Transient modeling of the ground thermal conditions using satellite data in the Lena River delta, Siberia

Westermann, Sebastian; Peter, Maria; Langer, Moritz; Schwamborn, Georg; Schirrmeister, Lutz; Etzelmüller, Bernd; Boike, Julia

doi:https://doi.org/10.5194/tc-11-1441-2017

Articles | Volume 11, issue 3

https://doi.org/10.5194/tc-11-1441-2017

© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 3.0 License.

https://doi.org/10.5194/tc-11-1441-2017

© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 3.0 License.

Articles | Volume 11, issue 3

Research article

|

28 Jun 2017

Research article |

| 28 Jun 2017

Transient modeling of the ground thermal conditions using satellite data in the Lena River delta, Siberia

Sebastian Westermann, Maria Peter, Moritz Langer, Georg Schwamborn, Lutz Schirrmeister, Bernd Etzelmüller, and Julia Boike

Download

Final revised paper (published on 28 Jun 2017)
Supplement to the final revised paper
Preprint (discussion started on 04 Jul 2016)

Interactive discussion

Status: closed

AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment

- Printer-friendly version

- Supplement

RC1: 'Transient modeling of the ground thermal conditions using satellite data in the Lena River Delta, Siberia by Westermann et al.', Anonymous Referee #1, 05 Aug 2016
- AC1: 'Reply to Reviewer 1', Sebastian Westermann, 12 Dec 2016
RC2: 'Manuscript of Westermann et al.', Anonymous Referee #2, 17 Oct 2016
- AC2: 'Reply to Reviewer 2', Sebastian Westermann, 13 Dec 2016

Peer-review completion

AR: Author's response | RR: Referee report | ED: Editor decision

AR by Sebastian Westermann on behalf of the Authors (13 Dec 2016) Author's response Manuscript

ED: Reconsider after major revisions (06 Mar 2017) by Stephan Gruber

Dear Sebastian and others

First, I apologise for the delay in making this decision. Given the potential for great progress offered by your manuscript on one hand – and the one very critical review it received on the other hand, I needed to have a good understanding of the issues involved before proceeding.

I will send this out for another set of reviews now. Specifically, I will ask reviewers to assess whether the criticism of Reviewer 2 has been addressed well and whether the corresponding caveats are expressed appropriately in the current manuscript. I will further ask, what breadth of objectives and conclusions they feel these limitations warrant.

When reading you reply to Reviewer 2, I appreciated in how much detail you have discussed many of the findings cited in the review. Not always does review and rebuttal go into as much detail; I feel this is very useful and it concerns a very important topic. At the same time, I worry that there is a lot of discussion and re-interpretation required to explain why your results should not be treated with more scepticism, given published inaccuracies in the data you use. This then leads to the requirement for making the conclusions of your manuscript more specific and to be more precise in formulating its objectives. As it stands, your conclusion “indicate that satellite-based modeling of the ground thermal regime in permafrost areas could eventually become feasible even on continental scales.” are either too vague (sure, we can model, but how well?), or taken to a larger spatial context from which you argue in your rebuttal that the LRD differs.

In any case, this is an exciting study and I look forward to the new round of review. Again, please excuse my delay. The reasons for this have partially cleared and I expect to be much more responsive with the next round of revisions.

Kind regards,
Stephan

Hide

ED: Referee Nomination & Report Request started (06 Mar 2017) by Stephan Gruber

RR by Anonymous Referee #3 (17 Mar 2017)

Suggestions for revision or reasons for rejection

This work updates a previous study by spatially distributing a point model scheme described in Langer et al (2013) across the Lena River Delta. The paper is clearly written and provides clear evidence of decent model performance despite the heterogeneous environment which complicates the characterization of snow cover (amongst other variables). The study clearly identifies the link between snow and the ground thermal regime, and the impacts of shortcomings in observed snow mass datasets. I suggest the following issues be addressed to improve the final manuscript:

Page 7 line 1: “CryoGrid 2 is capable of representing the annual build-up and disappearance of the snow cover with a variable number of snow grid cells…” Not clear what is meant by a ‘variable number of grid cells’. Does the model not simply simulate snow accumulation/ablation on a grid cell by grid cell basis?

Page 8 line 5: given the importance of snow to model performance of the ground thermal regime, some additional details on the snow component of CryoGrid 2 should be provided. Presumably the precipitation forcing is key to the simulation of a realistic snow cover? What snow processes are simulated by the model (i.e. sublimation loss during blow snow events is likely important in this environment)? It’s not clear in Section 3.4 how layering in the snowpack is treated.

Section 3.3 – I’m confused by the terminology in this section. To me, the MODIS LST product provides a skin temperature, while the reanalysis provides 2 m air temperature. Given the effort required to use the MODIS product, I assume it is skin temperature that the model requires for forcing, not surface air temperature? This also would explain the assumption that temperatures are fixed at 0 degrees when snow is present (which is reasonable for skin temperature, not for surface temperature). If I have indeed interpreted this correctly, please be consistent with the terminology in this section – for example (page 8 line 26) “For melting snow, surface temperatures…” should be changed to ‘skin temperatures’ etc.

GlobSnow SWE: The Lena River Delta is a very challenging environment for SWE retrieval given the high water fraction, organic soils, coastal location, etc. Given the distance to the nearest ground stations, I would expect a high level of uncertainty in the GlobSnow retrievals. On balance, however, I feel the authors provide a convincing case in the supplementary material that the pattern of SWE they derive from MODIS and GlobSnow is reasonable when compared with the CMC product and clear text is provided in numerous places throughout the paper on uncertainties in GlobSnow.

