An Integrated View of Greenland Ice Sheet Mass Changes Based on Models and Satellite Observations

Round 1

Reviewer 1 Report

The paper has been much improved. I have no more comments.

Author Response

Reviewer 1:

The paper has been much improved. I have no more comments.

RESPONSE: We thank the reviewer for their positive comments and efforts with reviewing

Reviewer 2 Report

Mottram et al. (2019) provide an overview of methods, results, challenges and opportunities that emerge from a reporting of ECV variables for the Greenland Ice Sheet. This is a difficult task that is well done in this article. Most of the notes below are minor. Section 4.5, is the only one that requires significant editing.  The other main issue is simply that of providing a better visual of the level of detail and as a result the capability of combining these data sets.  The data sets are typically presented only in a ice sheet wide view, which must be done. This does not however illustrate potential value, discrepancies or refinements that various data sets and model approaches provide.  Nor does it illustrate the synergy that can come from a combination of the ECV data sets that could be provided with a local visualization.  This can be best be accomplished by using a single specific outlet glacier example, such as Sermeq Advannarleq focused on each of the ECV variables for that location.  This figure could look like several of the others with multiple panels, but now of that location only for each ECV, though GLL and CFL can be shown on the same and GLL does not apply to all. Similarly given the detailed discussion of Petermannn Glacier why not present the various ECV for that location and there GLL is important.

 

8: “… using essential climate variables (ECV) ice velocity, surface elevation change, grounding line location, calving front location, and gravimetric mass balance.

16: Remove “In”.

30: Is it really just Arctic climate?  I would suggest global given the influences on both sea surface temperatures and jet stream position.  This is authors choice (AC).

69: Is it worth referencing englacial meltwater storgage in glacier ice, such as Kendrick et al (2018) observe?

86: Is it worth using a table to delineate the different sensitivities or are there too many?

98: Remove “though”

120:  Too many “and” usage in sentence. “…long-term trends providing key input for ice dynamic…”

122: What is resolution?

Figure 1: The four annual maps do illustrate the similar pattern of velocity from year to year and the level of detail.  Is it worth including a map, if already in hand, that indicates areas that have a level of interannual variability, not sure what threshold would be used velocity or % velocity.

168:  What percent of the 28 glaciers have this seasonal signal?

Figure 2a:  Would it be possible to overlay IV on Figure 2a, using a specific time such as 2015?  That would give a sense of the spatial resolution of the two CCI.

183:  Why not dates in between? Not asking for inclusion, just explanation.

202: Redundant sentence remove.

206:  It would be worth noting the resultant error or RMSE for the repeat and cross track, least squares approach.

246:  The last two sentences in section can be joined, and if possible quantify what the high level of agreement is.

266: Any specific examples of these differences? Just reference versus discussing in any detail.

Figure 5: The color scheme of daily melt maps that are typically published for Greenland such as at the Polar Portal (DMI) use red for melt and blue for accumulation.  I suggest this is a better color scheme for Figure 5.

346:  Can an overall map show this small surface increase better, than the five year time slice?

358:  Can a specific glacier be used to illustrate this point?  It looks like southeastern GIS outlet glaciers may best illustrate; Gyldenloves Glacier or Heimdal Glacier possibly.

388:  The increase in ice stream velocity and the greater focus, is important to illustrate.  I recommend a closeup of a single ice stream glacier to illustrate this point. At present Figure 6 does not have the resolution for meaningful identification of the significance of this parameter change.

454: Does this suggest that the a SMB model would benefit from including updates of the transient snowline position from MODIS or some other product?

498:  Is this equilibrium enabled by the significant reduction in the floating tongue spatial extent?

503: The other issue is the lack of focus on GLL. The CCI is GLL not ice shelf fracturing.  Time should be taken to briefly describe GLL variation vs the noted parameters of CFL, IV and SMB.  At present the GLL is not presented for the five key glacier noted. 

503:  The discussion of Petermann Glacier and other floating tongues is the only weak portion of this paper.  That is in part because little attention is given to the key to ice loss mechanism on Petermann which is basal melting. The majority of mass loss (80%) at Petermann is from submarine melting beneath the floating ice tongue (Rignot and Steffen, 2008). The melt rates peak at ~25 m/a 10 km downglacier of the grounding line and are accompanied by some interesting basal channels that suggest focused basal melt that can weaken the ice tongue (Muenchow  et al. 2014).  This is noted first in the conclusion line 620.  Should be in section 4.5 somewhere. I recognize that basal melting is not part of the CCI matrix, but given that ice shelf fracture mechanics is not a CCI and is discussed at length, the basal melting cannot be ignored.

590:  is there a percentage of the ice sheet that can be stated as having a significant negative SEC?

608:  Thinning of ice shelves due to basal melt also mechanically weakens them.

 

Kendrick, A. K., Schroeder, D. M., Chu, W., Young, T. J., Christoffersen, P., Todd, J., et al. ( 2018). Surface meltwater impounded by seasonal englacial storage in West Greenland. GRL,45, 10,474– 10,481. https://doi.org/10.1029/2018GL079787

 

Münchow, A., Padman, L., and Fricker, H. A. (2014). Interannual changes of the floating ice shelf of Petermann Gletscher, North Greenland, from 2000 to 2012. J. Glaciol. 60, 489–499. doi: 10.3189/2014JoG13J135

 

Rignot, E., and Steffen, K. (2008). Channelized bottom melting and stability of floating ice shelves. Geophys. Res. Lett. 35:L02503. doi: 10.1029/2007GL031765

Author Response

Reviewer 2:

Mottram et al. (2019) provide an overview of methods, results, challenges and opportunities that emerge from a reporting of ECV variables for the Greenland Ice Sheet. This is a difficult task that is well done in this article. Most of the notes below are minor. Section 4.5, is the only one that requires significant editing.  The other main issue is simply that of providing a better visual of the level of detail and as a result the capability of combining these data sets.  The data sets are typically presented only in a ice sheet wide view, which must be done. This does not however illustrate potential value, discrepancies or refinements that various data sets and model approaches provide.  Nor does it illustrate the synergy that can come from a combination of the ECV data sets that could be provided with a local visualization.  This can be best be accomplished by using a single specific outlet glacier example, such as Sermeq Advannarleq focused on each of the ECV variables for that location.  This figure could look like several of the others with multiple panels, but now of that location only for each ECV, though GLL and CFL can be shown on the same and GLL does not apply to all. Similarly given the detailed discussion of Petermannn Glacier why not present the various ECV for that location and there GLL is important.

RESPONSE: We thank the reviewer for their positive feedback on our article. While we agree that focusing on a single local glacier can give us a better overview of the benefits of using different datasets and models, the aim of the paper is to both give an overview of the total mass budget of the entire ice sheet and show how the ECVs can be used to narrow down different contributions to different processes. We have however considerably edited and restructured section 4.5 and the results from Petermann glacier to give a better overview of how these datasets can be used at a more local level.

