Sensitivity of glacier volume change estimation to DEM void interpolation

McNabb, Robert; Nuth, Christopher; Kääb, Andreas; Girod, Luc

doi:https://doi.org/10.5194/tc-13-895-2019

Articles | Volume 13, issue 3

https://doi.org/10.5194/tc-13-895-2019

© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/tc-13-895-2019

© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 13, issue 3

Research article

|

15 Mar 2019

Research article |

| 15 Mar 2019

Sensitivity of glacier volume change estimation to DEM void interpolation

Robert McNabb, Christopher Nuth, Andreas Kääb, and Luc Girod

Download

Final revised paper (published on 15 Mar 2019)
Preprint (discussion started on 01 Oct 2018)

Interactive discussion

Status: closed

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

- Printer-friendly version

- Supplement

RC1: 'McNabb et al review – Sensitivity of geodetic glacier mass balance estimation to DEM void interpolation', Nicholas Barrand, 02 Oct 2018
- AC1: 'Response to reviewer 1', Robert McNabb, 25 Jan 2019
RC2: 'Authors can turn this “good” study in a “benchmark” paper for the community', Anonymous Referee #2, 18 Oct 2018
- AC2: 'Response to reviewer 2', Robert McNabb, 25 Jan 2019
RC3: 'Review of ‘Sensitivity of geodetic glacier mass balance estimation to DEM void interpolation’ by McNabb and colleagues', Anonymous Referee #3, 19 Oct 2018
- AC3: 'Response to reviewer 3', Robert McNabb, 25 Jan 2019

Peer-review completion

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

AR by Robert McNabb on behalf of the Authors (25 Jan 2019)

ED: Referee Nomination & Report Request started (06 Feb 2019) by Daniel Farinotti

RR by Anonymous Referee #2 (18 Feb 2019)

Suggestions for revision or reasons for rejection

Mc Nabb have done a nice job at taking into account the three reviewer’s comment. The rebuttal letter could have been more “self-contained” but in the end it is important that they put most of their energy in revising the article. In particular I appreciated their effort to explore more aggressive threshold while creating data gaps and also the fact that they totally changes unit to report their findings. The study illustrates the robustness of some of the methods even when data gaps cover up to a large fraction of the glacier area. That said I still have a series of comments that will be easy to take into account and will not require an additional review from me, just an editorial check. Authors can nicely finalized their paper through a final careful read to improve the quality of the writing (in particular the abstract).

1.1 “recently” Hum the geodetic method finds its fundation back to the 1950s... for individual glaciers at least.
1.2 The term “full coverage” is not appropriate given that data gaps are the main topic! Skip.
1.7 this sentence reads as if the ASTER scenes were made of correlation values. Should be rephrased.
1.10. Final recommendation. This is quite a strong statement for something that has not been really explored in the article. Such a statement would be justified if the authors had proved that an “ensemble” of the 3 or 4 best performing interpolation methods would come closer to the true volume change than the methods taken individually. This was not demonstrated here. Authors may want to rather finish by an implication sentence. I think one strong implication of their study is that, with the 2 or 3 best methods, reliable glacier-wide mass balances can be obtained even with a limited coverage. This is a good news for our community.
2.1 How to say opposite things in the same sentence. I am sure authors can improve here.
2.29 "area" of data voids? I find the brief description of this specific method unclear.
2.32 I am still not convinced by global but I leave it to the editor to decide if this comment is relevant. My suggestion is still to change to “regional”.
3.1 See similar comment about the abstract. The correlation values comes from the DEM generation process not from the optical images themselves. Sorry for being picky here but I think it is important to be clear.
4.10. Is "provided" necessary? (there are many such small verbs that could be deleted I think)
4.20 check wording.
5.22 "for a select few interpolation" does not really sound OK for me.
5.22 Maybe mention also the 35% threshold already here (I was surprised to discover it later).
5.31 A glacier has an area, not an outline. Same comment applies to 6.1
7.13 "cf" not needed before references.
8.10 I find it a bit problematic that the authors refer several times to the Melkonian et al. 2013, 2014 studies as it has been proved that their estimates of the regional mass balance is severely biased positively (due to SRTM penetration).
8.20-24. Nice addition to put your region in perspective. Good job.
8.28. Is this the mean of all dh/dt values or the mean of the glacier-wide average dh/dt for all glaciers. Both values could be different due to the varying size of individual glaciers.
9.8 "83.9%" or even "84%"
9.30 I suggest replacing "trend" by "ones"
9.31 Distribution of what?
10.3 suggestion: "data used to fill voids for Taku Glacier are…"
10.18 What are "these"? The dataset from this study? Or from the cited references?
11.15 unclear what a "low method" is. Correct.
11.31 telling right away what this caveat is could help the reader I think.
12.12 This is an interesting test but in fact authors do not know what is the true value because of the different time span (3 years in one case, 13 years in the other case) and also the lack of correction for penetration. They should maybe only focus their discussion on the interpolation artefacts and not the mean dh/dt values.
12.22 some words missing in this sentence.
12.31 Suggestion "For the low noise level DEM used here, “
13.18 Is "introduced" needed? (another example of unnecessary adjective)
13.24 "The choice of a void interpolation method" (I think)
13.32. A bias is biased... correct sentence.
Figure 1. Names of glaciers in Red are hard to see (black would be best maybe). What about increasing the thickness of the outlines for these specific glaciers names in the text?
Figure 2. Indicate which threshold is chosen here (50% I assume)
Figure 10. mean over the glacier area?
Figure 11. This is the “void percent” for each glacier? Or for the whole region?
Figure 11 and 12 are great additions. I command the authors for exploring the effect of the % of voids.
Table 2. Authors could provide the true volume change (in m/yr) in the legend. It would provide a context for the interpretation of the values in the table.

Hide

ED: Publish subject to minor revisions (review by editor) (22 Feb 2019) by Daniel Farinotti

AR by Robert McNabb on behalf of the Authors (26 Feb 2019) Author's response Manuscript

ED: Publish as is (27 Feb 2019) by Daniel Farinotti

AR by Robert McNabb on behalf of the Authors (27 Feb 2019)

Short summary

Estimating glacier changes involves measuring elevation changes, often using elevation models derived from satellites. Many elevation models have data gaps (voids), which affect estimates of glacier change. We compare 11 methods for interpolating voids, finding that some methods bias estimates of glacier change by up to 20 %, though most methods have a smaller effect. Some methods produce reliable results even with large void areas, suggesting that noisy elevation data are still useful.