the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 19 Jul 2024
Tidal influence on flow dynamics of Dotson Ice Shelf, West Antarctica
Abstract. On Dotson Ice Shelf, Antarctica, ice velocities derived from satellite image pairs and in-situ GPS measurements reveal an oscillating flow pattern that is correlated with tide height. The tidally-affected flow pattern is of limited extent, in an area near the Wunneberger Rock nunatak in the outflow of Kohler Glacier. Comparing variations in the region’s flow velocity derived from a series of 16-day repeat-pass Landsat 8 image pairs spanning 2014–2020, and a 64-hour GPS record in 2022 with the CATS2008 and TPXO9 tide-height models, indicates a significant correlation between tidal uplift and the direction of ice-flow. During high-tide periods the ice-shelf flows in a true north direction, while at low-tide periods flow direction shifts towards the northeast, marking an approximately 40˚ change in flow direction. GPS measurements describe a continuous corkscrew-like motion of the ice-shelf surface, confirming the link between tide height and ice-flow direction. We attribute the observed pattern to tidally controlled changes of buttressing along the ice-shelf margins and the fin-like shape of Wunneberger Rock. This leads to a dual pattern: (i) fast flow across the grounding line of the tributary Kohler Glacier during high tides aligning with Wunneberger Rock’s summit ridge; and (ii) slow flow during low tide height facing its flanks. We suggest that the link between tides and ice dynamics is related to the rapid ice-shelf thinning in the area. In light of the continued thinning of ice shelves surrounding Antarctica, we anticipate similar variations in flow direction and speed arising from changes in tidal influence on buttressing from pinning points and grounding zones.
- Preprint
(6764 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-1895', Whyjay Zheng, 28 Aug 2024
Review of Collao-Barrios et al. (https://doi.org/10.5194/egusphere-2024-1895)
The in-situ GNSS observations are indeed a valuable part of this work. The observational length is short (64h), but the data shows varying ice movement along with the tidal cycle. The analysis of the GoLIVE velocity maps, however, is not as straightforward as the GNSS data due to the grouping strategy and the comparison with the temporally averaged tidal model. I have listed my major and minor comments below. I hope these are helpful for potential improvements to the manuscript.
Major comments:
- GoLIVE pairs and the grouping strategy. The authors use 23 pairs with a 16-day duration, which is considered reasonably sparse within seven years (~14% temporal coverage assuming no overlap). The grouping strategy is somewhat confusing. Do you sort these velocity maps (based on speed or flow direction?) and use the ones at 25% and 75% percentile? Or are the q25 and q75 composite maps with velocity and flow direction pulled from different maps? Based on L77-80, it seems that the latter method is used. If so, how can we make sure high speed corresponds to a 270-degree bearing? After all, they could be from different velocity maps.
Considering the small number of pairs used in this study, it would make more sense if the authors could show the distribution of the ice flow direction rather than just showing q25 and q75. It would help recognize whether two populations (peaks) are in the distribution. We need to find a way to highlight that many (if not all) high-speed flows have a bearing of 270 degrees. Maybe a rose plot could be useful here?
I can also see some statements that are hard-wired true because of this grouping strategy. For example, L144-145: “Speed also varies … higher speed in q25”; and L265-266: “velocity tracking indicates faster flow for the q25…”. I feel it’s strange to say this because q25 already represents the velocity map with high speed based on the grouping strategy. It’s the manual assignment, not something the results indicate, isn't it? - Averaging tide heights over 16 days. Again, it would be helpful to see the data from all 23 periods. The averaged values seem to be at the centimeter scale, which is much less than the error between the GNSS measurement and the tide model, according to Figure 6. How do you incorporate this large uncertainty into the significance test?
- Statistics. We need more information about how the authors obtain the p values for the GNSS data vs. tide height. GNSS data points are temporally correlated. I’d like to know how the authors deal with this issue (using an effective sample size or similar) for the correlation coefficient and the p-value.
- Issues of indicating directions. The authors typically use the EPSG:3031 northings and eastings for directions (e.g., L215: northern border; L147 & 232: southeast tributary), but we can also see directions in true geographic sense (e.g., L6: true north direction), and sometimes it’s not very clear (e.g., all directions in L203-209).
