the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Variability and drivers of winter near-surface temperatures over boreal and tundra landscapes
Vilna Aleksandra Tyystjärvi
Pekka Niittynen
Julia Kemppinen
Miska Luoto
Tuuli Rissanen
Juha Aalto
Abstract. Winter near-surface temperatures have important implications for ecosystem functioning such as vegetation dynamics
and carbon cycling. In cold environments, seasonal snow cover can exert a strong control on the surface temperatures.
However, the lack of in situ measurements of both snow cover and surface temperatures over high latitudes has made it difficult
to estimate the spatio-temporal variability of this relationship. Here, we quantified the fine-scale variability of winter
near-surface temperatures (+2 cm) and snow cover duration using a total of 441 microclimate loggers in seven study areas
across boreal and tundra landscapes during 2019–2021. We further examined the drivers behind this variation and the extent
to which surface temperatures are buffered from air temperatures during winter. Our results show that while average winter
near-surface temperatures stay close to 0 °C across the study domain, there are large differences in their fine-scale variability
among the study areas. Areas with large topographical variation, as well as areas with shallow snowpacks, showed the greatest
variation in near-surface temperatures and in the insulating effect of snow cover. In the tundra, for example, differences in
minimum near-surface temperatures were close to 30 °C. In contrast, flat topography and deep snow cover lead to little spatial
variation and decoupling of the near-surface and air temperatures. Quantifying and understanding the landscape-wide variation
in winter microclimates improves our ability to predict the local effects of climate change in the rapidly warming boreal and
tundra regions.
Vilna Aleksandra Tyystjärvi et al.
Status: open (until 14 Jun 2023)
-
RC1: 'Comment on egusphere-2023-576', Jonathan von Oppen, 08 May 2023
reply
General comments
The authors investigated winter temperatures near the ground surface and how they related to topography, vegetation, and snow cover, across several boreal forest and tundra landscapes. The study addresses an important topic, as winter microclimate has long been neglected despite its importance for Arctic plant biodiversity and potential crucial carbon feedbacks from soils. While overall not surprising, the findings add important evidence on terrain-snow-temperature relationships in boreal forest and tundra.
Overall, the manuscript is well written, the study was conducted thoroughly and used considerate methodology, and the authors generally communicate their findings in an appropriate way. I do, however, see potential for improvement, as I find some methodological choices insufficiently justified and described. Despite repeatedly referring to spatial variation at the landscape level, this variation is not presented or analysed for how specific spatial patterns lay out across the landscapes studied, which could add another important insight. Centrally, I suggest the authors to revisit the choice and check for consistent use of specific terminology, such as for temperature variables or study locations, to avoid confusion on the readers’ side. Also, I think the impact of the study could be improved if including not only near-surface, but also below-ground temperatures, which are readily available from the dataset. Some other shortcomings, such as the lack of in-situ snow depth data, cannot be easily overcome, but are being addressed in the manuscript. All that said, I think that the study is already presented well, and expect that these points will overall represent a minor revision effort. While some remarks concern very detailed points, that only indicates the already high level of the manuscript.
Specific comments
The structure of the introduction is overall logical and nice to follow, but some things are a bit confusing and should be clarified (a large part of which is related to terminology).
The methods are overall solid and thorough, but some statements lack precision and empirical support.
I find it a bit hard to extract the most important findings in the results section. There are many detailed findings being presented, and it is sometimes a bit hard to follow. I do not have a very specific suggestion to improve this though. Perhaps one point to clarify would be the specific level of variation that you are looking at, as there are at least four in parallel (sites, areas, regions, winter seasons). You could use that to structure the text (e.g., (1) across areas and within winter season, (2) within area and within winter season, (3) within area, but between winter seasons, etc). In addition, I think it is a pity that only relationships for above-ground temperatures are being presented. Given that recent research has shown e.g. the importance of vegetation cover for winter temperatures below-ground, and that this data is also available for this study, I think it could be interesting to repeat the analyses for below-ground temperatures and include the results either in the main text or in the appendix for comparison. I appreciate that this would mean some additional work on the authors’ side, but I think it would increase the scope of the study even more as it would allow conclusions with regard to winter belowground processes in boreal forest and tundra ecosystems. In my opinion, it is a key strength of the study setup with TMS loggers and elevated loggers that it enables comparison of temperatures between different heights/layers along the soil, snow and vegetation profile in the same spot, and this could be exploited even further. Finally, I would find it interesting to see how the actual spatial temperature patterns display on the different landscapes. This could be analysed through variograms, or simply by plotting the spatial layout of the study sites with an associated temperature variable of interest. There are repeated references to “spatial variation” in the text, but the evidence that is currently being shown only represents across-site variation irrespective of spatially explicit relationships.
