Although this paper has improved from its first iteration, it is still poorly written and requires major revision before it is in acceptable form.
My main complaints are the writing is very unclear in distinguishing behavior reproduced by the model and behavior that is actually observed. There are also numerous logical inconsistencies. For example, there are sentences where something is stated emphatically, and then controverted in the same paragraph.
While I have skimmed the prior reviews, as a new reviewer, I have consciously tried to put them aside and treat this as new submission (there are just too many comments to sort out whether earlier reviewers comments have been appropriately dealt with – I am focusing more on the quality of the revised paper). As such, some of my comments may seem redundant (probably a good indication comments by other reviewers were not fully addressed).
Below I include point by point issues throughout the paper, some of which are minor grammatical things and other which are major flaws.
In particular, the paper should make an effort to explain clearly to the reader why a particular forcing is causing a particular behavior. As a total made up example of how it should read: Warmer temperatures in the model induce more hydrofracturing in the model, causing more frequent calving, which accelerates terminus retreat. As just mentioned, this only mean to
Illustrate how something should be stated. The way such a statement would read in the current manuscript is. Climate produced more terminus retreat (see SI). Readers should not have to refer to the SI to get the gist of what is being said (the SI should have the details of the model, not the main results.)
Specific Comments:
P1L28 “We find that most of the JI retreat during 27 1990-2014 is driven by ocean forcing and bed geometry.” This is more or less what the model shows, but as noted in the paper there are other effects (e.g., meltwater in crevasses) that are not modelled. So a more accurate statement would be
“In our simulations most of the JI retreat during 27 1990-2014 is driven by ocean forcing and the glaciers subsequent response, which is largely governed by bed geometry. Other processes not included in the model (e.g., melt water driving hydrofracture) may also be important.”
P2L2 “We identify two major accelerations that are consistent with observations of changes in 2 glacier terminus.” This paper is not identifying any accelerations. Instead the text should say “Our model simulates two major accelerations….”
P2L7 “And as the slope steepened inland” This is totally unclear. I think what is meant is “as the terminus retreated over a reverse bed slope into deeper water….”
P2L8 “Our model provides evidence that the 1998 and 2003 flow accelerations are most likely initiated by the bed geometry.” This is an example of what I mean by logistical inconsistency. The bed is not doesn’t initiate anything. Instead it governs the response to the ocean or other forcing that initiates (triggers) the retreat.
P2L9 “reproduce” would be a better word than “capture”
P12 “Both modelled and observed results suggest that JI has been losing mass at an accelerated 12 rate, and that JI continued to accelerate throughout 2014.” Other papers have already shown this and this is not a new finding of this paper. Instead it should say something like “Our model is able to simulate the previously observed increase in mass loss through 2014”
P2L16 There is some ambiguity here between ice loss due to discharge and net ice loss (difference between discharge and smb). At a minimum it should say “net ice loss” since discharge has not doubled. Would make sense to include a reference to Enderlyn et al, 2014 here as it is the most current estimate.
P2L25 “Joughin et al, 2004” would be the better reference here because it was the first to covers this time period.
P3L12 “reduction in resistance (buttressing) at the marine front through thinning or retreat of the floating tongue of the glacier.” Much of the time it is the retreat of the grounded (not floating terminus).
P3L28 “climatic forcing and oceanic boundary conditions.” The ocean in large part determine climate. I think there are many places below where climate is used where “atmospheric is actually what is meant. In this case both are forcing – so “ocean and atmospheric forcing” makes this separation clearer.
P4L21 “thickness in the JI basin is computed as the difference between surface and bedrock elevation, which implies that at the beginning of our equilibrium simulation JI’s terminus is considered to be grounded.”
Part of the time Jakobshavn has had a long extended ice tongue, and part of the time it has had a short (<few km) or non-existent tongue (i.e., vertical calving face). There should be a clearer discussion in the main text as to how these two types (vertical vs. horizontal melting). Also in the case mentioned here, what does it mean to be “grounded” (is there melting or not, is there basal friction or not). This is an important aspect of the model and it should clearly stated.
P5L27 “The calving law is known to yield realistic calving” “appears to” rather than “is known” is a better way to put it.
P6L3 “which” should be changed to “that”
Section 2.1.3 There is a lot of detail here, yet the section says very little about how the ocean is represented in the model. A clear description of what the model does is in order (some of the detailed equations could go in the supplement).
