Seasonal land ice-flow variability in the Antarctic Peninsula
- 1Scott Polar Research Institute, University of Cambridge, Cambridge, United Kingdom
- 2ENVEO IT GmbH, Innsbruck, Austria
- 1Scott Polar Research Institute, University of Cambridge, Cambridge, United Kingdom
- 2ENVEO IT GmbH, Innsbruck, Austria
Abstract. Recent satellite-remote sensing studies have documented the multi-decadal acceleration of the Antarctic Ice Sheet in response to rapid rates of concurrent ice-sheet retreat and thinning. Unlike the Greenland Ice Sheet, where historical, high temporal resolution satellite and in situ observations have revealed distinct changes in land ice flow across intra-annual timescales, similar seasonal signals have not previously been observed in Antarctica. Here, we use high spatial and temporal resolution Copernicus Sentinel-1A/B synthetic aperture radar observations acquired between 2014 and 2020 to provide the first evidence for seasonal flow variability of the land ice feeding George VI Ice Shelf (GVIIS), Antarctic Peninsula. Our observations reveal a distinct austral summertime (December – February) speedup of ~0.06 ± 0.005 m d-1 (~22 ± 1.8 m yr-1) at, and immediately inland of, the grounding line of the glaciers nourishing the ice shelf, which constitutes a mean acceleration of ~15 % relative to baseline (timeseries-averaged) rates of flow. These findings are corroborated by independent, optically derived velocity observations. Regional contrasts in the onset of ice-flow acceleration and the overall timing of the speedup events across GVIIS fingerprint oceanic forcing as the primary control of this seasonality. Our findings imply that analogous ice-ocean interactions may be ongoing at the grounding lines of other ocean-vulnerable outlet glaciers around Antarctica. Assessing the degree of seasonal ice-flow variability at such locations is important for quantifying Antarctica’s future contribution to global sea-level rise.
Karla Boxall et al.
Status: final response (author comments only)
-
RC1: 'Comment on tc-2022-55', Anonymous Referee #1, 08 Apr 2022
The authors present a comprehensive analysis of the glacier flow variability of the GVIIS tributaries. The analysis relies on Sentinel-1 data and is backed up with independent Landsat measurements. Overall the paper is well structured and most sections of the analysis are well performed. However, there are some issues that must be addressed:
Most important, the authors state that surface meltwater cannot be the trigger of the observed seasonal variations. However, I am not convinced by the presented justification. Recent publications indicate the warming and also increased surface melt on the AP (e.g. Carrasco et al. 2021, Banwell et al. 2021). So, the authors should also consider surface meltwater in the discussion of their findings or provide evidence that surface meltwater can be neglected as a potential driver. (see also comments below, abstract, discussion, and conclusions need to be adjusted accordingly)
Moreover, the description of the methodology has some shortcomings. Please provide here more precise information and be always clear on which region (spatial extent), i.e. whole glacier or just the 10km² areas, is your analysis and interpretation based. Please justify the interpolation in Fig.6 and explain the applied approach. The error analysis should be also extended. See detailed comments below for some specific issues.
Here are also some questions that came to my mind regarding your analysis. Could you please address them?
Why is the ice flow higher in March-November for wide regions further inland of the grounding line (GL) and why is it lower during summer?
Why is the speedup only visible close to the GL. Why is there now speed up further up?
Is there any correlation of speedup with altitude (either the area affected by the speed up or the general hypsometric profile or hypsometric index of the glaciers)? Difference Alexander Island vs. AP?
Tides are also affected by the season. Could the seasonal changes of the tides affect the glacier, in particular the GL? E.g. stronger tides lead to a wider grounding zone.
Detailed comments:
l10: What about the short-term summer speed ups reported by Seehaus et al. 2015 and Seehaus et al. 2016 at Dinsmoor-Bombardier-Edgewoth Glaciers at Sjögren Inlet.
L39: Why is it vulnerable?
L53: Does the velocity field represent the long-term average?
L56: Source of flowlines?
L69: You list publications regarding meltwater lakes from 2017 onwards and say that such studies lead to the identification of GVIIS as a potential site for future ice shelf disintegration, identified in a study from 2013. That’s somehow inconsistent
L85: please explain “seaward extent”. The glaciers are flowing into an ice shelf.