Section 4.1: I’m not that familiar with this product, but is a cold bias in the MODIS LST data consistent with other literature? (it appears so, as you cite some studies later on page 15; I suggest you also include these citations in Section 4.1)
Page 11 line 24: note that the GlobSnow SWE retrieval uses a fixed snow density of 0.24. I assume this was used to back out snow depth from SWE for this product? If the assumption of density =0.24 is applied to GlobSnow to estimate depth, should that same value be used for simulations instead of 0.225? (Perhaps this small difference in density is not important, especially since 0.24 falls within the 0.2 to 0.25 range in density used for the simulations)

Figure 5 panel f: This panel essentially shows uniform snow depth across the domain because only 5 cm is covered by the 4 classes in the legend. Based on the spatial survey shown in Figure 4, this spatial variability is dwarfed by the sub-grid heterogeneity. Presumably the seasonal averaging smooths some of the spatial distribution, but based on the variability in Figure 4 and the pattern in Figure 5, one could argue that the 1 km resolution average snow depth is essentially the same across the domain despite a high degree of sub-grid variability. How sensitive are the ground thermal regime simulations (Figures 12 and 13 for instance) to the use of spatially varying snow depth versus fixed snow depth?

Section 5.1.2: This is a very thorough discussion of uncertainties in SWE retrievals. One isse:
Page 16 lines 13-15: The suggestion here is that a temporally continuous 10 mm SWE error results in a 2.5 degree error in temperature at 2.5 m depth. This is a sobering number (analogous to the issue in altimetric retrievals of sea ice thickness where errors in snow depth propagate to larger errors in freeboard/ice thickness retrievals). The conclusion is that SWE errors of this magnitude are unlikely in this study because ground temperature simulations had smaller errors than 2.5 degrees. This ignores the effect of compensating errors (i.e. SWE biased too high and too low during various times of the season; impact of errors in ground stratigraphy; biases in temperature forcing etc.) and I think it’s actually quite unlikely that SWE biases from GlobSnow were <10 mm due to all the uncertainties provided in the manuscript. I suggest the wording at the end of this paragraph be clarified just to be clear that uncertainty in the characterization of snow cannot be directly inferred from the overall simulation uncertainty.

General comment on the description of the model - various terms are used such as a “satellite-based model scheme” (Abstract line 20) versus “…transient ground temperature modeling scheme forced by remote sensing data…” (Page 2 line 26). I think the second is more accurate as the model itself is not based on remote sensing, but remotely sensed data comprise most (but not all) of the forcing data.

Hide

RR by Anonymous Referee #1 (20 Mar 2017)

ED: Publish subject to minor revisions (Editor review) (09 Apr 2017) by Stephan Gruber

Dear Sebastian and others

Thank you for your thorough revisions to this manuscript. It had received one detailed and critical review with the recommendation to reject (Anonymous Referee #2). That reviewer, understandably, did not want to assess the manuscript again.

Given your detailed rebuttal, I have sent the manuscript for a second round of reviews where a new reviewer came in (Anonymous Referee #3), with the knowledge required to speak to key points raised by the critical review. This reviewer has been largely satisfied with how you have addressed the concerns of Anonymous Referee #2. Anonymous Referee #1 likes your manuscript as is.

Based on the comments from Anonymous Referee #3 and my reading of the manuscript, I am happy that this manuscript can go ahead towards publication. I ask you to address the comments made by the referee as well as my comments below and included in the annotated PDF I have provided.

1) Temperatures can be high/low, but not cold/warm (objects are).

2) P18L21: Your snow density values were determined by in-situ measurements. Please mention this in Section 5.4 where you list challenges for extending this approach beyond the LRD. Not having these measurements may be difficult.

3) In the last few sentences of your conclusion, please point to Section 5.4. I believe that identifying many of the challenges for extending your study past the LRD make a big part of the value of this study – and the reviews it received.

4) P20L7: “If this interpretation is correct, surface temperatures derived from remote sensors have a significant advantage over data sets derived from atmospheric modeling…”. Even if this is not correct and pure coincidence, your statement on the value of remote sensing data would be true: they contain data that re-analyses do not currently resolve.

5) You often judge your model to be “adequate”. This needs to (a) have a purpose declared for which it is adequate and (b) criteria explained that will allow you to judge the model to be either adequate or inadequate. Without these two additions, stating adequacy has no value.

6) P14L7: Why is the accuracy defined with respect to only one snow density value if you cannot plausibly constrain the densities better than 200–250 kg/m3? It would be good to also give the pessimistic estimates derived from the extreme densities. Figure 9 is a great visualization of this.

This is a great manuscript and I am much happier with its more narrow conclusions. Furthermore, I found your careful and detailed discussion of the data used and its suitability for representing LRD conditions in model testing to be extremely valuable, and I hope, inspiring to other studies.

Some of the delays in handling this manuscript had to do with my schedule. These issues have now cleared up and I expect to provide a quick turnaround of your revised manuscript.

Kind regards,
Stephan

Comments to the Author: PDF

Hide

AR by Sebastian Westermann on behalf of the Authors (05 May 2017) Author's response Manuscript

ED: Publish as is (21 May 2017) by Stephan Gruber

AR by Sebastian Westermann on behalf of the Authors (23 May 2017) Manuscript

Short summary

We demonstrate a remote-sensing-based scheme estimating the evolution of ground temperature and active layer thickness by means of a ground thermal model. A comparison to in situ observations from the Lena River delta in Siberia indicates that the model is generally capable of reproducing the annual temperature regime and seasonal thawing of the ground. The approach could hence be a first step towards remote detection of ground thermal conditions in permafrost areas.