The authors also reproject the GNSS movement to a flow-based coordinate system. However, as the article suggests, the flow directions change from time to time, and I am not sure what direction the GNSS movement is reprojected to. In Figure 8b, the pink color does not look like a 135-degree bearing, as the caption suggests. (I guess they are still indicating the tidal heights?) The statement in L178-179 does not match the content in Figure 8b, either. I can’t see anything in the 270-degree bearing in Figure 8b. - Figure 8. The semi-diurnal tides also lead to high ice-flow speed (8c), contributing to a significant offset (8b; grey segments). This has made me think whether it would be more accurate if we compare the net flow direction during these 64 hours with the average tide height, in order to make a connection between the GNSS observations and the LS8 feature tracking results. (So that both are averaged to a certain temporal extent.)
Also, it would be good to add some timestamps in 8b along the GNSS track as additional connections to 8a and 8c.
Minor comments:
- I suppose replacing all “GPS” in the article with “GNSS” is more accurate based on the GNSS receiver you used?
- Maybe change all “pv” to “p-value” or italic p?
- I do not think the current data availability statement fulfills the requirements of TC. Plus, www.nsidc.org is not a good landing page for GoLIVE. Readers benefit the most if the authors could provide an accurate landing page (preferably with DOI) for the open data sets. This includes GoLIVE, ITS_LIVE, BedMachine, and the OIB radar data.
- L66: Does this paper identify the change in ice dynamics at a multiyear scale?
- L69: Any reference of this 0.1 pixels? Sciacchitano (2019) suggests it can be up to 0.2 pixels.
- L121: Fd in Eq. (4)?
- L128-136: This paragraph is not very clear to me. What are the variables the values in L130 correspond to? Where do you use B?
- L202-203, Figure 2, and Figure 9: It would be great if the authors could use and indicate the same name for the radar profiles. I have a hard time recognizing which is P1 and which is P2 in Figure 9.
- L223-227: This paragraph is not very clear to me. I could not see the deceleration of the q75 pattern that does not exist in q25.
- L228-229: day^(-1) instead of a^(-1)?
- L255-256: Why is the change of resistive force the causative factor of ice flow direction, not the opposite way around?
- L257-259: How did you obtain these percentages?
- L289: Where is the Kohler Rumple in Figure 1?
- L290-291; 297-298: Is there supporting evidence for this claim? Why is it less likely that the flow has been modulated for more than a few decades?
- Many of the figures are provided in low resolution. Is it possible to update them to a higher resolution?
- Figure 2: For a quick understanding, it would be great if the authors could mark the GNSS station location in panel c, just like they did in Figure 13.
- Figure 3: Multiple grounding lines of different periods co-exist with the same color, impeding the interpretation of the figure a little bit.
- Figure 5: Flow directions (120-180 degrees?) do not match what is presented in other figures and text.
- Figure 12: I don’t see any Grey arrows…
- Table 1: Fd for the q25 seems incorrect; it does not match Figure 12d.
Reference
Sciacchitano, A. (2019). Uncertainty quantification in particle image velocimetry. Measurement Science and Technology, 30(9), 092001. https://doi.org/10.1088/1361-6501/ab1db8
Citation: https://doi.org/10.5194/egusphere-2024-1895-RC1 - GoLIVE pairs and the grouping strategy. The authors use 23 pairs with a 16-day duration, which is considered reasonably sparse within seven years (~14% temporal coverage assuming no overlap). The grouping strategy is somewhat confusing. Do you sort these velocity maps (based on speed or flow direction?) and use the ones at 25% and 75% percentile? Or are the q25 and q75 composite maps with velocity and flow direction pulled from different maps? Based on L77-80, it seems that the latter method is used. If so, how can we make sure high speed corresponds to a 270-degree bearing? After all, they could be from different velocity maps.
-
RC2: 'Comment on egusphere-2024-1895 by Matt King', Matt King, 17 Sep 2024
The authors present an analysis of the tidally-related 3D motion of the Dotson Ice Shelf, from 16-day velocities over 6 years and a very short GPS record. The authors present evidence that the Wunneberger Rock in the floating ice shelf is an important control on this motion, with effects reaching well upstream. The paper is quite well written although there are a number of more minor issues of clarity and completeness. The introductory material and discussion lack a discussion of the wider literature on ice shelf tidal modulation and so misses an opportunity to extend the implications further afield. Parts of the conclusions leverage some brief and quite speculative parts of the discussion and in my view go beyond what is supported.
