the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
A low-cost, autonomous system for distributed snow depth measurements on sea ice
Abstract. Snow is a critical component of the Arctic sea ice system. With its low thermal conductivity and high albedo, snow moderates energy transfer between the atmosphere and ocean during both winter and summer, thereby playing a significant role in determining the magnitude, timing, and variability of sea ice growth and melt. The depth of snow on Arctic sea ice is highly variable in space and time, and accurate measurements of snow depth and variability are central to improving our basic understanding, model representation, and remote sensing observations of the Arctic system. Our ability to collect those measurements has hitherto been limited by the high cost and large size of existing autonomous snow measurement systems. We designed a new system called SnoTATOS (the Snow Thickness and Temperature Observation System) to address this gap. SnoTATOS is a radio-networked, distributed snow depth observation system that is 95 % less expensive and 93 % lighter than existing systems. In this manuscript, we describe the technical specifications of the system and present results from a case study deployment of four SnoTATOS networks (each with ten observing nodes) in the Lincoln Sea between April 2024 and January 2025. The study demonstrates SnoTATOS’ utility in collecting distributed, in situ snow depth, accumulation, and surface melt data. While surface melt varied within each network by up to 38 %, mean surface melt between networks varied by only up to 9 %. Similarly, whereas initial snow depth varied by up to 42 % within each network, a comparison of mean initial snow depth between networks showed a maximum difference of only 26 %. This indicates that floe-scale measurements made using SnoTATOS provide more representative data for regional intercomparisons than existing single station systems. We conclude by recommending further research to determine the optimal number and arrangement of autonomous stations needed to capture the variability of snow depth on Arctic sea ice.
- Preprint
(1553 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-187', Anonymous Referee #1, 23 Mar 2025
In this manuscript, the authors describe the design, validation, and deployment, of a low cost system for making distributed snow depth measurements. I have been working on similar systems that follows a similar design strategy, though the components for snow depth measurements are still a work in progress - congratulations to the authors on being first with a formal preprint and hopefully soon paper on this topic! I think that this is a very interesting and well done work, that it will be very useful to the community. Therefore, I am generally very enthusiastic of the work and supportive of publication. I have a few minor comments that I would like the authors to consider, and I think that, once these are addressed, this work should be ready for publication.
- I think that the introduction is interesting and well written. A couple of possible points to strengthen the claims for the value of these measurements. i) These measurements are very useful together with temperature profile measurements, since, as the authors point, the snow is an effective thermal isolation, so that it plays a large role in the temperature profiles obtained. The discussion on this aspect could be slightly extended. ii) there can be significant levels of snow accumulation, which is also adding a "perturbation" to some other measurements. For example, a 1m, or 1.5m, or 2m temperature and / or wind measurement is not any longer done at the same initial height if either 30cm of snow fall, or 30 cm of snow melt, or snow + ice melt takes place. Therefore, having good data on this aspect is also helpful for the accuracy of other kinds of measurements.
- The discussion about the cost savings and methodology are well presented. A couple of additions on this aspect that the authors can consider. Now, the SIMB3 buoy is the key expensive component and may be a major cost driver. Note that it is possible to build also low cost "main instruments", and that using for example iridium modems has become much easier: see, for example, the developments that are happening in the "small buoys" field in several groups around the world: e.g., https://doi.org/10.1016/j.coldregions.2019.102955 , https://doi.org/10.3390/geosciences12030110 , https://doi.org/10.1080/21664250.2023.2249243 , https://doi.org/10.1080/21664250.2023.2283325 , as also reviewed in e.g. https://doi.org/10.48550/arXiv.2410.07813 , and possibly additional similar developments that I may not be aware of. I know (from my own work) that it is easy to interface such solutions with a LoRa modem and / or microcontroller. This means that a further large cost saving could be obtained by also reworking the SIMB3 solution, substituting it by a low cost buoy similar to the ones discussed above. Given the cost of the SIMB3 buoy, this may allow to further drastically reduce the total cost of the system, and allow true larger scale deployment.
