the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Measuring the spatiotemporal variability of snow depth in subarctic environments using unmanned aircraft systems (UAS) – Part 2: Snow processes and snow-canopy interactions
Leo-Juhani Meriö
Anssi Rauhala
Pertti Ala-aho
Anton Kuzmin
Pasi Korpelainen
Timo Kumpula
Bjørn Kløve
Hannu Marttila
Abstract. Detailed information on seasonal snow cover and depth is essential to the understanding of snow processes, operational forecasting, and as input for hydrological models. Recent advances in unmanned aircraft systems (UASs) and structure from motion (SfM) techniques have enabled low-cost monitoring of spatial snow depth distribution in resolutions up to a few centimeters. Here, we study the spatiotemporal variability of snow depth and interactions between snow and vegetation in different subarctic landscapes consisting of a mosaic of conifer forest, mixed forest, transitional woodland/shrub, and peatland areas. To determine the spatiotemporal variability of snow depth, we used high-resolution (50 cm) snow depth maps generated from repeated UAS-SfM surveys in the winter of 2018/2019 and a snow-free bare ground survey after snowmelt. Due to poor sub-canopy penetration with the UAS-SfM method, tree masks were utilized to remove canopy areas and the area (36 cm) immediately next to the canopy before analysis. Snow depth maps were compared to the in-situ snow course and a single-point continuous ultrasonic snow depth measurement. Based on the results, the difference between the UAS-SfM survey median snow depth and single-point measurement increased for all land cover types during the snow season, from +5 cm at the beginning of the accumulation to -16 cm in coniferous forests and -32 cm in peatland during the melt period. This highlights the poor representation of point measurements even on the sub-catchment scale. The high-resolution snow depth maps agreed well with the snow course measurement, but the spatial extent and resolution of maps were substantially higher. The snow depth variability (5–95 percentiles) within different land cover types increased from 17 cm to 42 cm in peatlands and from 33 cm to 49 cm in the coniferous forest from the beginning of the snow accumulation to the melt period. Both the median snow depth and its variability were found to increase with canopy density; this increase was greatest in the conifer forest area, followed by mixed forest, transitional woodland/shrub, and open peatlands. Using the high spatial resolution data, we found a systematic increase (2–20 cm), then a decline of snow depth near the canopy with increasing distance (from 1 m to 2.5 m) of the peak value through the snow season. This study highlights the potential of the UAS-SfM in high-resolution monitoring of snow depth in multiple land cover types and snow-vegetation interactions in subarctic and remote areas where field data is not available.
- Preprint
(2174 KB) -
Supplement
(1179 KB) - BibTeX
- EndNote
Leo-Juhani Meriö et al.
Status: final response (author comments only)
-
RC1: 'Comment on tc-2022-242', Anonymous Referee #1, 11 Feb 2023
General comments
Authors analysed the spatial and temporal variability of snow depth and interactions between snow and vegetation in a subarctic landscape with coniferous forest, mixed forest, and peatland areas. To determine the variability of snow depth, they used high-resolution maps acquired from four UAS surveys in a one winter season verified by manual snow course measurements and one automatic snow depth station. Authors used interesting approach of creating a tree-mask to remove canopy areas from analysis due to poor penetration of the UAS camera. Authors found that both snow depth and its variability increased with the canopy density. Authors also described the snow depth increase and then decline with a distance from canopy, as well as the increase of the peak value distance from tree as the season progressed.
In my opinion, authors did an interesting work which certainly has a scientific relevance. I think this is an important pure and thorough field study. Therefore, the study has clear potential to be published. However, I do not see much novelty in the study both in terms improving our knowledge or methodological approaches. I see some methodological novelty in using the tree mask to limit the data, but the question is whether it is really novel. I am not saying that the study lacks novelty at all, but I think the authors should better define what is new in the study and how it goes beyond the previous studies. Besides, I have a few other comments listed below, which should be addressed before I can recommend the manuscript for publication.
Major comments
I did not fully understand why authors used CORINE land cover data since they worked with precise UAV based data describing the specific pixel distance from the canopy/trees (which were used to create canopy masks). Maybe I just did not understand it correctly from the text, but why they did not use accurate canopy structure data for the whole analysis? Or was CORINE data used only for general description of the land cover types in individual plots? Please explain it in more detail (probably in methods).
Result section 4.3 contains three figures; however, the related text doesn’t contain detailed explanation and interpretation (it consists only in three short paragraphs). Please extent the related text substantially to provide the reader with detailed description and interpretation of the related figures.
In my opinion, conclusion section is too brief and general. I would suggest including more details (including numbers) about individual conclusions. As it is now, it looks like a summary describing what authors did rather than main study conclusions.
Specific comments
L 21: One of the study conclusion is that differences between UAS and ground measurements highlights “the poor representation of point measurements even on the sub-catchment scale”. This might be certainly true; however, this may also show that point measurement location is not fully representative for the wider area. Could you please add some more discussion related to this issue?
