SUMMARY
This is my second review of this paper, which describes a GPS-based technique to make estimates of basal mass balance (BMB) and surface mass balance estimates for Pine Island Glacier ice shelf (PIGIS). The inclusion of GPS-derived snow surface height relative to the GPS antenna is a particularly interesting addition to standard ice-shelf GPS analyses, and the results are tied in well to other utilized datasets including the stereo-imagery of the DEM and the surface snow/firn state from a RACMO-based firn density model (FDM).
The paper is much clearer than the previous version, due largely to simplifying the set of vertical coordinate variables.
Specific comments are divided into “Major”, where I’d like to know what you did in response, and “Minor”, which don’t require documentation in revision. Numbers refer to original page.line.
-- Laurie Padman
GENERAL COMMENTS
1. I think “PIG shelf” should always be “PIG ice shelf”, in which case you might want to introduce “PIGIS” early and always use it.
2. I don’t like the constructions “between/from 2012-2014” as it requires reading the dash as “and” or “to”. Better to replace the dash with a word, or say “the period 2012-2014”. Similarly for ranges of dimensional values.
3. I’d try to avoid the construction “Figure X shows that …”. It suggests that a pre-existing set of figures is driving the paper, rather than the figures supporting the facts. It’s usually possible to say something like “The elevation at each site (Figure X) …”
MAJOR COMMENTS
3.20-3.22: You almost, but didn’t quite, finish the mass balance statement for the ice shelf itself. If 95-101 GT/yr for basal melting is 70-80% of mass loss from the ice shelf, then calving is 20-30 Gt/yr (although Rignot et al. 2013 says 62 Gt/yr), and SMB is 5 (Rignot). Does that add up to give the observed thinning rate?
4.32-4.34: Both bedrock GPS sites are outside the domain for Figure 1, right? Are they good choices, and are there different penalties for the different sites? Regardless, maybe tell us the typical distance away from the ice shelf GPS for each site.
5.2-5.3: I don’t understand why you subset the 30-s positions to 10-minute intervals: you’re throwing away 95% of the data?
5.11-5.18: This question was asked by both reviewers last time around: Why use a tide model when your data contains the exact tide, and the records are long enough to analyze? I’d expand this to suggest that a better IBE correction might be possible following Padman et al. (2003) rather than just using 1 cm/hPa. The correlation of the height data with the pressure from ERA-Int or the AWS would give you a better model. If you continue to use CATS2008, then a better expression for citing it is “an updated version of the model described by Padman et al. [2002]”
7.22 and elsewhere: I am not familiar with NMAD. (1) You need to tell us something about it (what does it tell you that RMS doesn’t), and (2) does it have units? If so, why “normalized” ?
8.4-8.8: (1) v_fc is fairly important in your study, but there’s not much information here to understand it other than calling it “dry firn compaction”. My possibly wrong interpretation is that it is the height change associated with firn compaction *below* the pole base, and that this rate is set by the amount of new SMB above the pole base. But this needs to be clearer, and I’d also want to know how the choice of a fixed “firn air content” of 12 m relates to what v_fc can be. (2) related: How does all this get us to an estimate of uncertainty on v_fc?
9.3-9.8: This is confusing as, at first thought, the relationship of the pole base to the moving firn-ice transition is not obvious. I think I understand v_fc now (see previous comment), but then the assumption of constant firn air (12 m) creates a relationship between the firn-ice transition and the total mass content of the firn layer, implying that compaction for the firn layer depth range under the pole base is determined by recent precip above the pole base.
9.13-9.14: The density ratio term in eq. (7) explains why you say that BMB is 9 times more sensitive to z_surf than to adot. However, this ignores the scaling of div.u and the relationship between z_surf and adot. Set divergence to zero, on a flat ice shelf with no BMB, and z_surf relates to adot, but the divergence term doesn’t even show up.
10.28-10.29: Not “in the upper few meters of the firn column”, just “in the firn column above the pole base”. Right? What happens *below* the pole base doesn’t affect the antenna to surface distance.
12.10-12.12: This seems a bit disingenuous. The signal is being interpreted entirely as though all the assumptions are correct. You explain later (Section 5) that there are reasons it might be wrong, especially Section 5.1, but the uncertainties need to be addressed briefly here.
13.10-13.17: This is not very convincing (to me). My reading of your paper, without being steeped in the PIGIS literature, is that longitudinal extension might be biased towards transverse basal channels, and that hydrostatic equilibrium of short scales might be being slowly approached over the decade since the ice began to float. Can you really resolve the dynamic term, given the scales of the transverse rifts seen in the right-hand panels of Figure 10?