- The authors use a MCU (AVR family) + a LoRa modem. Naturally, this is perfectly fine. But I would like to recommend to the authors to consider solutions such as the STM32WL family of MCUs for their future work, that combine the MCU (actually, a significanlty more recent, more power effective, and more powerful one than the "old" 8bit AVR one used here) and the LoRa modem, on a single chip. Having worked both with solutions involving MCU + Lora modem, and solutions involving the integrated STL32WL chip, I can warmly recommend the second one - it is just overall more robust, lower cost, and better to work with; in addition, the stm32duino library support is excellent. But naturally, this is only a minor technical recommendation that the authors may or may not follow in their future design - this is just a minor tips from my experience working with similar solutions.
- The choice of the power solution (NiMh batteries) is interesting and a bit surprising to me. Again, "if it works it works", and I do not ask the authors to do any additional work, so this is not a criticism, just sharing my experience. One usually gets significantly better energy density and performance in the cold using primary, non rechargeable Li batteries, such as (there are other suppliers too selling similar batteries) Saft LSH20 or Saft LS33600 batteries (these are also available in different form factors and capacities), depending on the peak power requirement. In the case when no solar panel is attached to the small modules, and they should survive low temperatures (both of which seem to be the case here), this can be an interesting solution to consider in the future.
- It would be useful to gather all the power consumption data also in a table, and not have these only in the text - this is easier to read for technical people interested in getting a quick overview of the technical facts.
- The system is described in quite some details, and the main lesson I "take home" is that i) this works reliably, ii) the sensor suggested is probably a good choice (I have been looking at another sensor, but I will investigate this one too in the future). With this information, it will be easy for me to build my own system that is very similar to this one. However, there is no reason to redundantly repeat work across groups. Therefore, my question: it does not seem like you are offering this as an open source project? Would you be able to do so, and to provide mechanical and electrical CAD files and source code on for example a github repository or similar, together with detailed assembly and programming instructions? This would increase a lot the impact of your work and save time and money to the community as a whole.
Apart from these minor points, I think that the study is very well done and reported, and I have no major comments.
Citation: https://doi.org/10.5194/egusphere-2025-187-RC1 -
RC2: 'Comment on egusphere-2025-187', Anonymous Referee #2, 25 Mar 2025
The paper presents an innovative extension of the autonomous snow depth measurements. The presented method finally delivers a feasible possibility of long-awaited expandable spatially distributed snow depth sensors that is applicable to sea ice and terrestrial environments. The method is energy-efficient and low-cost. The prototype deployment provided convincing evidences that the method supplies good data. I am very supportive of the author’s work and encourage the journal to publish their work after a few major improvements:
-
The method is presented as an extension to an existing autonomous system SIMB3 – which is costly, heavy and requires relatively long deployment time. The authors suggests that their extension can be modified into a stand-alone system. I would like this paper to include more information on this second option. Most of the stand-alone system particularities are very briefly mentioned or even hidden in the captions.
-
The authors make a claim about the representativity of the network (nod size) for the mean snow conditions on the sampled sea ice floes. This should be better supported by comparisons from the spatially distributed in-situ observations (snow lines, Magnaprobe transects, stake fields) taken during the deployments or (as statistical analysis) at previous seasonal sea ice drifts (SHEBA, N-ICE, MOSAiC).
I also suggest some minor and technical improvements:
-
line 16: 95 and 93% is very exact (and therefore likely fast outdated), I suggest to modify this to something like ‘nearly an order of magnitude’
-
line 24-25: This same claim is repeated then two more times in the paper. Can you at one of those places (discussion) elaborate this a but better: how?