8: “… using essential climate variables (ECV) ice velocity, surface elevation change, grounding line location, calving front location, and gravimetric mass balance.

RESPONSE: Added

16: Remove “In”.

RESPONSE: Removed

30: Is it really just Arctic climate?  I would suggest global given the influences on both sea surface temperatures and jet stream position.  This is authors choice (AC).

RESPONSE: As the exact influence of Greenland on global, as opposed to regional, climate is still contentious, we have elected to remain focused on the regional climate

69: Is it worth referencing englacial meltwater storgage in glacier ice, such as Kendrick et al (2018) observe?

RESPONSE: Added

86: Is it worth using a table to delineate the different sensitivities or are there too many?

RESPONSE: We refer interested readers to the references given in the methods sections for more details

98: Remove “though”

RESPONSE: Removed

120:  Too many “and” usage in sentence. “…long-term trends providing key input for ice dynamic…”

RESPONSE: Edited for clarity

122: What is resolution?

RESPONSE: It varies according to time period and location, full details are given within the different datasets on the ESA CCI website.

Figure 1: The four annual maps do illustrate the similar pattern of velocity from year to year and the level of detail.  Is it worth including a map, if already in hand, that indicates areas that have a level of interannual variability, not sure what threshold would be used velocity or % velocity.

RESPONSE: The purpose of figure 1 was to demonstrate the high quality data and excellent spatial coverage now achievable in the CCI dataset rather than show significant results.

168:  What percent of the 28 glaciers have this seasonal signal?

RESPONSE: It’s rather difficult to say as the presence or absence of the ice mélange is quite variable from year to year depending on where in Greenland the glacier is. We have clarified this in the text.

Figure 2a:  Would it be possible to overlay IV on Figure 2a, using a specific time such as 2015?  That would give a sense of the spatial resolution of the two CCI.

RESPONSE: We would rather keep the data products separate in this part of the paper where the methods are being discussed rather than the results.

183:  Why not dates in between? Not asking for inclusion, just explanation.

RESPONSE: The following text explains the requirements for the repeat period. We have clarified this further in the text.

202: Redundant sentence remove.

RESPONSE: Removed

206:  It would be worth noting the resultant error or RMSE for the repeat and cross track, least squares approach.

RESPONSE: Full details of both methods and results are given in Sørensen et al and we have clarified this further

246:  The last two sentences in section can be joined, and if possible quantify what the high level of agreement is.

RESPONSE: Edited for clarity

266: Any specific examples of these differences? Just reference versus discussing in any detail.

RESPONSE: Added reference to Hermann et al., 2018

Figure 5: The color scheme of daily melt maps that are typically published for Greenland such as at the Polar Portal (DMI) use red for melt and blue for accumulation.  I suggest this is a better color scheme for Figure 5.

RESPONSE: This has been a point of contention for some time and there appears to be no standard currently. However, we will take this into account for new plots in coming years.

346:  Can an overall map show this small surface increase better, than the five year time slice?

RESPONSE: Using a shorter time period can introduce anomalies related to noise in the signal which need to be carefully analyses, therefore we prefer to use a 5 year period here.

358:  Can a specific glacier be used to illustrate this point?  It looks like southeastern GIS outlet glaciers may best illustrate; Gyldenloves Glacier or Heimdal Glacier possibly.

RESPONSE: Added

388:  The increase in ice stream velocity and the greater focus, is important to illustrate.  I recommend a closeup of a single ice stream glacier to illustrate this point. At present Figure 6 does not have the resolution for meaningful identification of the significance of this parameter change.

RESPONSE: These results will be the focus of a paper currently in preparation that will focus on the ice sheet modelling results specifically and therefore we have decided not to go into detail here.

454: Does this suggest that the a SMB model would benefit from including updates of the transient snowline position from MODIS or some other product?

RESPONSE: This is a good point and in fact the HIRHAM5 SMB model does include assimilations of MODIS data as given in Langen et al 2017 which we have modified the text to include but in this case the Hermann et al reference show that much of the error in the models comes from overestimates of precipitation at the margins. This is discussed later on in the paper so we have not amended it here.

498:  Is this equilibrium enabled by the significant reduction in the floating tongue spatial extent?

RESPONSE: We have updated this discussion as other studies have found a change in grounding line location and we therefore suggest the apparent stability may be a consequence of a lack of a long time series of data.

503: The other issue is the lack of focus on GLL. The CCI is GLL not ice shelf fracturing.  Time should be taken to briefly describe GLL variation vs the noted parameters of CFL, IV and SMB.  At present the GLL is not presented for the five key glacier noted. 

RESPONSE: Analysis of the GLL data product showed significant variability between different detection algorithms that was difficult to account for physically and suggesting that more work is necessary to refine these in Greenland. However as there are now only 2 remaining ice shelves due to the loss of three of the 5 initial shelves during the course of the CCI project. We therefore felt that putting more time into developing these algorithms was pointless.

503:  The discussion of Petermann Glacier and other floating tongues is the only weak portion of this paper.  That is in part because little attention is given to the key to ice loss mechanism on Petermann which is basal melting. The majority of mass loss (80%) at Petermann is from submarine melting beneath the floating ice tongue (Rignot and Steffen, 2008). The melt rates peak at ~25 m/a 10 km downglacier of the grounding line and are accompanied by some interesting basal channels that suggest focused basal melt that can weaken the ice tongue (Muenchow  et al. 2014).  This is noted first in the conclusion line 620.  Should be in section 4.5 somewhere. I recognize that basal melting is not part of the CCI matrix, but given that ice shelf fracture mechanics is not a CCI and is discussed at length, the basal melting cannot be ignored.

RESPONSE: This is the subject of another paper by some of the same team members and currently in review. We have therefore opted to update the text to reflect these points and we have restructured the section to focus on the subject of the integrated mass balance of Greenland from models and remote sensing observations.

590:  is there a percentage of the ice sheet that can be stated as having a significant negative SEC?

RESPONSE: Added.

608:  Thinning of ice shelves due to basal melt also mechanically weakens them.

 RESPONSE: Good point, we have added this in to the discussion in the results section

Reviewer 3 Report

Review comments on "An integrated view of Greenland Ice Sheet mass changes based on models and satellite observations" by Ruth Mottram et al.

 

General comments:

 

This paper presents recent changes observed in the Greenland ice sheet based on satellite remote-sensing and numerical modeling studies. Satellite observations on ice mass, surface elevation and ice speed are compared with those obtained by numerical models. The results are discussed for the mechanisms of the ongoing changes and possible weakness of the models. The emphasis of the paper is a remote-sensing data set covering the last ~30 years, which is coordinated by the European Space Agency's climate change initiative (ESA CCI). The authors demonstrate the value of the ESA CCI data set, and stress the importance to use the data in combination with numerical modeling works.