15.10-15.11: Maybe I don’t understand this all well enough; but in what way is PIG2 SMB “in balance with” ongoing firn compaction and basal melt during this period? In mass, or height? It seems like the balance is that ~2 m per year of fresh firn is deposited, but Table 1 suggests BMB is ~2 or ~4.4 m/yr (depending on method) of ice, and BMB is only ~1 m.w.e. per year, so it’s only a balance in terms of thickness. But then compaction is fixed to only allow 12 m of firn air at all times?
15.16: You can’t say “appear to be uncorrelated”. You either mean just “unrelated” (or “causally unrelated”), or you calculate the correlation and decide if it passes a statistical threshold or not.
16.11-16.31: I recommend rolling these two subsections together, and starting with the discussion of spatial variability so that, when you compare with the two direct measures of BMB, you already have the justification explained.
16.33-17.3: “that appear to display a lagged …”. If this is true, then show it. However, once you start the sentence “Our analysis …”, you seem to be stepping away from accepting the lagged correlation with ocean T. Overall, you seem to set up a belief that the ocean matters and that you have evidence for it, but then say “Actually, no, it’s something else.”
MINOR COMMENTS
1.27: “limited” is unnecessarily vague here.
2.6: “relatively coarse grid”. By most standards, these grids are fairly “fine”. You need to tell us what grid you need, and why.
2.7-2.8: This sentence seems to imply that installing GPS (the topic of this paper) is easier than these are things, but the logistics are similar.
2.10: Why “cumulative” balance?
2.22-2.23: dynamic firn models are forced by a lot more than just SMB.
2.35: “temporally dense” seems complicated: Why not just say “continuous” ? Overall, the issue is whether the single-to-noise combined with sampling characteristics gives you more valuable results than other methods at time *and space* scales you want to resolve.
3.16: This net mass loss (40-50 GT/yr) applies to the entire PIG grounded-ice catchment, right?
4.2: Don’t see the need to hyphenate “ice sheet” and “ice shelf” here
4.27: As expanded upon, “fortuitous” seems like the opposite of what you mean!
4.29: I don’t understand the ‘(co)’
5.8-5.9: What are the “8 km GPS paths”? Do you mean “receiver separations”?
6.30 (but check everywhere): Consistent use of italics for Lagrangian derivative D/Dt. Note that some oceanography texts would say D{\itX}/D{\itt}.
8.21: If rho_ice =917 +/- 5, then how can you use 917 as the threshold for identifying the height of the firn-ice transition? Doesn’t that become a noisy estimate?
8.23: units for ‘d \approx 12 *m*’
9.24: The sign of the shear “dextral (right-handed)” doesn’t tell me anything. All I care about is that velocity is higher towards the center of the trunk flow, right?
10.9-10.10: I really don’t like the format “increased (decreased) … increase (decrease)”. First, it’s hard to read. Second, it’s obvious, right?
11.11: Italics for z_surf etc.
11.19-11.20: I don’t think you mean “scaled” temperatures; you mean “calibrated” temperatures, or local temperatures estimated from the relationship between PIGIS AWS and Evans Knoll.
11.27-11.29: Would have preferred to see a graphic of the historical context for warm periods in a longer time range. Seems important, especially if it figures into total firn state.
14.2: “We now consider the possibility that” => “It is possible that” ?
14.13: “Finally, we assume that …” Out of style with this section, which starts each para with a discussion fo what could go wrong, not what you assumed. So here, maybe “It is possible that the poles tilt over time.”
14.33: My reading is that v_fc represents “firn compaction below the pole base”, not total firn compaction.
16.4-16.5: *Why* is there missing antenna-surface distance data during this period? What causes loss of this valuable information?
17.27: spell out “cal/val”
17.30: “Set GPS elevation mask to 0^o”. Only makes sense if you tell/remind us what the present setting is, and *why* that’s the present setting; i.e., is there a penalty on other measurements when you change the mask specifically for antenna-to-surface heights?
18.3: Does NMAD have units?
FIGURES
See RC1 for some other figure comments.
F.4: This figure would be a good place to define v_fc.
F.6: What does “annualized velocity magnitude” mean, when it clearly isn’t “annualized”?
F.7 and F.9: full-page-width would be good. Especially for Fig. 9, where the text asks us to see short-time-scale events in the 2012-2014 period. |
- Printer-friendly version
- Supplement