-
Line 27-28: here and elsewhere, citations must be ordered chronologically, likely this will be handled by technical editing
-
Lines 39-41: This is known from the times of SHEBA (see Matthew Strum et al, 2002, doi:10.1029/2000JC000400). Also recently pointed out by Itkin and Liston in this journal (https://doi.org/10.5194/egusphere-2024-3402)
-
Lines 55-57: please double check the years of this experiments: N-ICE was from January 2015 to June 2025, MOSAiC was from October 2019 to October 2020.
-
Line 58: Here you can also add the Sever data: National Snow and Ice Data Center. (2004). Morphometric Characteristics of Ice and Snow in the Arctic Basin: Aircraft Landing Observations from the Former Soviet Union, 1928-1989. (G02140, Version 1). Romanov, I. P. (Comp.) [Data Set]. Boulder, Colorado USA. National Snow and Ice Data Center. https://doi.org/10.7265/N5B8562T. [describe subset used if applicable]. Date Accessed 03-25-2025.
-
Table 1: I would like to see SnoTATOS in two versions here: as extension of SIMB3 and as stand-alone. Also, consider adding IMB buoys developed by bruncin.com. Also, are there no comparable terrestrial systems for snow depth measurements?
-
Lines 125-126: I don’t understand the statement ‘while conforming to less-than-truckload…’ Please, explain how exactly this is relevant. Are you talking about how much weight the ice can bear and how you calculate it relative to long/heavy the nod is? It might e useful to return to this claim in your discussion about the Lincoln Sea case. How it it work there? You did not revisit, but do you have any assumptions based on the data? Did you do any experiments (‘in the yard’, maybe on a local lake) where you could observe this?
-
Figure 3: similar to above comment: The last part of the caption belongs to the main text. If increase of omega with time (melt) all the to the 35 does not matter, this is convincing. How did you measure that?
-
Line 171: Russian Federation not Russia
-
Line 226: replace reliable by durable?
-
Lines 230-231: Maybe reformulate to be more clear to non-expert reader: theoretical limit for the installation on a perfect flat surface and sensor 1 m over the ground would yield in 10-20km range. Maybe say something about how high is the stake and how high are normally the pressure ridges.
-
Line 233: Still, spend a sentence on what is mesh network: I assume nods relay the information…
-
Line 239: Is this a good place to explain the stand-alone option (or give reference to a different section, appendix)?
-
Line 251: ‘hanging’? Is this really how we say it or is it just colloquial expression?
-
Figure 5: Last sentence of the caption deserved to be in the main text and expanded.
-
Line 266: ‘it pulls the CS line to ground’ – I don't understand.
-
Lines 281-282: Can it be deployed on terrestrial systems?
-
Line 306: ‘mean installation conditions’: Can you explain how did you collect these data? Was this just one snow depth under each sensor or did you also a floe survey?
-
Figure 8: Why 10-day average?
-
Lines 348-353: Can you compare this also to the variability at SHEBA, N-ICE and MOSAiC floes?
-
Line 371: Can such difference be justified by looking into weather reanalysis or sea ice deformation (leads) data?
-
Line 383-384: Can you give some specific recommendations on the further research and experiment design?
-
Lines 404-408: This is a valuable motivation to develop such system in the first place. Is worth saying this already in the Introduction?
Otherwise the paper is very well structured, concise, understandable, well-written and generally easy to read.
Citation: https://doi.org/10.5194/egusphere-2025-187-RC2 -
Data sets
SnoTATOS snow depth and surface melt observations from the 2024 ARCSIX campaign Ian Raphael, Donald Perovich, and Christopher Polashenski https://doi.org/10.18739/A2RX93G1Q
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
113 | 53 | 7 | 173 | 10 | 13 |
- HTML: 113
- PDF: 53
- XML: 7
- Total: 173
- BibTeX: 10
- EndNote: 13
Viewed (geographical distribution)
Country | # | Views | % |
---|---|---|---|
United States of America | 1 | 73 | 39 |
Germany | 2 | 19 | 10 |
Norway | 3 | 16 | 8 |
China | 4 | 11 | 6 |
Finland | 5 | 7 | 3 |
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1
- 73