I enjoyed reading. The paper covers a broad area of this important research topic, which is useful for those who are interested or engaged in the Greenland ice sheet research but not an expert of the ice sheet mass change. The presented data are not novel, and results are discussed only superficially. Because the paper covers diverse data sets and methodologies, it does not help us to understand the details of each subject. Nevertheless, I understand that the objective of the paper is not to report a new data or findings, but to introduce the ESA CCI data set and review recent advance in the research field. Assuming this kind of a paper is within the scope of the journal, the paper will draw attention from a broad range of journal readers.

I list below my relatively minor concerns and suggestions. I hope they help the authors to revise the manuscript.

 

1. Introduction and Background

I do not see a reason why the first two sections are separated. The sections Introduction and Background both explain general background of the study, previous studies, and the objective of the paper. These are usually summarized in the first section as "Introduction". These two sections have only one big paragraph for each, and they are not well coordinated. For example, important processes (ice speed, surface elevation, grounding line…) are described in the 1st section, before the mechanism of the ice sheet mass change is explained in the 2nd section. I suggest the authors to reorganize these two sections into one, describing in a good order of general background, key mechanism/processes, observational/modeling approach, and the ESA CCI data set overview.

 

2. Calving glacier processes

In the section Result and Discussion, authors nicely presented examples of the ESA CCI data set and compared the observational data with modeling study. This manner gradually deteriorates in the latter part of the section. In "4.4. Calving glacier retreat and ice sheet mass budget", only satellite derived glacier front variations are presented without comparison with a model. Contrary to my expectation from the title, this subsection is not well connected with ice sheet mass budget. I am also confused at the ice shelf fracture processes described in the second half of 4.5. I suppose the authors wanted to show how we can model calving, by using SMB from RCM and satellite derived ice speed. Nevertheless, Figures 10A and 10B refer different time periods, and Figures 10C and 10D are not well explained in the text. The two paragraphs describing this issue (line 504-521 and line 522-547) are disconnected from each other. Moreover, I do not find a connection between the modeling and actual calving or frontal variations of Peterman glacier. My suggestion is to combine 4.4. and 4.5. to describe available satellite data on glacier front and grounding line migrations. The crevasse depth theory can be incorporated if the authors can better explain the model and demonstrate its relevance to satellite data and ice sheet mass budget.

 

3. Abbreviation

The paper uses so many abbreviations, which are not consistently used throughout the text, defined multiple times, and some are not commonly used in scientific papers. Please inspect the text to make sure all abbreviations are used in a consistent manner. Please also consider to reduce abbreviations. For example, I have never seen before ECVs (essential climate variables) and they are used only three times in the text. Also, it is much clearer to use ice velocity instead of IV, and abbreviation is not necessary for such short words, I think. Further, I see sometimes GL (grounding line) in a paper, but never seen GLL (grounding line location) which is used only three times in the text. I want to see an abbreviation only if it is necessary because a phrase is lengthy and appears many times.

 

 

Specific comments:

 

Title: Should be "Greenland ice sheet (lowercase letters)" to be consistent with the main text.

 

Line 2-3: "The cryosphere as a whole …" >> Showing the contribution of the Greenland ice sheet (%) is more useful here.

 

Line 11: "-2.65 m/yr" >> From when to when?

 

Line 16: Delete "Using".

 

Line 21: "Regional climate model" >> RCM is already defined in line 9. I suggest not to use abbreviations in abstract.

 

Line 29: "The Greenland ice sheet" >> Define "GrIS" here.

 

Line 32-33: "ice sheet dynamics" >> I would write "ice dynamics".

 

Line 33: "submarine melt" >> "ice-ocean interaction"?

 

Line 37: "ECV" >> ECVs (I do not like this abbreviation…)

 

Line 49-57: The first and second points sound similar to me. Can you refine the text to make the points clearer?

 

Line 63: "2 K" >> Why not "2 degree C"?

 

Line 71: "surface mass budget (SMB)" >> SMB is defined in line 54-55 as surface mass "balance". Do you distinguish "balance" and "budget" in this paper? Please be consistent to avoid confusion.

 

Line 71-73: "The dynamics component of the mass budget…" >> Is this a common definition of "dynamic component of the mass balance"? I am not sure what do you mean by "dynamic".

 

Line 87: "for example, …" >> Please provide some more examples because there are so many such assumptions.

 

Line 94-96: "Even differences …" >> Can you simplify the sentence as "Even simple choices of ice sheet masks and topographic dataset have significant impact on estimates of SMB."

 

Line 104: "Promice" >> Please define this acronym. Should it be "PROMICE"?

 

Line 120 and 124: Delete "(IV)".

 

Line 125, 129, 137, …: "Sentinel-1 (S1)" >> Please consistently use the abbreviation once defined. I think this abbreviation is not necessary.

 

Line 150: Is the location of [27] correct? I prefer to see the reference at the end of this sentence.

 

Figure 2: Please give (a) and (b) on the plots.

 

Line 186: "Coherence" >> "coherence"

 

Line 186: You need a space between the number "200" and unit "m". Please check throughout the text.

 

Line 199 and 202: Delete "(SEC)".

 

Line 201-202: "… to reflect climate variability and not weather." >> It sounds odd to me. What about "to smooth year-to-year variations."?

 

Line 205: "least-square-methods" >> Please provide a citation.

 

Line 220: Please define CSR, GFZ, JPL and TU.

 

Line 223: "Gravimetrically derived Mass Balance, also known as Gravimetric mass balance…" >> I think these capital letters should be in lowercase.

 

Figure 4a: Please remove the Greenland map showing 8 basins because it is not relevant to the plot. I know the text later refers to the basins, thus I suggest to show this map somewhere else.

 

Line 281: "Ice Sheet Modelling …" >> Please be consistent in capitalization of the section head.

 

Two lines before Equation (2): "SIA" is already defined earlier in the same paragraph.

 

Line 296: "…the fill value for…" >> I do not understand this sentence. Please revise.

 

Line 297: The dot between 2.0 and 10^5 should be replaced by a multiplication sign.

 

Figure 5: Indicate "Surface elevation change" next to the color scale.

 

Line 330: "1 m per year" >> 1 m/yr

 

Line 340: "(see 1)" >> (see Fig. 1)?

 

Line 369: What do you mean by "final dynamics of the ice"? Please clarify.

 

Line 374: "the sliding case" >> the enhanced sliding case?

 

Figure 6. In the figure caption, text for (d) comes after (e) and (f). I would move the plot (d) to the right end and call it (f).

 

Line 393-394: "a linear and quadratic model" >> Do you mean "a linear or quadratic model"? Can you provide a reference?

 

Line 395: "ice-dynamical mass changes included in the GMB products are largely removed." >> Why do you think so? Ice dynamics can cause rapid change in the ice mass. I think the effect of ice dynamics can be discussed by comparing the SMB and gravity derived mass balance.

 

Line 427: "occurs later…" >> occurs later in summer?

 

Figure 8: Please consider showing the locations of these glaciers on a map. A possible option is to prepare a map, which shows drainage basins (inset of Fig. 4a) and glacier locations.

