differ from the surface freezing/thawing.
L200-205 – Is the period used to determine the air and surface indices the same or do you consider that the period may be different (i.e. don’t have a fixed period and only consider the total freezing or thawing degree days based on the temperature data).
L202-209 – Be clear that it is the ground surface temperature that is used rather than the deeper ground temperatures that are used to determine the DD indices and n-factors.
L211 – General comment on section – Much of this is already known such as damping of temperature variation with depth (decrease in amplitude of wave) and the natural filtering of higher frequency variation with depth.
L215 – What is meant by thermal structure. Reference is usually made to the thermal regime.
L218 – Is this the mean air temperature? – be clear
L222 – In addition to damping there is also a phase lag with depth.
L226 – How do you determine if conditions are isothermal? What is the threshold value with respect to temperature variation? When determining the depth of zero annual amplitude for example we consider that the annual variation in temperature is <0.1°C. Are you considering the mean over the entire period here OR considering the variability that occurs on individual days because according to the data records the temperature does vary at the greatest depth at some sites more than 0.1°C during the day. Even if there is negligible diurnal variation, surface forcing is still having an effect at the deepest measurement depth because at all sites temperatures vary on the scale of a few days to seasonally (by more than 5°C). It is not really correct to say there is strong attenuation of surface forcing.
Figure 5 – The actual depths should be provided in a legend for each graph, so the reader does not have to search in the paper for them.
L250 – See comment above – temperatures are still changing so unclear what is meant by stable and predictable thermal regimes.
Table 6 – Give the units for mean temperature in the header for the table. What is Frozen Frac.? Over what period is FT Count given.
Figure 6 – Using colours for ranges in duration (i.e. have categories) is better then using a colour scale when using symbols – it is easier for the reader to see differences between sites. The map isn’t that effective given so many of the dots look the same.
L252 – Comment on entire section – Much of what is discussed in this section has been covered in existing literature, so it is surprising that that no literature has been cited when providing interpretations of the results. For freezing of the active layer, the snow cover is important as is the moisture content of the active layer. The timing of the onset of snow cover also plays a role. The zero curtain duration is often used in descriptions of ground freezing. Moisture/ice content will also influence the rate of thaw as well. Moisture/ice content is important because it controls the latent heat requirement. There have been other studies that have considered factors that influence ground freezing (including timing of active layer freeze-back), as well as relationships between freezing n-factors and snow cover and substrate conditions including Throop et al. (2012); Palmer et al. (2012); Riseborough and Smith (1998); Sladen et al. 2019; Smith et al. (2016). I suggest you consult these papers along with other relevant literature and reference where appropriate.
L266-268 – Other studies have considered this and also the duration of the zero curtain. Moisture content and snow cover are important influences (see some of the references provided in the comment above).
L276 – are your referring to the zero curtain?
L295 – This is the surface offset an effect that is well known and largely due to the effect of snow.
L295-313 – See comment above for relevant references. The impact of snow cover and moisture content influences winter ground temperature as has been shown by others. This includes the difference in range of air temperature and shallow ground temperature. Throop et al. (2012) consider ranges in air temperature and ground temperature using a similar graph to figure 9 – the paper should probably be consulted regarding interpretations etc.
L320 – The n-factor should be mentioned in this discussion.
L326-328- How is the thaw depth being determined – the measurements of soil temperature are shallower than these depths.
L329 – Is this air or surface FDD – be clear
L341 – It would make sense to compare with data from nearest meteorological station rather than the ERA data given the met station data is observational. The skin temperature isn’t the same as the surface temperature that you are measuring as it will be dependent on the vegetation conditions so not at the ground surface.
L361-365 – This should probably be in the methods section. When were sites visited?
L382 – The difference between the air and surface temperature is referred to as the surface offset. The thermal offset refers to the difference between the surface temperature and the temperature at the top of permafrost. See for e.g. Smith and Riseborough (2002).
L383-394 – Conifers can intercept snow resulting in colder ground conditions.
L397-398 – The lack of shade in combinations with wind scour in winter (limited snow cover) means that surface temperature will closely track air temperature.
L402 – Is Biskaborn et al. an appropriate reference here. Smith et al. (2022) comment on limitations of models including using global scale data to represent regional to site scale etc.