Although relatively brief, the discussion covers the immediate aspects related to the study’s findings well. However, I would suggest to expand a bit on the ecological implications, which are also mentioned in the first paragraph of the introduction (for instance with regard to vegetation dynamics or ecosystem processes such as permafrost dynamics or soil processes). In this context, the discussion currently only includes a brief reference to effects of snowmelt date on the start of the growing season.
Technical corrections
Abstract:
L2 you use “near-surface temperature” before, so it’s not clear if this is the same?
L2f it sounds more like you are looking at snow cover thickness, but only in L5 it becomes clear that it’s about snow cover duration. I suggest to make that clear from the start.
L10f “In the tundra” – it’s unclear if these differences were at the site/plot/… scale
L11 “lead” should be “led” to match the past tense used otherwise
L12 add a comma after “variation”, else you are saying that there was little decoupling with flat topography
Introduction:
L30ff In this paragraph, I find it a bit hard to distinguish when you are talking about forest and when about tundra. I suggest re-structuring the paragraph to make it easier to follow.
L45 Why are you talking about ground temperatures here, while otherwise only about near-surface temperatures? If the reason is to introduce how you deduced snow cover duration, I recommend to be explicit about it.
L46 “absense” should probably be “absence”
L46f “ground surface” is a confusing term if otherwise distinguishing between “ground temperatures” and “near-surface temperatures”.
L49 “ice particles that affect vegetation growth” – in what way? Positively, negatively, why?
Methods:
L74 Why are we looking at February specifically here?
L83 Please include the abbreviation for Hyytiälä as well
L88 How were the sites laid out in the landscape, i.e. how many plots over what size of an area? That information is essential if assessing spatial variation with regard to scale. Also, it sounds more like site locations were determined stratified randomly rather than strictly randomly, if they were aimed to cover these environmental gradients?
L90f Is 15 cm height really “near-surface” in tundra environments, where often much of the vegetation is below that height? If you want, you could have a look at von Oppen et al. 2022 Global Change Biology for a suggestion for terminology.
L97f How did you select the 40 plots for air temperature logging out of the 100 overall plots, and how did you ensure a balanced selection?
L99f Why did you choose such different logging intervals? Could that affect the data collected, e.g. underestimation of temporal variation in air T when measured through HOBO?
L105 Maybe “weather data” should be “weather station data”? Else I don’t find it intuitive that snow depth is included.
L105ff The paragraph doesn’t make it clear to me why both point and gridded macroclimate data were used, or if there were differences in their use.
L112f So snow depths of less than 15 cm were considered snow-free? I assume there would still be some insulating effect even from a thin snow layer?
L113ff “The loggers were estimated … snow covered periods.” Could you provide an empirical justification for this assumption – either from your data or citation? Why did you choose these specific moving window lengths or temperature thresholds? Also, this is a very complex and dense, yet central sentence to the paper, and I would recommend to restructure and simplify to make it easier to understand.
L122f “… these situations were rare in our study domain, and the algorithm was considered to detect periods of snow cover reasonably well” – This is a very vague statement that in my opinion does not serve to increase trust in the method. Do you have e.g. in situ snow depth data that could provide empirical support?
L123 “are” should perhaps be “were”? I suggest to double-check use of tenses – I prefer past tense in the methods to refer to what was done to reach the conclusions, but that might be personal preference to some degree.
L134 “between a point and its surroundings” – perhaps rather between a grid cell and its surrounding cells?
L139 if only incorporating vegetation > 2m, were treeless tundra sites essentially assumed to have no canopy cover?
L141 was it really spatial variation that was assessed in SEMs? From my perspective, variation among plots and sites, yes, but perhaps not explicitly spatial?
L147 “two-week averages” – so you only used 2 weeks out of 8 months winter data for some sites?
L148f Does that mean that the end-of-snow season temperature is the two-week average before the end of snow cover season?
L149f “Snow cover … late-season models” – As indicated above, I think “snow cover duration” would be more accurate here. Also, this sentence is quite complex and would benefit from simplifying.
L152 I think the grouping approach is absolutely valid, but did you pool the data or average the effect sizes within groups of study areas?
L157f “SEM is … based on prior knowledge on how the system functions” – I think it could be useful to spell that prior knowledge out in a specific hypothesis / schematic figure etc, to clarify your expectations. Also, perhaps this descriptive bit could be combined with the background on the SEM method above (L142ff)?
L158f “We expected solar radiation to have only a marginal effect in mid-winter” – that is probably fair to assume, at least for the Northern study areas, but is that expectation backed up by any empirical data? Why not just include it and test this expectation?
L160 “similar” is too vague here in my perspective. If not identical, I suggest describing the differences in model structure explicitly.
L160 see my remark above on spatial arrangement of the sites. I think some background info on site distribution in each area would be helpful.