 

Line 467: I think Petermann glacier is a tidewater glacier as well.

 

Figure 9: Please give (a) and (b) to the plots. Please connect the data points in (b) with lines. Indicate in the caption that the plots are from Petermann glacier.

 

Line 479: "basal melt rates" >> Are you talking about basal melting of ice shelves? I think "submarine melting of glacier front" is more relevant to Greenlandic glaciers.

 

Line 485: "tidewater type" >> Is this commonly used to refer tidewater glaciers without floating tongue? I cannot find this term in the reference [27].

 

Line 497-498: "… response to a climate forcing may be delayed by other glaciological processes." >> What kind of processes are you talking about? Please clarify with an example.

 

Line 499: "…recent years…" >> When?

 

Line 516: "the run-off is larger than zero" >> This sounds odd to me.

 

Figure 10: I think it is not useful to define "SIF" as it appears only twice.

 

5. Outlook: Because you several times mentioned the importance of submarine melting of calving glacier fronts, can you write more about the future possibility to use satellite data to study ice-ocean interaction?

 

Line 606-607: "analysis of the strain distribution across the ice shelf…" >> It is not clear from the text if strain rates and surface meltwater are really important in fracture propagation, because no validation was given for the analysis based on the theory. I think the second half of 4.5. should be improved or removed from the text.

Author Response

1. Introduction and Background

I do not see a reason why the first two sections are separated. The sections Introduction and Background both explain general background of the study, previous studies, and the objective of the paper. These are usually summarized in the first section as "Introduction". These two sections have only one big paragraph for each, and they are not well coordinated. For example, important processes (ice speed, surface elevation, grounding line…) are described in the 1st section, before the mechanism of the ice sheet mass change is explained in the 2nd section. I suggest the authors to reorganize these two sections into one, describing in a good order of general background, key mechanism/processes, observational/modeling approach, and the ESA CCI data set overview.

RESPONSE: We thank the reviewer for their thoughtful review which has helped to strengthen the manuscript. We have reorganized and edited along the lines suggested.

2. Calving glacier processes

In the section Result and Discussion, authors nicely presented examples of the ESA CCI data set and compared the observational data with modeling study. This manner gradually deteriorates in the latter part of the section. In "4.4. Calving glacier retreat and ice sheet mass budget", only satellite derived glacier front variations are presented without comparison with a model. Contrary to my expectation from the title, this subsection is not well connected with ice sheet mass budget. I am also confused at the ice shelf fracture processes described in the second half of 4.5. I suppose the authors wanted to show how we can model calving, by using SMB from RCM and satellite derived ice speed. Nevertheless, Figures 10A and 10B refer different time periods, and Figures 10C and 10D are not well explained in the text. The two paragraphs describing this issue (line 504-521 and line 522-547) are disconnected from each other. Moreover, I do not find a connection between the modeling and actual calving or frontal variations of Peterman glacier. My suggestion is to combine 4.4. and 4.5. to describe available satellite data on glacier front and grounding line migrations. The crevasse depth theory can be incorporated if the authors can better explain the model and demonstrate its relevance to satellite data and ice sheet mass budget.

RESPONSE: We have reorganized and merged the two sections as suggested. We have also made more explicit the link to both ice sheet mass balance and the other essential climate variables of the calving front and grounding line datasets. As the modelling is a spin-off of another study, the details of which are given in some depth in a spin-off paper currently in review, we have added a reference to the follow-on study and added some more details on the ice fracture model that give a better overview of the importance of ESA CCI data in determining ice sheet mass loss processes

3. Abbreviation

The paper uses so many abbreviations, which are not consistently used throughout the text, defined multiple times, and some are not commonly used in scientific papers. Please inspect the text to make sure all abbreviations are used in a consistent manner. Please also consider to reduce abbreviations. For example, I have never seen before ECVs (essential climate variables) and they are used only three times in the text. Also, it is much clearer to use ice velocity instead of IV, and abbreviation is not necessary for such short words, I think. Further, I see sometimes GL (grounding line) in a paper, but never seen GLL (grounding line location) which is used only three times in the text. I want to see an abbreviation only if it is necessary because a phrase is lengthy and appears many times.

 RESPONSE: This article was originally submitted to the special issue on ECVs and the abbreviations are commonly used in the documentation of the data products, however we take the point and have edited for clarity!

 

Title: Should be "Greenland ice sheet (lowercase letters)" to be consistent with the main text.

 RESPONSE: Fixed

Line 2-3: "The cryosphere as a whole …" >> Showing the contribution of the Greenland ice sheet (%) is more useful here.

 Line 11: "-2.65 m/yr" >> From when to when?

  RESPONSE: Fixed

Line 16: Delete "Using".

  RESPONSE: Fixed

Line 21: "Regional climate model" >> RCM is already defined in line 9. I suggest not to use abbreviations in abstract.

 RESPONSE: Fixed

Line 29: "The Greenland ice sheet" >> Define "GrIS" here.

RESPONSE: We have decided not to use the abbreviation of GrIS and have removed it from the manuscript

Line 32-33: "ice sheet dynamics" >> I would write "ice dynamics".

 RESPONSE: As the peripheral glaciers are not covered in this dataset we prefer to keep ice sheet dynamics to be clear

Line 33: "submarine melt" >> "ice-ocean interaction"?

 RESPONSE: Fixed

Line 37: "ECV" >> ECVs (I do not like this abbreviation…)

RESPONSE: Fixed. We have also limited the use of ECV in the manuscript

Line 49-57: The first and second points sound similar to me. Can you refine the text to make the points clearer?

RESPONSE: Edited to clarify the difference between using models to infer first and second order mass balance processes.

 

Line 63: "2 K" >> Why not "2 degree C"?

RESPONSE: We prefer to use the Kelvin scale.

Line 71: "surface mass budget (SMB)" >> SMB is defined in line 54-55 as surface mass "balance". Do you distinguish "balance" and "budget" in this paper? Please be consistent to avoid confusion.

RESPONSE: We do not distinguish between surface mass balance and budget and have therefore changed all to be consistent with surface mass balance.

Line 71-73: "The dynamics component of the mass budget…" >> Is this a common definition of "dynamic component of the mass balance"? I am not sure what do you mean by "dynamic".

 RESPONSE: we define the dynamic component on line 72 ad this is the usual definition as it relates to the dynamics of ice flow, however we have clarified that we also mean the ocean contribution too.

Line 87: "for example, …" >> Please provide some more examples because there are so many such assumptions.

 RESPONSE: we have added several more examples

Line 94-96: "Even differences …" >> Can you simplify the sentence as "Even simple choices of ice sheet masks and topographic dataset have significant impact on estimates of SMB."

 RESPONSE: Fixed

Line 104: "Promice" >> Please define this acronym. Should it be "PROMICE"?

 RESPONSE: Fixed

Line 120 and 124: Delete "(IV)".