L 29-31: Authors stated that “This study highlights the potential of the UAS-SfM in high-resolution monitoring of snow depth in multiple land cover types …”. UAS is nowadays standardly applied and well-established method for snow depth monitoring (even in diverse vegetations). Therefore, the statement that “it has a potential” might be relevant perhaps 5-7 years ago, but not nowadays. Please consider reformulation.
L 44-45: Although I agree that individual factors control snow depth at different spatial scales, I do not think that such distinct limit (100 m) can be defined. Maybe consider reformulation.
L 79-87: Here authors explain the novelty of their study. Besides others, authors see the novelty in applying UAS imaging in boreal regions. Why this is specifically novel? How the UAS imaging in boreal regions differs from imaging in other areas? I think that application of UAS in boreal regions just because it was never used there before, doesn’t mean novelty per se. Please consider more specific explanation.
L 85: One of the research questions is how UAS can be used for snow depth imaging during poor light conditions (probably because the study plots are far beyond the artic cycle). Authors addressed this question rather marginally in Section 5.4, but maybe this might be one of the novel issues which might deserve more attention (see also my general comment and the previous comment).
Fig. 2: Lines with min/max snow depth means snow depth evolution of the one winter season with highest/lowest snow depth or each date on x-axis means maximum/minimum value for this date from all winters at the study period? Please clarify.
L 123-124: Could you be a bit more specific why two of the surveys were discarded?
Table 1 (and maybe also Table 2 and 3): Consider adding also absolute values of snow depth and not only differences between point measurements and UAS data.
Fig. 6: What is the physical explanation of increasing differences of the snow depth near canopy with progressing season? How important is the longwave radiation emitted by trees which increases the snowmelt rates near tree trunks? Please discuss shortly.
L 338: While I generally agree with provided explanation of highest snow depth in forested areas, do you have any data to support this interpretation?
L 345: While this is rather trivial conclusion, I think it might be beneficial for end users (e.g., operational services) and thus it may appear also in the conclusion section.
L 357-361: I see the point, however, why it should be interesting? Could you explain it in more detail?
Technical corrections
L 65: Instead of “submitted to the same journal”, I would specify its name.
Fig. 2: Please consider change of individual line colours/types to increase readability.
L 362: “For open peatland landcover, this peak may be explained …” It is not clear what “this” refers to. Please consider reformulation.
Citation: https://doi.org/10.5194/tc-2022-242-RC1 - AC1: 'Reply on RC1', Leo-Juhani Merio, 13 Apr 2023
-
RC2: 'Comment on tc-2022-242', Anonymous Referee #2, 16 Feb 2023
General Comments
The authors present a comparison of snow depths observed using UAV-based structure from motion. They explore the relationship of snow depth through a snow accumulation and melt season across various land cover types within a sub-arctic environment, paying special attention to the interactions between forest canopies and snow depth. The main comparison techniques include presenting differences between UAV-SfM observations and a single observing station, between land cover classes, and based on the distance from the canopy. I believe the time-series component mixed with both the spatial coverage and diversity of land cover types makes this work a valuable addition to the literature. I also agree with Reviewer 1 in that the tree masking was an interesting component and additionally think that the exploration into snow depth as a function of canopy distance was particularly informative. The use of a summary figure at the end of the manuscript was also valuable.
With that being said, I do believe there is also substantial room for improvement. The objectives stated in the introduction did not seem to align with the presented results. Also, while the dataset is impressive in its own right, the analysis and statistical methods were unclear or questionable at times. I have included some specific suggestions outlining where improvements can be made within the following sections. I think the paper is novel enough to warrant publishing with some revision.
Major Suggestions/Comments
L84-87: These objectives don’t seem to line up with what was addressed in the study. The first point seems to be the focus of the accompanying paper (not this one). The second and third points should be kept. Though, consider adjusting them to be more in line with the actual analyses done (1 – comparing spatiotemporal snow depth variability across different land cover types, and 2 – exploring the controls canopy has on this variability). There is also a large component in the way the results are presented that presents all observations relative to the point observation. Since assessing the spatial representativeness of the single-point site is such a focus of the analysis approach, I suggest adding a clear mention of this within the objectives as well.
The word ‘variability’ is used throughout when referencing the difference between the 5th and 95th percentiles of the snow depth distribution. This serves more as a measure of the range, not variability (like standard deviation/variance). Please revise your use of the word ‘variability’ throughout (or update the statistical method to better reflect variability). In most cases, it could be replaced with the word range.
L230-244: This section and potentially the statistical approach should be restructured. Initially you mention that all classes are significantly different with high confidence (very low p-value), then process to counter this claim when using the smaller random samples of snow depth data. What is the takeaway here? I suggest selecting a single appropriate statistical test and sticking with it.
With such a considerable focus on comparisons between the ultrasonic sensor at Kenttarova and the UAV-SfM observations, there needs to be a more comprehensive discussion as to the land cover surrounding this site (i.e., distance from canopy, understory, Corine class canopy type etc.).