Fig A2 – snow depth doesn’t seem to be shown although it is measured at 2 stations (also during periodic site visits at all sites)
Table A1 – Isn’t it the surface temperature that is being considered in this table.
Data set – It would be better to give the actual depth of measurement as the column header rather than just a sensor number. Some of the sites do not have the depths in the correct order, i.e. not presented from shallowest to deepest. The location data etc. could also be provided as csv file.
Palmer, M.J., Burn, C.R., and Kokelj, S.V. 2012. Factors influencing permafrost temperatures across tree line in the uplands east of the Mackenzie Delta, 2004–2010. Canadian Journal of Earth Sciences, 49: 877-894. doi:10.1139/E2012-002
Riseborough, D.W., and Smith, M.W. 1998. Exploring the limits of permafrost. In Proceedings of Seventh International Conference on Permafrost. Yellowknife, Canada. June 1998. Collection Nordicana Vol.57, pp. 935-941.
Sladen, W. E., Wolfe, S. A. & Morse, P. D. (2019). Evaluation of threshold freezing conditions for winter road construction over discontinuous permafrost peatlands, subarctic Canada. Cold Regions Science and Technology, 170, 102930, 1-11. https://doi.org/10.1016/j.coldregions.2019.102930
Smith, M.W., and Riseborough, D.W. 2002. Climate and limits of permafrost: a zonal analysis. Permafrost and Periglacial Processes, 13: 1-15. https://doi.org/10.1002/ppp.410
Smith, S.L., Riseborough, D.W., Bonnaventure, P.P., and Duchesne, C. 2016. An ecoregional assessment of freezing season air and ground surface temperature in the Mackenzie Valley corridor, NWT, Canada. Cold Regions Science and Technology, 125: 152-161. doi:10.1016/j.coldregions.2016.02.007
Throop, J., Lewkowicz, A.G., and Smith, S.L. 2012. Climate and ground temperature relations at sites across the continuous and discontinuous permafrost zones, northern Canada. Canadian Journal of Earth Sciences, 49: 865-876. doi:10.1139/E11-075
Citation: https://doi.org/10.5194/essd-2026-213-RC2
Data sets
Alaska-COLD Aymane Ahajjam et al. https://doi.org/10.5281/zenodo.17980271
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 171 | 41 | 17 | 229 | 19 | 14 |
- HTML: 171
- PDF: 41
- XML: 17
- Total: 229
- BibTeX: 19
- EndNote: 14
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Review of "Alaska-COLD: Linking Surface Temperatures and Subsurface Thermal Dynamics in a Multi-Year Hourly Dataset From Interior and Northern Alaska" by Ahajjam et al.
The paper describes a dataset of air and soil temperature at 12 sites and additional meteorological variables at two sites at Arctic Alaska near the Dalton Highway. The data is of poor quality and not worth an ESSD publication. Please see figures and details below. Additionally, the data is short term and 5 of 12 stations failed with less than 2 years of data. This indicates that long-term operation is not possible with the equipment used and the added value to existing data networks is very small. The snow depth described in the paper is very error-affected and raw distance is not converted to depth. Furthermore, the statistics used to describe the data in the paper are misleading as sensors with different measurement periods are compared. E.g. the diurnal variation is different for sensors which failed without observing a summer period as compared to data series covering a complete year. Also comparing sensors at very different depths (between 30 and 70cm) named 'depth 4' is not helpful. Furthermore, the data is missing important metadata information. Except for (rounded) coordinates, all site information is derived from large scale maps or gridded products. This is not sufficient given the high local scale heterogeneity in the study region. To use the data, I would need local information on vegetation type and height, maximum thaw depth, surface cover (bare/moss/lichen/litter), soil type and moisture conditions and moss/organic layer thickness. The latter is particularly important as the sensor at the soil surface is actually buried below the moss/organic layer. A detailed description of what this means at each site is missing.
Detailed comments:
(1) Data publication on Zenodo:
'License' file is about software not data
metadata file: coordinates rounded too much to assess local vegetation. More metadata would be essential, like site vegetation (local, on site observations) and moisture, soil properties
'DateTime – timestamp (see metadata for time zone and format)': this is the metadata file, where else should I look?