 RESPONSE: Fixed

Line 125, 129, 137, …: "Sentinel-1 (S1)" >> Please consistently use the abbreviation once defined. I think this abbreviation is not necessary.

 RESPONSE: Fixed

Line 150: Is the location of [27] correct? I prefer to see the reference at the end of this sentence.

RESPONSE: Fixed 

Figure 2: Please give (a) and (b) on the plots.

 RESPONSE: Fixed

Line 186: "Coherence" >> "coherence"

 RESPONSE: Fixed

Line 186: You need a space between the number "200" and unit "m". Please check throughout the text.

 RESPONSE: Fixed

Line 199 and 202: Delete "(SEC)".

 RESPONSE: Fixed

Line 201-202: "… to reflect climate variability and not weather." >> It sounds odd to me. What about "to smooth year-to-year variations."?

 RESPONSE: Fixed

Line 205: "least-square-methods" >> Please provide a citation.

 RESPONSE: Fixed

Line 220: Please define CSR, GFZ, JPL and TU.

 RESPONSE: Fixed

Line 223: "Gravimetrically derived Mass Balance, also known as Gravimetric mass balance…" >> I think these capital letters should be in lowercase.

 RESPONSE: Fixed

Figure 4a: Please remove the Greenland map showing 8 basins because it is not relevant to the plot. I know the text later refers to the basins, thus I suggest to show this map somewhere else.

 RESPONSE: A previous reviewer asked for this plot to be included here. We have decided to keep it.

 Line 281: "Ice Sheet Modelling …" >> Please be consistent in capitalization of the section head.

 RESPONSE: Fixed

Two lines before Equation (2): "SIA" is already defined earlier in the same paragraph.

  RESPONSE: Fixed

Line 296: "…the fill value for…" >> I do not understand this sentence. Please revise.

RESPONSE: Fixed, we have clarified this sentence

Line 297: The dot between 2.0 and 10^5 should be replaced by a multiplication sign.

RESPONSE: Fixed

Figure 5: Indicate "Surface elevation change" next to the color scale.

RESPONSE: Fixed

 

Line 330: "1 m per year" >> 1 m/yr

RESPONSE: Fixed

Line 340: "(see 1)" >> (see Fig. 1)?

RESPONSE: Fixed

Line 369: What do you mean by "final dynamics of the ice"? Please clarify.

RESPONSE: Fixed, we have removed the word final.

Line 374: "the sliding case" >> the enhanced sliding case?

RESPONSE: Fixed

Figure 6. In the figure caption, text for (d) comes after (e) and (f). I would move the plot (d) to the right end and call it (f).

RESPONSE: We have reorganized and restructured this caption to make it easier to follow

Line 393-394: "a linear and quadratic model" >> Do you mean "a linear or quadratic model"? Can you provide a reference?

RESPONSE: We have reorganized and restructured this paragraph to make it easier to follow

 Line 395: "ice-dynamical mass changes included in the GMB products are largely removed." >> Why do you think so? Ice dynamics can cause rapid change in the ice mass. I think the effect of ice dynamics can be discussed by comparing the SMB and gravity derived mass balance.

RESPONSE: We have reorganized and restructured this paragraph and include discussion on this point now

Line 427: "occurs later…" >> occurs later in summer?

RESPONSE: Yes. Fixed

Figure 8: Please consider showing the locations of these glaciers on a map. A possible option is to prepare a map, which shows drainage basins (inset of Fig. 4a) and glacier locations.

 Line 467: I think Petermann glacier is a tidewater glacier as well.

 RESPONSE: Classically a tidewater glacier does not have a large ice shelf like Petermann glacier has

Figure 9: Please give (a) and (b) to the plots. Please connect the data points in (b) with lines. Indicate in the caption that the plots are from Petermann glacier.

RESPONSE: Fixed

Line 479: "basal melt rates" >> Are you talking about basal melting of ice shelves? I think "submarine melting of glacier front" is more relevant to Greenlandic glaciers.

RESPONSE: yes, we have clarified this in the text

 Line 485: "tidewater type" >> Is this commonly used to refer tidewater glaciers without floating tongue? I cannot find this term in the reference [27].

 RESPONSE: yes

 

Line 497-498: "… response to a climate forcing may be delayed by other glaciological processes." >> What kind of processes are you talking about? Please clarify with an example.

RESPONSE: fixed with more examples.

Line 499: "…recent years…" >> When?

RESPONSE: fixed 

Line 516: "the run-off is larger than zero" >> This sounds odd to me.

 RESPONSE: fixed

Figure 10: I think it is not useful to define "SIF" as it appears only twice.

RESPONSE: fixed

5. Outlook: Because you several times mentioned the importance of submarine melting of calving glacier fronts, can you write more about the future possibility to use satellite data to study ice-ocean interaction?

 Line 606-607: "analysis of the strain distribution across the ice shelf…" >> It is not clear from the text if strain rates and surface meltwater are really important in fracture propagation, because no validation was given for the analysis based on the theory. I think the second half of 4.5. should be improved or removed from the text.

RESPONSE: More detailed analysis of these two points is given in a separate paper currently in review. We have expanded and reorganized this section however to clarify the important processes and added a reference to the hopefully soon to be forthcoming paper

Reviewer 4 Report

I read the paper with interest, however in some sections I feel little confused. I believe authors may explain that parts clearly dispelling my doubts.

I mainly concern about used in the study the Parallel Ice Sheet Model (PISM). As far as I understand that approach from Ed Bueler and Jed Brown, it is ice dynamic model capable of performing simulations for whole ice sheet scale. It is however producing realistic ice velocities for ice sheets where bed gradients are “modest” as authors said, and further referring to NE Greenland. I wonder if this model will perform well for the entire Greenland, since the bed slopes on other part of the coast varies form relatively high to very mild. In that case the strass may not be always vertical-dominated. Please discuss it shortly, to avoid misunderstanding of your approach.

I think it will be also important to provide information about the coordinate system for ice velocities: please add what you mean for u1 and u2, as well as v1 and v2.

Also I wonder what made authors to use parameters: E = 1.5 and n = 3.0 for both the SIA and the SSA case.

Author Response

I read the paper with interest, however in some sections I feel little confused. I believe authors may explain that parts clearly dispelling my doubts.

I mainly concern about used in the study the Parallel Ice Sheet Model (PISM). As far as I understand that approach from Ed Bueler and Jed Brown, it is ice dynamic model capable of performing simulations for whole ice sheet scale. It is however producing realistic ice velocities for ice sheets where bed gradients are “modest” as authors said, and further referring to NE Greenland. I wonder if this model will perform well for the entire Greenland, since the bed slopes on other part of the coast varies form relatively high to very mild. In that case the strass may not be always vertical-dominated. Please discuss it shortly, to avoid misunderstanding of your approach.

I think it will be also important to provide information about the coordinate system for ice velocities: please add what you mean for u1 and u2, as well as v1 and v2.