L161 “study area as a random intercept” – if I understand it correctly and “study area” = “landscape”, this random intercept will not account for spatial aggregation within a study area?
Results:
L175f I don’t think the “length of the snow cover season” is actually shown anywhere explicitly, or at least it is very hard to see with the non-transparent polygons in Fig. 2, but that would be interesting and useful information. Perhaps it could be added to Fig. 3?
L176 “At the ground surface” vs. “near-surface” in the next sentence, but I assume they are referring to the same layer – again, I suggest you keep the terminology more consistent to avoid confusion.
L176 looking at Fig. 3 a/b, there are three levels of variation that this statement could be referring to - sites, areas, and years - and if seen across sites within areas, they actually varied more (as you are also mentioning further down), so this statement is not generally true. I suggest to be explicit about which level of variation you are referring to.
L176 Why “mean February”? in the Methods, you only mention two-week averages. Is this referring to the same variable?
L181 “There was also more variation in winter minimum near-surface temperatures” – where was that, and more than what/where?
L221/223 See comments above on the use of “spatial variation” – I would use “across sites” here.
L225 Perhaps it would be worthwhile mentioning the negative exponential shape of the relationship here?
Discussion:
L232 Either choose the term “heterogeneity” or “varied considerably” – both do not make sense here
L235 “low-lying vegetation strongly influence snow accumulation patterns” – I agree from a theory point of view, but yet, SEMs did not identify a relationship between canopy cover and snow cover duration. This could be indicating the limited use of the canopy cover variable of choice for the tundra (see my remark in the methods section).
L241f It might be better to stick to "canopy cover" here for consistence (as you do below) As I see it, “vegetation structure” is more complex than the way canopy cover was measured in this study (e.g. including cover at multiple heights, maximum height etc, so essentially three-dimensional).
L256f I am not sure if I understand this sentence. As far as I am aware, De Frenne et al. (2019) actually found a positive buffering effect on minimum temperatures (i.e., a positive temperature offset). If this statement is meant to refer to offsets in mean temperatures, it should be rephrased to make that clearer. Also, importantly, I am in doubt if De Frenne et al. (2019) is a very appropriate reference in this context, as their analyses were mainly based on growing-season temperature records.
L257ff For canopy-snow interactions, you might also find the works of Malle et al (https://doi.org/10.1029/2018JD029908) and Mazzotti et al (https://doi.org/10.1029/2019WR024898, https://doi.org/10.5194/hess-2022-273) interesting, which represent some more recent developments in the field than the sources already cited.
L261f The last sentence in this paragraph does not make sense as it is now, I suggest revisiting it.
L274 Maybe reiterate for the readers that these snow depths were measured at weather stations and not in situ. With regard to spatial variability, some of the above-mentioned references might be relevant here, too.
Figures:
Fig. 1 for panel c) and d): it could be nice to have the comparison with the 1991-2020 period here as well, like in Figure 2. In the legend, it says “study areas”, whereas in the text, I think these have been referred to as “sites? Please indicate the data source in the figure caption.
Fig. 2 Please indicate the data source in the figure caption. I find lines a bit misleading if showing monthly means for temperature, as they give the impression of continuous data. Maybe use dots instead or in addition? Adding outlines to the polygons (colour = ... argument in ggplot), or adjusting the colour scheme and/or transparency could make the snow depth data more readable. Also, I suggest to include the keyword macroclimate at the outset of the caption since that is used to refer to the figure in the text paragraph.
Fig. 3 I suggest to make it clear that there was (apparently?) no snow in KAR in 2020-21. I recommend to add units to the temperature axes. Also, I find it difficult to compare variation in snow cover start vs end date with different axes on panels e) vs f), I think aligning them would make it easier to verify the statements made in the text (L193f).
Fig. 4 Aren’t the shaded areas the snow cover-free periods, contrary to what it says in the caption? In addition, I noticed that this figure is actually being referred to very little in the text, and I think it adds relatively little to the information shown in Fig. 3, so you might consider moving it to the appendix.
Fig. 5 I recommend explaining the variable abbreviations in the caption. That would help to make the figure more stand-alone, so readers don’t have to look them up in the text. Also, I suggest renaming the response variable “surface T.” to “near-surface T.” to increase coherence with the text.
Fig. 6 It is not clear from the plot if all vertical axes show beta or temperature differences. I suggest adding a label for clarification.
I hope that the authors will find my remarks helpful. I wish them good luck and all the best!