P8Section 3: While I realized the parameters are in a table, something more descriptive about what processes are represented by the parameter space should be included.
P8L26” The first speedup is caused by a retreat…” Change to “The first speedup PRODUCED BY THE SIMULATION is caused by a retreat….
P9L4 “Starting in 1992 we obtain a good fit between modelled and observed frontal 4 positions.” Try “Starting in 1992, the modelled and observed terminus positions agree well.”
P9L7 “From 1993 a stronger seasonal 7 velocity signal begins to emerge in our simulation that continues and intensifies in 8 magnitude during 1994 and 1995” It looks like there is sub-annual variability, not necessarily seasonal variability (i.e., the period of the variation is not necessarily close to annual).
P9L14 “The modelled velocities for 1992 and 1995 are consistent with observed” There are only limited velocities during this time and they show constant winter velocity. Since you have just noted sub-annual variability, which observations are too spare to confirm or refute, you need to be clearer here about what the agreement is (for example you mean-annual velocity may agree well with the 1992 and 1995 data).
P9L24 “of the terminus in 1997-1998” add “ice tongue’s” before terminus to remind readers it was floating.
P928These findings are corroborated both by observations 28 (SI, Fig. S15) and modelling results (Fig. 3). Although thinning appears to have 29 increased in our model during three continuous years we find little additional 30 speedup during the period prior to 1998 (Figs. 2, 6, and S7).
Since you have just mentioned both observations and model results in the earlier part of the paragraph, it would we clearer to say “The modelled behavior agrees well with the observations of the observed behavior [CITE REFS]. Rather than “we find little additional 30 speedup during the period prior to 1998” Although thinning appears to have 29 increased in our model during three continuous years , it produced only minor additional speedup during the period prior to 1998 (Figs. 2, 6, and S7).
P9L31 Change “According to…” to “In our…” The way it sounds as written, you are speculating that this is what actually happened, based on inference from the model. A better way to state this would be
In our simulation, JI’s speed increased in the summer of 1998 by ~ 80% relative to the summer of 1992 (Fig. 3), at which time the grounding line position starting starts to retreat thereafter (Figs. 2, 6, and S7). Observations (cite luckman) do not show this level of speedup and there are no observations of the grounding line position at this time with which to assess our model performance.
P10L3”between 1999 and 2002 is in our simulation characterized by a temporal uniform 3 flow, with no episodes of significant terminus retreat” This is not an accurate statement of the behavior shown in figure 3, particularly at “S1”. There is not much apparent trend, but the speed is varying substantially with time.
P10L12 “In the late summer of 2003, an increase in flow velocity is observed (Fig. 3),” don’t use the word “observed” for your model, its too confusing since there are many references to actual observations.
Try “In the late summer of 2003, the simulated flow velocity increases….”
P10L16 “By December 2003 the terminus has retreated back to the position of the 16 grounding line” Would be better to say something like “By December 2003 the modeled ice tongue has completely disintegrated so that the terminus is grounded.
P10L17 Add “simulated” before retreat.
P10L20 “During the final breakup of the ice tongue, JI reached unprecedented 21 flow rates, which in our simulation are as high as 20 km a-1” During the final breakup of the ice tongue, the simulation produces speeds high as 20 km a-1(substantially higher than the sparse observations from that time[REF]).
L10L23 “The velocities…” insert “modelled”
P10L24 replace “seasonal” with “sub-annual”
P10L25 replace “observed at” with “modelled for” (This is a good example where the text is written such its hard to tell modelled from observed behavior – be very clear when you are describing the model, when you are describing data, and when there is and isn’t good agreement between the two).
P10L32 “seasonal” -> “sub-annual”
P11L1 “the terminus remained close to the grounding line” add “(i.e., the terminus remained grounded or only a small floating ice tongue existed)”
P11L7 “In terms of seasonality, our results suggest that most of the seasonal signal in the 7 model is climate driven (see SI, Sect. 1.4 and Fig. S12).” For a concluding sentence, I shouldn’t have to go see these other sections. Basically I really don’t really know here what is meant by “climate driven”, especially since the “seasonal” variation don’t seem seasonal. Elsewhere in the paper a better explanation is given, which is that the coarseness of the grid induce much of this variability (i.e., 4km^2 calving events).
L11L18 Remove all reference to the elastic loading, it really add nothing to the paper. Figure 4 does a nice job of making the point that your reproducing the mass loss. The loading stuff doesn’t add anything new and just adds unnecessary complexity to the paper.