Also I wonder what made authors to use parameters: E = 1.5 and n = 3.0 for both the SIA and the SSA case.

RESPONSE: We thank the reviewer for their response. PISM is a large scale model for the whole ice sheet scale, but still able to include vertical shearing as well as membrane stresses. It was demonstrated by Aschwanden et al 2016 that PISM is able to capture the complex flow pattern of outlet glaciers in Greenland. We use their best fit recommended parameters for the simulations presented here.

Reviewer 5 Report

This review paper gives an overview over the results of the ESA Greenland Ice Sheet CCI project and compares the results to regional climate models (RCM) and ice sheet models to improve the understanding of the processes that cause the observed changes. The manuscript starts with an introduction into the CCI project and the background of observations and models of the Greenland Ice Sheet (GIS). In the following section, the datasets used in this study are described. This comprises the different observations of essential climate variables (ECV), which quantify the response of the ice sheet to a changing climate, as well as RCM and ice sheet models. Section 4 compares the data and the models and discusses the conclusions that can be drawn from this investigation. At the end of the manuscript, an outlook is given for a continuation and extension of the observations of the ECVs in the CCI+ project as well as some improvements which can be expected for the models.

The manuscript gives a very good overview over the different observables of ECVs and how they compare to recent modeling results. However, I see a remaining issue in precisely describing, what the data and methods show and how they interact. Section 2 gives an introduction into the mass balance of the GIS, the processes involved and why modeling is necessary here. In this general overview, however, I missed the connection to the five ECVs to the processes that cause changes in the mass balance. I would suggest to explain here (very briefly), why each ECV was chosen and how these observations contribute to the understanding of the processes. Furthermore, as the introduction starts with the contribution of the GIS to sea level, it should be explained why e.g. the calving front location (which describes already floating ice and does not anymore directly contribute to sea level) is still important for the mass balance of the GIS. Furthermore, at the end of the paper, it is stated several times that the manuscript gives an overview over the current mass budget of the GIS. This study summarizes observations and models, contributing to this mass budget, but does not contain the total mass budget (besides GRACE, which is, however, not discussed towards this point).

There are several minor issues, explained in more detail in the following comments. After these points are addressed, I think the manuscript is suitable for publication.

Detailed comments:

l.16: Remove “Using”

l.54: An explanation of “second order processes” would be useful. Comparing mass balances from RCM and GMB immediately arises the question how they can be compared without any information about ice dynamic changes. I would suggest to mention that this is a comparison of the short to intermediate term variations of the mass balance, excluding the long term signal.

l.63: The dynamic component of mass balance, which is relevant for sea level, is the ice flux over the grounding line. Iceberg calving and submarine melting surely influences this flux, but have no direct effect on the sea level as they affect the floating ice. This should be formulated more precisely.

l.72: See comment above.

Fig.2: How is the distance defined (distance from what)? Does 0km refer to the grounding line or is this just a randomly selected point?

l.172: What makes those 5 glaciers “key glaciers”? As this is an ECV to study the state of the GIS, it should be explained why this parameter has not been obtained for other glaciers or ice streams of Greenland.

l.200: The 5-year running mean over 1992-2017 implies (as it is also stated on the CCI website) “an unbroken time series from then to present”. However, following the description in Sandberg Sørensen et al. (2018), this is not the case. According to Tab. 1 therein, the running mean covers 1992-1996, 1993-1997, 1994-1998 and so on until 2007-2011. However, the next interval, obtained from CryoSat-2, is 2011-2015. Referring the trend to the central epoch of this five year interval means a gap between 2009 and 2013. This is especially critical for Greenland due to the strong melt season in 2012, which is not covered entirely using this method.

l.205: The manuscript describes the SEC product as “combination of cross-over-, repeat-track-

and least-squares-methods”. I would suggest to remove the “least-squares”, as this is in general the same as the repeat-track-method. The only difference here is the choice of some of the parameters and how the observation for this fit are selected.

l.211: I think, describing the products correctly should also include, that for SEC from conventional pulse-limited radar altimetry (all data before 2011), was masked out where the topography slope was > 1.5°. This is an important detail to this dataset, as it means, that there exist no results in the very margins, where the largest elevation changes would be expected.

Eq. (1): “n” is not explained in the text.

l.322: “makes for challenging evaluation” check wording

l.330: Please consider, that the results in the margins (>1.5°) have been masked out in the SEC.

l. 348: “modelled SMB model” check wording + New sentence after l.347.

l.351: If I understand it right, after the spin up to equilibrium, the ISM is driven by the recent SMB. So I think, if the SMB does not contain any long term changes, it not not surprising that also the ISM does not see any changes.

l.357: This comparison always takes the observed SEC as the truth. However, it should at least be briefly discussed, to what amount also uncertainties in the SEC observations could be responsible for the mismatches.

Fig.6: This figure is quite hard to read. I do not understand, why different color scales have been chosen for the different plots. A logarithmic color scale from 0.01-10 m/day (as for a,b and e) can also be used to display the differences. This would allow an easier comparison between the magnitudes of the data and the differences. I also suggest to change the labels of the difference plots from “Diff” to “Obs. - Low sliding” and so on, making the figure easier to understand.

l.390: This section does not deal with a total mass budget, as implied by the heading. I suggest to change it to “mass budget variations”. (the same applies to the beginning of the text on l.391)

l.392: Which version of the GRACE data? DTU, TUDR or something else? No final GRACE product is presented in Sect. 3.5.

l.394: What was the reason for different reference periods? Why was this “long-term signal” not calculated over the same reference period, covered by both observations?

Fig.7: The caption should also note, that this figure shows no “mass changes” but “variations of mass changes excluding a long-term signal”.

Fig.8: How is this rate of retreat calculated? Is it a mean rate, fitted through all epochs of CFL or is it calculated from the first and last epoch only? As discussed in the text, this can make a large difference.

l.463: Please give at least some examples for these “multiple processes”.