The topic is certainly appropriate for TC and the results yield new insights into an important part of Antarctica.
More substantial remarks:
It seems important to know how sensitive choices of result are to absence of smaller scale features that BedMachine will not capture. Or at least what is the formal uncertainty of the various strain estimates? There is a general low level of providing data uncertainties in the paper.
The radar profiles were almost impossible to interpret as the panels of Figure 9 were not explained in the caption and Figure 1 shows P1 and P2 but Figure 9 does not say which profile is which. Figure 1 caption does not mention P1, P2.
The phase colour scale does not seem to be perceptually uniform. As per https://www.fabiocrameri.ch/cycliccolourmaps/ for example. This makes (if I am correct) the colours difficult to interpret. Please update throughout.
The methods for computing the tidal height are not clear to me and the rationale for choosing the average needs a bit more text. I presume the thinking here is that the state at t2 relative to t1 16 days earlier is due to the time-integrated effects of the tidal height? And is this the height relative to the mean (in which case why is the mean not close to zero) or the range or the magnitude? I got more confused by Figure 5 - why for instance is the blue line down at -1.4m or so when none of the tidal range goes to 1.4m? What have I missed? So I didn't understand a key part of the paper.
I suggest the manuscript would benefit from engaging with the literature on tidal modulation of flow of ice shelves. For example, Doake et al 2002 and Makinson et al 2012 and Rosier and Gudmundsson 2020, although others exist focused on Ronne and Larsen C, and Brunt et al 2010 for Ross Ice Shelf.
GIven the widely observed non-linear propagation of tides into ice flow (to biweekly or semi-annual for M2 for instance) what impact would such a modulated have on the analysis here? Is anything observable? Thinking here particularly of biweekly band. This may be irrelevant given the largely diurnal tides.
The final two sentences of the abstract leverage the most speculative part of the manuscript, and I am not entirely persuaded by that brief argument that enough is known to make such a prominent conclusion.
minor remarks:
L84 was 1cm/hPa used?
Secvtion 2.6 many of the terms in Eq 1-4 are not defined. L135 B is meant to be Beta? L129 delete "and"
L139 the analysis may have been multiyear, but I guess you mean 'analysis of multiyear ice-flow data'?
L159 pv is not common notation
L179 Fig 8 would benefit from a new panel between b and c of the detrended along-flow displacement.
L181 the work of Robel seems to be theoretical rather than an observation for this ice shelf. And is certainly not the only observation of such things (thinking here of Gudmundsson's extensive work on Rutford).
L184 it was curious to reference a movie that is not available for the review.
L202 are the angles relative to horizontal or vertical?
L211-212 the wording of this sentence is confusing.
L218 'a[n] along-flow' also L233. Is this along a 'central flow line'?
L223-227 the point regarding Wunnegerger Rock wasn't clear here. I see you got to it in section 3.5 but it left me a little confused as to the direction and point here
L228-231 add the Figure panel references into this para
L258-259 what does 'dependency' mean here? it is not a standard term. I think this is the variance explained being referred to (R^2)?
L289 I do not see Kohler Rumple marked on Fig 1
L297 D6 is only mentioned in the introduction once and now here. Surely if it is important it should be in the main body?
L304 'will [be] available]
Figure 1 - hard to read the grey labels. Add a marker for the GPS site. could maybe add an inset with a zoom to WR region
Figure 5 the caption '16-day flow direction average for correlation' is not clear what this means
Figure 7 a colour scale passing through white isn't a good choice here.
Figure 8 panel b does not seem to have a different colour scale to a) or c) as intended. Is there a trend in panel c? if so, is that because the reference direction is not reflective of the flow in this period? add to the caption the reference direction used to rotate into across and along flow. The caption says the arrows refers to acceleration periods whereas they point to max speed which is not max acceleration.