L12L20 “The glacier terminus in 20 1990s is known to have been floating” so remind us again why its not floating in the model (e.g., Figure 6).
L12L20 “but details 21 regarding its thickness are not known.” There are lots of lidar lines on the tongue in this period, from which thickness can be inferred (with some uncertainty).
L12L26 “thickness as the difference between the surface elevation and the bed. This implies that our 26 simulations start with a grounded terminus.”
Why is this being done. If the terminus is grounded, how can the ocean force the glacier in a realistic way. As mentioned above, what basal traction is used (loss of this traction will have an effect on the speedup, which didn’t occur on true tongue since it was floating).
L29 As expected, the difference in 29 geometry results in modelled basal melt rates slightly larger than those obtained by Motyka et 30 al. (2011).
Motyka et al, had something like a m/day, over an area of roughly 5x15 km^2, you have an ocean melt rate that is applied to just a calving face. You purport to be reproducing behavior well pre-1999, but only then has a floating ice tongue developed. Your simplistic assumption is introducing all sorts of unrealisting forcing (i.e., ungrounding of 5x15 km section, instead of loss of an isolated pinning point or sidewall traction on a floating ice shelf).
P13L1 “the model is able to simulate with much accuracy its breakup that occurred in late summer 1 2003 and the subsequent glacier acceleration.”
This statement can hardly hold true since you modeled something completely different during this period than was actually the case (grounded terminus vs floating ice tongue).
P13L3 add “modelled” before “the retreat”
P13L4 “The terminus and the grounding line retreat does not cease after 2010” Modelled or observed????
P13L7 “suggesting that additional feedbacks and/or 7 forcings must continue to disturb the glacier.”
This statement is too strong. It could just be that the bed rock in the model is wrong or the model has other deficiencies, which you acknowledge below. Change to “suggesting that additional feedbacks and/or forcings may affect the glacier. Alternatively, it may represent missing physics, inaccuracies in atmospheric/oceanic conditions, or other various limitations (e.g. bed topography model constraints and 10 grid resolution; see SI, Sect. 1.3 for more details).
P14L11 add “,” before “which”
P14L13 Change “climate” to “atmosphere”
P14L15 “Our results suggest that most of the sub-annual signal in the 15 model is climate driven” This statement should be supported by more than a reference to the SI. It should include a description of what the processes are important and why forcing on annual time-scale produces more rapid variation.
P14L18: “and a simulation with constant climatic forcing (mean 1960-1990 temperature and SMB) 18 indicates that the two accelerations, in 1998 and 2003, are related to bed geometry and ocean 19 melt.” What was it about these two simulations that led to this conclusion (remind us with words or at least call out a result in figure if one exists). Moreover, page 14 is one long paragraph dealing with multiple subjects. Break it up.
P14L20: “Furthermore, our results show that some seasonal velocity peaks could potentially be 20 influenced by the climatic forcing applied (see Figs. S9 and S12(A,B)” This statement seems to relate to the L15 statement I commented on above. So why are the statements separated by a statement related to the long-term forcing (don’t ping pong back and forth, start with a theme, discuss it thoroughly, then move on to the next).
P14L24: “accelerations. The modelled sub-annual signal in terms of terminus retreat and velocities 24 does not always correlate with the observed signal, suggesting that potentially different 25 seasonal forcings (e.g. ice mélange variability, seasonal ocean temperature variability) may 26 influence the advance and retreat of the front at seasonal scales. The” Back to the seasonal cycle, with some explanation (I am not sure if buttresses early explanations or offers alternative).
P14L30 “Furthermore, the 2 km resolution 30 used in this study may not be sufficient to accurately model the seasonal retreat and advance 31 of the front. The smallest calving event in our model is 4 km2, which is larger than most of the 32 calving events observed at JI (see SI, Sect. 1.3.1)”
Now we are getting to what may be driving much of the issue – the poor resolution of the model. Given that 50-runs were done to pick the right model, it seems like some computational resources should be available to run the model at say 0.5 km to test out the sensitivity such large events. Although there have been 2by2km ice bergs, often this is when the ice is floating.