Jonathan von Oppen
Citation: https://doi.org/10.5194/egusphere-2023-576-RC1 -
RC2: 'Comment on egusphere-2023-576', Anonymous Referee #2, 23 May 2023
reply
Review of manuscript “Variability and drivers of winter near-surface temperatures over boreal and tundra landscapes” submitted to The Cryosphere
The manuscript presents an analysis of winter near-surface temperatures along a south-north transect, going from boreal forest to tundra regions in Finland. The manuscript studies the drivers affecting the near-surface temperature in winter using a Structural Equation Modelling framework for the boreal and the tundra regions. Results show that snow cover duration has a strong control on soil temperature, but with opposite effect for mid-winter compared with late-winter. Site with shallow snowpack show stronger spatial variability, while site with flat topography and deep snow show strong decoupling between air temperature and soil temperature.
The manuscript is overall well written and generally show expected results. The findings of the study are not necessarily new; however, the dataset is quite extensive along a south-north gradient, which add some value to the study. The analysis is based on an interesting statistical approach which allows to see the interactions between the different drivers. However, such approach limits the applicability of the findings and in that sense a discussion on the possibility to use the dataset to improve physical modelling of snow and soil temperature could be interesting. Some aspects of the methodology should also be clarifier. Overall, the manuscript is suitable for The Cryopshere, but proper improvement should be brought to the manuscript.
- The abstract can be clarified. For example, mentioning “seven study areas across boreal and tundra landscapes”, it seems that the study was made across the northern hemisphere. It is important to clarify the study extent. There should be one or two sentence on the method used to get to the results (statistical approach).- Also, the results based on snow cover duration and the SEM is not quite clear in the abstract.
- Line 38: “slow down snow melt during spring through energy balance controls”. It is more complex than that. There are melting related to tree radiation around the trunk. It is mentioned in the discussion. Need to be clarify here.
- Figure 1. There is a need to clarify what represent c and d. Is it a histogram of all study sites for each study area?
- It is important in the text to well distinguish between “study area” and “study site”. Sometimes, it can get confusion. Maybe using clear acronym for each could help?
- Line 91: There is a need to clarify what “-6” means. Is it 6 centimeters under the surface. So it means that the 2 cm is above the surface? So it means that the means surface is 2 cm above the ground? Needs to be clarify. How the sensors were kept above the surface?
- Snow Cover duration: I have some doubt about the snow cover duration calculation. Why using the 15 cm? Even if the 15 cm is not cover by snow, it doesn’t mean the 2 cm is not cover by snow? But the problem with using 2 cm to get the snow cover duration is that you would use the same measurements to get the snow cover duration and the impact of snow on near surface temperature. This point needs to be clarified/discussed.
- In addition it is not clear if the snow cover duration was calculated for each “study site” or each “study area”. Figure A1 is confusion because it shows all the near surface temperature at 2 cm (? need to be clarify) and the snow cover duration. However, that would be interesting to show the 15 cm temperature and air temperature to see how the snow cover duration was calculated.
- Line 147-150: These sentences are confusing. It seems that the calculations were done for each “study area”. However, all the data is available to make the calculation at each “study sites”. From these sentences, I understand that the snow cover duration is calculated for a study area, when you can calculate it at each study site. It would be very important to clarify this point and clarify how many “N” are used in the SEM.
- It also seems that the total snow cover duration is used as a variables in the SEM to explain the near-surface temperature in mid-winter. It seems inadequate to use a full winter snow cover duration as a variable to explain near-surface temperature in the middle of the winter? Maybe looking at the beginning of the snow cover would make more sense?
- Would be important to mentioned if the study area are in permafrost regions. It will have a impact on the thermal regime of the soil and thus on the near-surface temperature.
- Would be useful to give a more representative acronym for the “beta”.
- line 207: “Snow cover duration had a strong positive effect (0.73) in mid-winter”. It is quite surprising to get such a strong relation when the end of snow season should not have any impact on the mid-winter soil temperature?
- Figure 6: Should clarify what is on Y axis. Also “linear regression model calculated from a two-week moving window”, add “(beta)”
- In the discussion, it will be important to mention the soil thermal regime. Your measurement are above the ground (2 cm) if I understood well. However, it is well known that a wet soil will stay at zero curtain longer because of the latent heat. Permafrost can also alter the soil thermal regime. Even if the measurement are not done in the soil, the author have to recognize the potential strong impact of soil on the results, which are not considered in the study.
- As mentioned earlier, the results are interesting, but not quite new. Ideally, the study would have been conducted using snow physical modelling. But I understand that it is not the scope of the work. However, should be important to relate the results to possible improvement in soil temperature modeling.
16: Figure A2: not clear what “predicted” mean in that context?
Citation: https://doi.org/10.5194/egusphere-2023-576-RC2
Vilna Aleksandra Tyystjärvi et al.
Vilna Aleksandra Tyystjärvi et al.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 147 | 59 | 8 | 214 | 2 | 4 |
- HTML: 147
- PDF: 59
- XML: 8
- Total: 214
- BibTeX: 2
- EndNote: 4
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1