Figure 9 define Heq. explain panels. link to P1 and P2 in Figure 1
References
https://tc.copernicus.org/articles/14/17/2020/
https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2012GL051636
https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2001GL014606
https://www.cambridge.org/core/journals/journal-of-glaciology/article/flow-of-the-ross-ice-shelf-antarctica-is-modulated-by-the-ocean-tide/300FF223B355B9E7F9D2B25B74285A80
Citation: https://doi.org/10.5194/egusphere-2024-1895-RC2 -
RC3: 'Comment on egusphere-2024-1895', Anonymous Referee #3, 17 Sep 2024
Collao-Barrios et al. present GPS data from Dotson Ice Shelf and analyze available satellite-derived velocity information from the ice shelf. They identify 2 different ice-flow direction regimes. They correlate flow direction and tidal heights, both quantities averaged over 16-day satellite window, finding low, but significant correlation. From the mean ice velocity fields, they calculate resistive forces around a pinning point, finding that the magnitude and direction of the resistive forces depends on the ice velocity field. There is some tidal analysis of the GPS data, but that needs substantial revision and at the moment doesn't support any of the claims in the paper. There are some speculations about present and future relationships between tides and mean ice flow towards the end of the manuscript, but they are not supported by any mechanism, or model.
I think the results that identify 2 different ice-flow direction regimes are really interesting. However, there is no proposed mechanism for why the flow rotates at times. Figure 13 describes the correlations, but doesn't offer a mechanism. Related to that, some information about the timescale and frequency of the occurrence of the two regimes would be useful. There is a lot of focus on the nunatak in this paper, but it is a bit unclear why, because its role in the temporal change in the flow field isn't identified. I think the result that a different flow field direction is associated with a different resistive force direction is expected.
There are some significant misconceptions about tides, as far as I can tell. For example, the authors divide the time series in segments in time space and state that some portions have diurnal tide and and others have semidiurnal tide. Both (and possibly other) tidal frequencies are present at all times. To analyze the behavior of a given tidal constituent, the authors need to do tidal analysis. This should anyway be done, I think, especially, since the tidal height and flow speed relationship might be different for different tidal constituents, as suggested by Figure 8. The authors say that for diurnal tide there is a correlation in the way that is consistent with longer timescales, but then hand wave a different segment of the time series as 'more complex'. This complexity should be analyzed appropriately and any discrepancies explained. All this is relevant only if the authors decide to keep the GPS data as part of the paper and connect it to the longer term dynamics somehow. Also some justification for why the response on short and long timescales should be the same needs to be provide, given the viscoelastic ice rheology.
The correlation between longer term ice velocity direction and tidal heights is relatively weak, and in the absence of a mechanism connecting them it is questionable to make a statement about tidal influence on flow dynamics, as the title of the paper suggests. Anyway it is not clear whether the authors talk about tidal influence on tidal flow or tidal influence on mean flow. It should also be shown somewhere or discussed that tides are the dominant contributor to the sea surface height changes in this particular region of Antarctica - as that assumption underlies some of the claims.
Any claims about expected future change should be supported by a mechanism.
Other comments are below and in the attached pdf:
Perhaps an ice flow model could shed light on the importance of the nunatak that seems to be a major focus of this paper?I think the authors should clearly separate out robust results from interpretation and speculations. Specifically, rather than talking about low/high tide regime they should discuss this as regime 1 and regime 2 throughout the paper, and only in the discussion bring in the subsequent interpretation that involves tides, because that part is not very robust. The tides do not come in into the resistive force estimates at all, so need to bring them in early on.
The drag calculation is diagnostic from the velocity fields. It doesn't actually help explain the change in the velocity field.
Methods:I have a hard time following the satellite velocity methods. I think at the moment it is not clearly stated what product is used for what specific purpose. It would be good to separate that out before describing details about the products.
Results:
Do the quartiles based on direction vs flow speed overlap? If not, then larger velocity variation than described might take place, and that probably needs to be addressed somewhere somehow. Showing the distributions would be informative too.
When talking about properties (speed, flow, direction) it always needs to be specified if you talk about tidal or background (non-tidal) flow
Figures:
All figures need x-y labels and tick marks for some reference! At the moment most of them are missing this absolutely essential piece of information.
Statements in the paper should reference figure panels. More frequent figure referencing would be appropriate.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 257 | 104 | 26 | 387 | 12 | 10 |
- HTML: 257
- PDF: 104
- XML: 26
- Total: 387
- BibTeX: 12
- EndNote: 10
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1