Can the vegetation classification (using Corine) be enhanced by using ortho mosaic data & your tree masks? As is, the resolution is somewhat limiting, and it is difficult to tell how effectively this captures the different canopy types. There would be considerable value in adding forest type & density information into the analysis, without adding much additional work.
Minor/Technical Suggestions
L24: First instance of using variability in place of range. Please adjust the terminology here (and throughout)
L30-31: This point should be modified to reflect the fact that even if there is field data (collected in a classic way through point sites/snow courses) snow analyses are still limited. Doing so would make this more in line with the points made in the discussion and conclusions later in the paper.
L47-49: “In forests,….” This sentence is a bit challenging to read, please consider revising
L59: remove ‘the scale of’, is redundant
L68: Please specify more clearly that a more comprehensive review of UAV-SfM studies is included in the accompanying paper
L71: For the likely audience of this work, an explicit definition of tree wells is unnecessary
L80: Just because the region is locationally different doesn’t necessarily make this work different. Please mention some of the unique considerations (like lighting, forest structure, snow properties) that make the subarctic region a unique study area.
L113: Please clean this up a bit, the use of parenthesis is excessive and challenging to associate the numbers mentioned with the cited works
L149: I assume ‘high-quality and moderate depth filtering settings’ are specific to the software used. Can you please provide a reference to what these parameters mean, or briefly add description herein
L165: You introduce DoDs here, then quickly shift to referring to them as snow depth maps. Am I correct that they are the same? If so (or if not) make sure it is clear that they are interchangeable terms (or not).
L178: What does ‘i) approved vegetation classification’ mean in this context?
L199: Please re-iterate that the ‘point site’ refers to the single automated station located within the forest
Table 1: Please clearly indicate which depth is subtracted from which in the figure caption. Also, while presenting only the differences between depths at the courses and to the observing site is interesting (and relevant to the spatial representativeness question), I think this table would greatly benefit from the inclusion of the actual median depth values by class. For example, include the median depth followed by the difference relative to the point site (i.e., 55 cm (-10 cm))
L216: Please clarify ‘point measurements’ – is this plural to represent the time-series at the single observing site, or does it refer to the snow course obs. Please try to make this clear throughout the paper.
Table 1 -> Section 4.2 (and other locations): Please make sure your use of units is consistent (pick m or cm)
L219-L223: I was confused by the statistics here. Why not just present the median depths?
L224: “difference in variability” (?)
L225: ‘difference’ compared to what? The point reference?
Table 2-3: Again, in my opinion, there is an unnecessary added layer of complexity here. I suggest presenting the true depth values (5-95%) within each field and timestep, then presenting the observations for the same timesteps at the point site & the range/median of the snow courses. One idea… a time-series figure (or 1 per plot) with the data bounds may be a good additional way to visualize changes occurring across each class (& their relationship with the observations). -> it also would capture similar information to what is shown in Figure 5
Figure 5: This is a great figure, please ensure it is referenced/discussed sufficiently in the text
Line 263: Again, variability needs to be clearly defined
Figure 6-8: Since these plots are showing similar information, I think their number can be reduced. It is not clear to me what Figure 8 adds to the paper. Consider revising.
Section 5.1: This section is well written but lacking a bit in terms of the actual findings regarding spatiotemporal variability during the accumulation and melt season. Consider adding to the discussion using relevant statistics from the paper relevant to the spatiotemporal variability
L308, 310: careful using ‘variability’ here
L313-314: sentence here is a bit wordy
L326: Note the magnitude of the variability/range
L342: Do you mean ‘similar’ types here? Or ‘different’?
L344-345: Snow courses are good, but I think it would be useful to reiterate that they are limited in their ability to describe the types of canopy interactions observed in this work (10’s-100’s of obs, vs. thousands)
L365: ‘….by the limited canopy effect,…’
L380: errant ‘(‘
L388: completely up to the authors, but I think a more reasonable range of magnaprobe survey observations is in the 1,000-10,000 range. 100,000 seems a bit high from my experience. This also helps to make the value of the UAV-SfM clearer
L411: flip local & medium
L419: consider removal of ‘illumination conditions’. While mentioned in the part 1 paper, this does not seem to be something addressed in this manuscript
Citation: https://doi.org/10.5194/tc-2022-242-RC2 - AC2: 'Reply on RC2', Leo-Juhani Merio, 13 Apr 2023
Leo-Juhani Meriö et al.
Data sets
Unmanned aircraft system (UAS) snow depth mapping at the Pallas Atmosphere-Ecosystem Supersite Rauhala, A., Meriö, L. J., Korpelainen, P. and Kuzmin, A. https://doi.org/10.23729/43d37797-e8cf-4190-80f1-ff567ec62836
Leo-Juhani Meriö et al.
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
368 | 112 | 16 | 496 | 37 | 5 | 5 |
- HTML: 368
- PDF: 112
- XML: 16
- Total: 496
- Supplement: 37
- BibTeX: 5
- EndNote: 5
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1