Time zone not specified
'TCDT_C – additional temperature channel (°C), see metadata' Not clear, which metadata? This is contradictory to the statement in the paper of 'Snow depth (TCDT)' The abbreviation is not clear
Height of air temperature relevant and missing for all sites
data:
column names should have the depth at this location as it is different for all sites
date formatting: it would be better to use yyyy-mm-dd HH:MM
in some files, like site 11, the column order is different (first soil 1, second air)
low sensor quality is related with coarse resolution of 0.025°C, which is not ideal for monitoring freeze/thaw around 0°C and hinders the interpretation of diurnal cycles at lower depths (as done in the paper)
site 4: soil temperature 1 is clearly wrong as it exceeds air temperature on most summer days around noon
site 14: soil temperature 1 is clearly wrong as it exceeds air temperature on most summer days around noon
site 18: soil temperature 3 has a zero curtain of roughly 1°C and therefore does not satisfy the accuracy description in the paper
site 14: soil temperature 3 has a zero curtain above 1°C and therefore does not satisfy the accuracy description in the paper
site 11: soil temperature 3 has a zero curtain of 0.4°C and therefore does not satisfy the accuracy description in the paper
site 18: indications of rain/melt water intrusion along the installation leads to early warming peak
site 15: indications of rain/melt water intrusion along the installation leads to early warming peak
site 10: effects of installation not fully removed
site 3: physically impossible relative humidity > 100% (data >1000%)
site 6: physically impossible relative humidity > 100% (data >1000%)
site 6: physically impossible pressure
site 3 & 6: snow depth has many outliers and zero is not defined
I show these issues in quick graphs attached to the review.
Conclusion: the data is of poor quality and not quality controlled. Clearly erroneous measurements should either be removed or flagged.
(2) Paper:
general: brackets missing around citations
references should be hyperlinks already in the review stage to make reviewing easier
almost all fonts in figures are too small and the figure design is inconsistent
the full temperature time series should be shown (focus of the paper)
Instrumentation: The sites are equipped with comparably low-cost instrumentation which is not scientific standard
Abstract: According to Table 2, only one site has soil temperature sensors at 72cm while multiple sites have less than 40cm maximum depth. I find that the abstract and introduction over-sell the data with generic statements of "soil temperatures at four depths (0-0.7 m)". Furthermore, it is a bit hidden that some sites have less than a year of data. "Seven sites provide 714-726 days of continuous records (Summer 2023-2025)" please add the periods of the other 5 sites as the "12 sites" is prominent in the first sentence.
l 3: height of air temperature missing (and important)
l 24: 'governed by conductive heat fluxes' is a bit too simplified as latent heat is very important
Figure 1: This is common knowledge and not needed here. If the figure is kept, a citation is missing to some of the older papers showing the same schematic.
l 44: sentence confusing
Table 1: the MEB database has many Arctic air and soil temperature sites included although its scope is global: https://meb-network.com/access-to-the-database/
what is DOI in the context of DOI/GTN-P?
the Table is very Alaska-focused (with some global/polar examples). There are other regional networks, e.g. on Svalbard (Isaksen, K.; Lutz, J.; Sorensen, A. M.; Godoy, O.; Ferrighi, L.; Eastwood, S. & Aaboe, S., Advances in operational permafrost monitoring on Svalbard and in Norway, Environmental Research Letters, 2022, 17) and NW Canada (Kokelj, S. V.; Palmer, M. J.; Lantz, T. C. & Burn, C. R. Ground Temperatures and Permafrost Warming from Forest to Tundra, Tuktoyaktuk Coastlands and Anderson Plain, NWT, Canada, Permafrost and Periglacial Processes, 2017, 28, 543-551 and Throop, J.; Lewkowicz, A. G. & Smith, S. L. Climate and ground temperature relations at sites across the continuous and discontinuous permafrost zones, northern Canada, Canadian Journal of Earth Sciences, 2012, 49, 865-876 ). Either include other networks or specify in header and text that you study Alaska only.
l 67: "most existing networks operate at relatively coarse temporal resolution" that is a bold statement and not true from many stations e.g. contributing to GTN-P
l 80: no need to cite so many single LiDAR datasets, not relevant
introduction in general: a bit lengthy, could be shortened. The figure is not relevant and could be replaced by a citation. The table could be moved to the appendix.