L98: Did you apply any multi-looking or filtering? What about the coregistration of the images? Some more technical information would be nice.
L100: Could you please provide an overview of the used imagery
L103: Did you prove this assumption? You should use your velocity measurements to prove it.
L119: Please describe here briefly how the uncertainty was estimated and what is a “valid pixel”. This would be beneficial for the reader
L122: This information should be provided in section 3.1. and here you can refer to 3.1.
L127: Here you can refer to Friedl et al. 2021 as well. Their study is based on the same satellite data.
L130: what is sigma? The average of all pixels?
L136ff: Unclear explanation. You are using intensity tracking, thus you measure also displacements in azimuth direction and not only in range (LOS) direction. For sure, the shifts in the phase center depth can affect your measurements. But please rephrase this section to be more clear. Did you account for this shift in LOS direction? How much would it be? Any suggestion on how to estimate the bias? A brief statement would be nice at the end of this section.
L152: On which spatial scales did you apply the analysis. Throughout the whole glacier area? Only for the 10km² areas next to the GL? Please clarify
L153: Is this analysis based on the monthly mosaics or single velocity fields?
L165: Do you remove pixels that had no coverage for a specific month or even for single SAR image pairs? Please clarify.
L167: What about very slow-flowing regions? Will they be discarded? (or did you analyze fast-flowing regions only?, see comment above)
L193: feature tracking
Fig.3: Why is the pattern so noisy? Any explanation? Could you also include the glacier numbers in the upper maps?
Fig.4: How did you compute the error bars? How did you compute the mean monthly velocity for the period 2014-2020? Please provide more information or a link to the respective section. Fig. A2 indicates that for several glaciers the availability of monthly means was quite limited (1-3 measurements, e.g. flowline 16, 10 ...) How did you account for this issue in your analysis?
L262: Maybe there was a switch between effective and ineffective subglacial drainage. This might explain the late-summer slowdown. At some other glaciers, a late summer or even March/April minima is also visible.
And the late winter slowdown might be caused by a lack of bed lubrication at all. Well, that is just pure speculation from my side. Some studies at Columbia Glacier or also in Greenland revealed similar patterns. (e.g. Moon et al 2014, Vijay and Braun 2017…)
*L271: Maybe surface melt onset is earlier on Alexander Island as compared to the glacier's origination from the AP. Any correlation with average glacier altitude or surface melt data from climate modeling data?
L287: Why did you apply any interpolation? Just show the pure data.
Fig.6: Please use different colors or line styles to illustrate the SAR derived average velocity contours. How did you generate the heat-map? Please provide more information on how you computed the density. Please do not interpolate the density, if the interpolation is causing such strong artifacts (see comment above).
L291ff: Please show at least one example in Fig.6. Otherwise, it is difficult to figure out this issue.
L320: You should also mention the more recent warming on the AP which overlaps strongly with your observation period (reported by Carrasco et al. 2021). This should be considered in your discussion. There is also a strong surface melt anomaly in 2019/2020 reported by Banwell et al. 2021 on Alexander Island and at least close to the GL next to the AP. So you should consider also the option of surface meltwater as a driver for seasonal fluctuations
L324ff: Please revise and account also for potential surface meltwater availability (see comment above)
Fig. A1: On many panels, the glacier names are covered by black lines.
Fig. B1: Could you please include the central flowlines of the glaciers and glacier numbers.
Fig.C…: what about 2019-2020?
Table D1: Could you also include the most dominant frequency
-
RC2: 'Comment on tc-2022-55', Ted Scambos, 12 May 2022
Review of Boxall et al., The Cryosphere;
Seasonal land ice-flow variability in the Antarctic Peninsula
The paper uses a combination of Sentinel 1 A/B velocity data and ITS_LIVE Landsat-8 velocity data, along with a careful mapping of the grounding line, to assess the scale and extent of a clear seasonal variation in flow speed for glaciers inflowing on both sides of the George VI ice shelf. The paper makes a strong case for the validity of the signal they see, and the seasonality is quite sharp and clear, albeit not large. The authors attribute this to variation in ocean forcing.