P14 General comment. Start with something like this outline “We see such and such type of seasonal variability. Some of the greater than annual frequency could be an issue with resolution in the model. We examined this sensitivity by doing a few runs at higher resolution. Although some of the variability does appear to be due to this effect, some of it is due to physical processes. In particular, the model shows process A is important because … It also shows process B could be important too. At this point we can’t precisely determine the relative importance of each process. Finally, there could be additional contributions to the seasonality from processes X, Y, and Z” (this likely should take more than one paragraph)
Then start with something like “At the decadal time scale,….something like the above where you discuss the relevant processes, with clear descriptions of the physics revealed by the model (fine put the details in the SI, but the reader should be able to go through and understand what is being said without resorting to the supplement, which is for the nitty gritty details modelers will want to see.
P15L4 Change “determine” to “determined”
P15L10 “and not by an increase in e.g. ocean temperature.” Change to “and not by an increase in variability in ocean temperature.
P15L17 “most likely responds to changes in ocean temperature that are sustained for longer time 17 periods (ADD PARENS WITH TIME SCALE DECADAL, MULTIDECADAL??).
P15L21 “This generated” Add something after “this” such “this increase generated”
L15L19-27. Here you demonstrate that there is a sensitivity to temperature increases of ~1 deg over timescales of a few years. Yet your model reproduces the observed behavior with no variability. Could this be a consequence of the fact that you had 50 runs to choose from?? As I expressed above, how can we believe you have the ocean forcing, when the ice tongue is not properly represented in the model (e.g., in 1998 melt is not applied to a 15-km long tongue). Even if ocean conditions were constant, the were be a lot of variability based on whether there is a tongue or a grounded terminus. Some of this is in the model, but given that Figure 6 shows ice tongue/terminus configurations vastly different than observed, how can we draw firm conclusions from the model????
P16L5 “change, and GPS derived elastic uplift of the crust (Figs. 3 and 5).” Again remove this part as it adds an unnecessary constraint (if you got the observed mass changes right, then you don’t need this).
P16L6 “Our results suggest that most of the JI retreat during 1990-2014 is ocean and bed geometry 6 driven “
Its not really driven by geometry – rather the glacier response to ocean or other forcing is function of its bed geometry. Given the above discussion about how melt is applied – its not really that ocean forcing is well established as a major factor.
P16L7 “and that the overall variability in the modelled horizontal velocities is a response to 7 variations in terminus position.” It should be noted this confirms earlier work that reached this same conclusion”
P16L7 “The seasonal variability observed in our simulations is climate driven.” Like the discussion this is very weak (again, do you mean atmospheric forcing)” Should be worded more like “The seasonal variability observed in our simulations is likely is driven by processes related to atmospheric forcing (e.g., mélange variability, melt induced hydrofracture …).
Page16L16 “provides evidence” should be “reproduces”
Page16L19 “During this period, JI attained unprecedented velocities reaching as high as 20 km a-1.” This is a model result not a statement about what JI did. Should be rewritten as “During this period in the simulation, JI attained unprecedented velocities reaching as high as 20 km a-1, which are considerably higher than observed speeds for this period (CITE PAPERS).
P16L21 “Over the last decade, as the slope steepened inland, sustained high 21 flow rates were observed at JI.” This isn’t really a finding of the model as worded (it restates earlier findings), so cut or make relevant to this study (e.g, the model confirms this finding).
L16L29 “In our model, the terminus retreat is mostly driven by the 29 sub-shelf melting parametrization applied. Thus, our results suggest that ocean forcing is the 30 principal driver for the retreat observed over the last 2 decades.” This seems much strong a statement about the ocean’s role than is provided in the main text (not to mention the problems with how the melt is applied).
L1631 “Further, our model provides 31 evidence that the rapid accelerations of JI in 1998 and 2003 could be triggered by the bed geometry and internal glacier dynamics, and not by a sudden increase in e.g. ocean 1 temperature.” This more or less contradicts the rather strong statement in the model above. Whatever the forcing (ocean or atmospheric), the sensitivity of the glaciers response is governed by the bed geometry (its not a case of ocean/atmosphere or bed). If the ocean is responsible one of two things likely happened. 1) Constant ocean forcing over a long period led to gradual thinning, which caused the glacier (like many others) to eventually retreat of its bedrock high and snap back. This seems more like what your model indicates given you used a steady forcing. 2) The glacier was in a position of precarious stability on a bedrock high and ocean or atmospheric forcing triggered a retreat (your sensitivity studies indicate this could have happened, but the actual simulations). |
- Printer-friendly version
- Supplement