Figure 2: Better show one picture for each station (12 instead of 8). The pictures should be more indicative of the landscape than picture of site 7. Site name and latitude would be more helpful than site Id
Figure 3: most fonts too small, in particular the scalebars and the source reference (which should go into the caption). The site IDs are missing. It would be good to combine with Figure 2 to highlight where the pictures are from
Table 2 and 3: The order is really confusing (to an out-side person the IDs are arbitrary). Please order south-north or north-south as you describe the gradient.
Table 3: I'm missing site details not derived from published maps but as seen on site. E.g. "TT (LS or H)" reading "Tussock Tundra (Low Shrub or Herbaceous)" does not help much as these vegetation types are very different in terms of thermal insulation properties. Also "Soil type (at 100m depth)" is not really relevant for measurements within the active layer. Other important pieces of information are missing, such as active layer thickness/maximum thaw depth or surface cover (e.g. presence and thickness of mosses)
Table 4: The slope column is reasonable, but all the other columns are not helpful at all. Snow density is highly variable at a single site within one season and has a strong local variability. Giving a single value is not helpful and, I find, not correct. Also, ERA-5 may not be the best source of information on snow density. The same is true for snow depth and albedo.
ll 157-167: Installation along a pipe has several disadvantages which need to be discussed: (1) preferential water flow along the pipe can transfer heat into the ground (I think you can see this in the data of sites 15 and 18), (2) long-term operation is not possible as the pipe moves out of the soil during freeze-thaw cycles, (3) heat conduction along pipe material and cables. It is better to dig soil pits and install temperature sensors in the pit wall before closing the pit carefully.
l 165: "We note that the upper probe (nominally 0 cm) records near-surface temperature beneath the moss or organic layer,..." This is highly relevant and needs to be described in detail for each site including particularly the thickness of the moss or organic layer. Also the properties like peat moss vs. other moss, litter vs dead moss is critical.
Table 5: separate table into the part of the climate stations (Campbell) and the Hobo stations and add site names the cells "ClimaVUE 40 / S-TMB-M002" are confusing as these are two different sensors not used at the same station I suggest 5 columns: Variable | Sensor Campbell | Accuracy Campbell | Sensor Hobo | Accuracy Hobo and additional rows with 'Installed at station' and 'Datalogger'
l 174: the thresholds are not adequate if the sensor accuracy is 0.2°C at best
ll 179f: Sentence double
Section 3.1: This has been done by many other scientists before. Please cite them.
l 219: "reaching above freezing during afternoon hours." Not clear here which day you describe. Or the mean of which period. If the period is not exactly a multiple of one year, the mean data is screwed (like summer warm temperatures can be under represented and the 10 Aug 2023 - 08 May 2024 mean diurnal cycle is not comparable with a 11 Jan 2025 - 29 Jul 2025 mean or a 23 Jul 2024 - 28 Jul 2025 mean)
Figure 5: not helpful as different periods are considered and the sensors are at different depths; fonts too small, depth and period need to be specified in the panels of each site
l 235: "maximum air temperatures exceed 30°C" this may be an issue of the low-cost sensors heating up under direct sunlight
Table 6: hart to compare, at least the periods should be specified again for each site, site order should be changed as in all tables
Figure 6: blurry, fonts too small, color scale not helpful as differences are hard to see, no need to include >365 in the range for the color bar if the maximum value is 254 days. The caption should state why not all sites are included
Figure 7: Fonts way too small, groups of sites should be highlighted with color
Figure 8: I do not understand the 'cold/warm permafrost' labels. The paper does not present any permafrost measurements (only active layer); label and legend fonts too small
Figure 9: fonts too small
ll 314-319: This may partly be due to the moss/organic layer above the top sensor. This issue should be quantified and measured using a top-of-moss/organic layer measurement
l 374: "post-processing quality checks" these should be described in detail. Looking at the data, I am skeptical.
l 400: "potentially thin cryptogamic layers": this should be measured not speculated
l 402: reference not fitting here
ll 421f: limitations also include missing site metadata such as moss thickness, maximum annual thaw depth, vegetation height,...
Figure A2a&b: the figure highlights poor data quality with many physically impossible outliers not removed (Pressure, relative humidity, distance)