This is a well-written, well illustrated and described study that breaks new ground on sensitive detection of seasonal velocity signals (a -few- other studies are out there now for some other regions). However, the attribution of this signal to ocean forcing in untenable. While this means that the paper absolutely needs to be revised, in fact 80% of the paper is ready to go. It is necessary that the paper revise the attribution to discuss the pros and cons of ocean forcing and surface melt percolation to the bed equally. That is, if the following considerations do not convince the authors that surface melting has in fact a far stronger case for this speed-up. I would like to point out that such a conclusion, or preferred but qualified causal process (surface meltwater reaching the glacier bed), would still make this paper a significant contribution to Antarctic glaciology.
The sharpness and regularity of the signal, spanning the entire GVIIS cavity within one or two months (Figure 4 and 5), is the first indication that this is related to summertime melt rather than ocean flow. Peaks in ice flow in December and January are timed closely with peaks in surface melting. Moreover, these timings occur sharply year after year (Figure A1). Nearly all of the glaciers showed a significant spike in 2019-2020, a major melt year for the region. As the paper notes, the -potential- for surface water to induce glacier acceleration is well-proven. It is not essential that the water be visible on the surface as pools (see papers by Harper, Humphrey, Pfeffer; by Koenig, O. Miller, Miégè, Forster.)
On the other hand, oceanographic signals along the Antarctic coast are rarely so sharply seasonal. The cited papers do not (-can- not) discuss seasonal variations in cavity currents or changes in the depth of the CDW layer. The authors infer and favor ocean forcing, but don’t discuss how it would occur – would it be related to sea ice losses? (far more variable and uncertain than the surface melt season) Or wind patterns moving the polar water layer and changing isopycnals in that fashion? (also not reliable enough to provide a signal like Flowlines 2, 3, 4, 5, 17, 18, 19, 20, and 21 in Figure A1). Note that if the change in CDW depth or flux is related to a south-to-north current, the speed required would be an order of magnitude faster than that discussed in Jenkins and Jacobs, 2008 (and it would have to be a continuous laminar flow or wave in the isopycnal).
At the very least the authors need to discuss the two possibilities as equally likely. Personally, the case for summer melt influence is far stronger in my view. However, there are data that might save the ocean discussion: instrumented seals. Data collected by instrumented seals and analyzed by, e.g., Lori Padman or Lars Boheme, might be able to show strong seasonal ocean variations. Have a look at Padman et al., 2012, JGR-Oceans – perhaps in the data used in that paper there is an indication of seasonality (but I don’t think it is mentioned in the paper).
The sharp downturns in the ice velocity just before, or just after, the seasonal speed-up pulse are not easily explained in the ocean scenario.
Also – the authors missed something really cool in the data shown in Figure A1. Look carefully at the signal of Flowline 3 and Flowline 21, and their geographic position. These glaciers are influencing each other across the ice shelf. The earlier acceleration of Flowline 21 (Grotto Glacier, west side, Alexander Island) -slows- the outflow of Flowline 3; then Flowline 3 (Millet Glacier, east side, with a later melt-season peak, perhaps?) accelerates and forces Flowline 21 to slow down. You can see a similar but less clear influence in Flowline 2 and 4, and then Flowline 20 and 19. Like an angry uncle at Thanksgiving, one glacier is shoving the ice shelf table, turkey and all, towards the unsuspecting nephew; the nephew then makes his final point, and shoves the table back toward the uncle. (I suppose it stars with the aunt dumping her drink, meltwater, on each of their heads in succession.)
The last parts of the paper should be re-written with these considerations in mind, but the majority of the paper is publishable as is. I suggest moving some of the figures to the appendices or supplemental information, but overall this is a very well done study, that needs to revise the attribution to a wider perspective at least; if not outright favor surface melt-driven acceleration.
I would like to review the revised paper.
Also, the authors are invited to Thanksgiving at my house.
Detailed comments: Many comments are embedded in the annotated .pdf of the paper, submitted with this review.
Karla Boxall et al.
Karla Boxall et al.
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
499 | 173 | 11 | 683 | 9 | 10 |
- HTML: 499
- PDF: 173
- XML: 11
- Total: 683
- BibTeX: 9
- EndNote: 10
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1