Fig.9: The glacier name should be mentioned in the caption of the figure. I also suggest to use a gradual color scale for the left subfigure. This random (and sometimes almost repeating) selection of colors makes it impossible to distinguish the temporal evolution of the calving front.

l.500: This conclusion does not agree with Rückamp et al (2019;https://doi.org/10.1029/2018JF004775). They find another fracture further upstream, suggesting another large calving event in the near future.

l.503: I guess, this should be Fig.10.

l.549: I do not agree that this paper gives “an overview of the current mass budget of the Greenland

ice sheet”. It gives an overview over different ECVs and models, describing different processes contributing to the total mass budget. However, the only place in the paper where the total mass budget occurs is Fig. 4a, but this figure is not discussed with respect to the mass budget trend.

l.576: The same as above applies here. The study summarizes the present state of observations and models of processes affecting the mass budget but is does not summarize the budget itself.

l.578: The number of -255±15Gt/yr appears in the abstract and the conclusion but nowhere else in the manuscript. There is no explanation how this rate is derived from GRACE. Even more important, there is no discussion how the uncertainty estimate is obtained w.r.t influence of GIA, filtering or other sources of errors.

l.605: See comment to l.500.

p { margin-bottom: 0.1in; direction: ltr; color: rgb(0, 0, 0); line-height: 115%; }p.western { font-family: "Liberation Serif", "Times New Roman", serif; font-size: 12pt; }p.cjk { font-family: "AR PL SungtiL GB"; font-size: 12pt; }p.ctl { font-family: "Lohit Devanagari"; font-size: 12pt; }a:link { }

Author Response

The manuscript gives a very good overview over the different observables of ECVs and how they compare to recent modeling results. However, I see a remaining issue in precisely describing, what the data and methods show and how they interact. Section 2 gives an introduction into the mass balance of the GIS, the processes involved and why modeling is necessary here. In this general overview, however, I missed the connection to the five ECVs to the processes that cause changes in the mass balance. I would suggest to explain here (very briefly), why each ECV was chosen and how these observations contribute to the understanding of the processes.

RESPONSE: This is a good suggestion and we have added a couple of sentences.

 Furthermore, as the introduction starts with the contribution of the GIS to sea level, it should be explained why e.g. the calving front location (which describes already floating ice and does not anymore directly contribute to sea level) is still important for the mass balance of the GIS.

RESPONSE: We have clarified this in the text, changes in calving front position generally indicates that the glacier is out of balance with current climate and observing the position over a long period can indicate by how much and in what direction (positive or negative) the balance is.

 Furthermore, at the end of the paper, it is stated several times that the manuscript gives an overview over the current mass budget of the GIS. This study summarizes observations and models, contributing to this mass budget, but does not contain the total mass budget (besides GRACE, which is, however, not discussed towards this point).

RESPONSE: We have added a short paragraph including these points in the final section and brought this into the main text in other sections to emphasise the point in the paper about determining the mass budget of the ice sheet.

There are several minor issues, explained in more detail in the following comments. After these points are addressed, I think the manuscript is suitable for publication.

 

Detailed comments:

 

l.16: Remove “Using”

RESPONSE: Fixed

l.54: An explanation of “second order processes” would be useful. Comparing mass balances from RCM and GMB immediately arises the question how they can be compared without any information about ice dynamic changes. I would suggest to mention that this is a comparison of the short to intermediate term variations of the mass balance, excluding the long term signal.

RESPONSE: This is an excellent point and we have clarified this further in the text.

l.63: The dynamic component of mass balance, which is relevant for sea level, is the ice flux over the grounding line. Iceberg calving and submarine melting surely influences this flux, but have no direct effect on the sea level as they affect the floating ice. This should be formulated more precisely.

RESPONSE: We have clarified this in the text and added a discussion on the dynamical effects of floating ice shelf loss in section 4.5

l.72: See comment above.

Fig.2: How is the distance defined (distance from what)? Does 0km refer to the grounding line or is this just a randomly selected point?

RESPONSE: The distance is from a point selected to be easily identifiable on all imagery.

l.172: What makes those 5 glaciers “key glaciers”? As this is an ECV to study the state of the GIS, it should be explained why this parameter has not been obtained for other glaciers or ice streams of Greenland.

RESPONSE: These 5 glaciers are the only ones with permanent floating ice shelves in front of the glaciers. We have stressed this point in the text.

l.200: The 5-year running mean over 1992-2017 implies (as it is also stated on the CCI website) “an unbroken time series from then to present”. However, following the description in Sandberg Sørensen et al. (2018), this is not the case. According to Tab. 1 therein, the running mean covers 1992-1996, 1993-1997, 1994-1998 and so on until 2007-2011. However, the next interval, obtained from CryoSat-2, is 2011-2015. Referring the trend to the central epoch of this five year interval means a gap between 2009 and 2013. This is especially critical for Greenland due to the strong melt season in 2012, which is not covered entirely using this method.

 

l.205: The manuscript describes the SEC product as “combination of cross-over-, repeat-track-and least-squares-methods”. I would suggest to remove the “least-squares”, as this is in general the same as the repeat-track-method. The only difference here is the choice of some of the parameters and how the observation for this fit are selected.

RESPONSE: fixed

l.211: I think, describing the products correctly should also include, that for SEC from conventional pulse-limited radar altimetry (all data before 2011), was masked out where the topography slope was > 1.5°. This is an important detail to this dataset, as it means, that there exist no results in the very margins, where the largest elevation changes would be expected.

RESPONSE: we have added this to the text

Eq. (1): “n” is not explained in the text.

RESPONSE: fixed

l.322: “makes for challenging evaluation” check wording

RESPONSE: fixed

l.330: Please consider, that the results in the margins (>1.5°) have been masked out in the SEC.

RESPONSE: fixed

l. 348: “modelled SMB model” check wording + New sentence after l.347.

RESPONSE: fixed

l.351: If I understand it right, after the spin up to equilibrium, the ISM is driven by the recent SMB. So I think, if the SMB does not contain any long term changes, it not not surprising that also the ISM does not see any changes.

RESPONSE: this is actually the point of this section, that SMB does drive the short-term modelled surface elevation except for some areas where ice dynamics are more important.

l.357: This comparison always takes the observed SEC as the truth. However, it should at least be briefly discussed, to what amount also uncertainties in the SEC observations could be responsible for the mismatches.

RESPONSE: We have added some lines to explain uncertainties on the SEC data

Fig.6: This figure is quite hard to read. I do not understand, why different color scales have been chosen for the different plots. A logarithmic color scale from 0.01-10 m/day (as for a,b and e) can also be used to display the differences. This would allow an easier comparison between the magnitudes of the data and the differences. I also suggest to change the labels of the difference plots from “Diff” to “Obs. - Low sliding” and so on, making the figure easier to understand.l.390: This section does not deal with a total mass budget, as implied by the heading. I suggest to change it to “mass budget variations”. (the same applies to the beginning of the text on l.391)

RESPONSE: We have changed the heading to Total ice sheet mass budget down to the basin scale

l.392: Which version of the GRACE data? DTU, TUDR or something else? No final GRACE product is presented in Sect. 3.5.

RESPONSE: fixed with a reference to the TU Dresden data used in this analysis

l.394: What was the reason for different reference periods? Why was this “long-term signal” not calculated over the same reference period, covered by both observations?

RESPONSE: We have clarified this section further

Fig.7: The caption should also note, that this figure shows no “mass changes” but “variations of mass changes excluding a long-term signal”.

RESPONSE: fixed

 

Fig.8: How is this rate of retreat calculated? Is it a mean rate, fitted through all epochs of CFL or is it calculated from the first and last epoch only? As discussed in the text, this can make a large difference.

RESPONSE: Lines added explaining this in the caption

l.463: Please give at least some examples for these “multiple processes”.

RESPONSE: fixed

Fig.9: The glacier name should be mentioned in the caption of the figure. I also suggest to use a gradual color scale for the left subfigure. This random (and sometimes almost repeating) selection of colors makes it impossible to distinguish the temporal evolution of the calving front.

RESPONSE: fixed

l.500: This conclusion does not agree with Rückamp et al (2019;https://doi.org/10.1029/2018JF004775). They find another fracture further upstream, suggesting another large calving event in the near future.

RESPONSE: We have clarified and updated this section, also with regard to a paper currently in review from some of the co-authors that looks at fracture patterns on Petermann glacier more closely.

l.503: I guess, this should be Fig.10.

RESPONSE: fixed

l.549: I do not agree that this paper gives “an overview of the current mass budget of the Greenlandce sheet”. It gives an overview over different ECVs and models, describing different processes contributing to the total mass budget. However, the only place in the paper where the total mass budget occurs is Fig. 4a, but this figure is not discussed with respect to the mass budget trend.l.576: The same as above applies here. The study summarizes the present state of observations and models of processes affecting the mass budget but is does not summarize the budget itself.l.578: The number of -255±15Gt/yr appears in the abstract and the conclusion but nowhere else in the manuscript. There is no explanation how this rate is derived from GRACE. Even more important, there is no discussion how the uncertainty estimate is obtained w.r.t influence of GIA, filtering or other sources of errors.

RESPONSE: We agree that these points were left too implicit and have added a paragraph in section 4.3 that describes the total mass budget figure, where it comes from and how the rest of the datasets can be related to this. The references and discussion in section 3.5 do however give a full description of the derivation of the budget number and the associated uncertainties and relevant processes. The Mass budget of the ice sheet is thus much more explicitly related to the different datasets and processes presented here.

l.605: See comment to l.500.

RESPONSE: see above

 

 

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

Review of “An integrated view of Greenland Ice Sheet mass changes based on models and satellite observations” by Mottram et al.

 

Mottram et al. review recent changes in Greenland Ice Sheet mass using an impressive variety of satellite remote sensing, regional climate and ice sheet models. The paper provides a useful review of the data and techniques used to understand how the ice sheet is responding to natural variability and long-term climate change. There are some interesting discussion points that raise important questions about the use of SMB models for forecasting Greenland Ice Sheet mass. I think, however, that some of the conclusions are too strong. This is mainly due to the limitations of PISM for investigating dynamical surface elevation changes. I encourage the authors to tone down some of these conclusions before publication of the manuscript.

Main comment

Use of PISM to partition role of ice dynamics vs. surface mass balance on elevation change. Unless I am missing something, the model does not include calving front change so cannot account for the retreat and acceleration observed at some of Greenland’s major outlet glaciers during the 21st century, where most of the dynamical surface elevation changes are occurring (Fig. 6; top right). The authors admit this limitation but only after they conclude that surface mass balance dominates overall elevation change (repeated in the abstract). While I think this statement may be justified in the slow-flowing interior, I don’t think the authors can rule out the role of dynamics for surface elevation change at the margins and should think about adjusting their conclusions accordingly.

Below I detail some specific comments which I hope the authors will find useful.

Background

L57-60: Long, convoluted sentence. The way its worded suggests precipitation is responsible for sea-level rise when really runoff and calving are the two dominant processes. Suggest rewording.

L65: I agree that there is significant uncertainty in model estimates but this paragraph does not describe them. Suggest going into more detail about processes which are complex or known to be poorly modelled. This should set the next sections of the paper up better.

Methods

L90: First time “IV” is used, define. What is meant by “unprecedently dense”, in comparison to what? Suggest rephrasing.

L140: “for determining”

L160: Mention the sensors used to measure this variable.

Figure 4: Confirm that this is showing difference between 2016 and 2017?

L195-202: Suggest moving this into next section.

Results

L263: I don’t know how this conclusion can be justified. In the upper panel there are large areas of elevation gain in the interior and elevation loss at the margins of West Greenland that appear to not be captured by the SMB model.

L272-282: OK now I understand.

L284-297: The statement about SMB dominating elevation changes critically depends on this paragraph. Suggest moving this section to the start of the results.

L307-310: This conclusion is misleading since you cannot rule out dynamical processes in this basin.

Fig. 8: I would be interested to see how 2012 compares for Greenland Ice Sheet and Basin 6. Suggest increasing the y-axis limits.

Fig, 10: Can’t see half of it.

L384: Edit reference to Benn 2007

 

Author Response

Please see attached file

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper provides an overview of the current state of the mass budget of Greenland based on satellite gravimetry and remote sensing observations of surface elevation change, ice sheet velocity and calving front positions. It is well-written and organized in good format. My main concern is that the contribution of this study should be better highlighted.

General comments:

ESA CCI has done excellent work to gather together valuable datasets for Greenland, which is the basis for this paper. As such, I suggest the additional new contribution of this paper should be claimed explicitly (what new knowledge can we obtain from this paper?). Also, I suggest that the authors reconsider the strong conclusions made in this paper.

Specific comments:

1. Abstracts, line 8, is it proper to say "poorly understood ice sheet dynamical and surface mass processes", given lots of efforts have been paid on this topic? line 4, explicitly describe the sensitivity. line 16, perhaps add the number before 2002 for comparison?

2. Key words: "Essential Climate Variables" is not a common word for me.

3. Lines 33-37, therefore, the contribution of this study should be better illustrated.

4. Lines 38-39, "evaluate the importance of different ice sheet processes", this is very strong...I suggest the authors reconsider this claim.

5. Lines 42, "to correct and refine estimates of physical processes", also very strong description.

6. Lines 38-52, there are several strong descriptions in this paragraph. I suggest that the authors should reconsider these descriptions and evaluate the contribution of this study properly.

7. Line 54, what is "the ice sheet contained on the island"?

8. Line 62, is it necessary to evaluate the relative importance of these two components?

9. Line 108, make Figure 1 bigger.

10. Line 113, it will be interesting to add 1-2 sentences to compare the two datasets.

11. Lines 123-124, it may be useful to add a table showing the spatial resolutions of these images.

12. Figure 2b, it is difficult to understand the y-axis (distance)...Distance from where?

13. Line 139, need citation.

14. Line 146, "two distinct epochs"?

15. Line 157-158, why?

16. Figure 3, add north arrow.

17. Figure 5, DTU and TUDR GMB products are very different. A brief explanation may be useful.

18. Lines 195-202, this paragraph does not talk about GRACE and should be removed to Section 3.6. Additionally, different SMB models perform differently for short-term surface runoff simulations, as shown in Smith et al., 2017, PNAS.

19. Line 225, perhaps add a comment to compare PISM with ISSM?

20. Line 266-267, I did not get the point. Perhaps highlight in Figure 6?

21. Lines 272-273, add the numbers.

22. Line 285, how are the velocities simulated? PISM? perhaps this should be explained in the method section.

23. Line 283, Section 4.2, what is the point to compare modelled and observed ice flow velocities? I think it may be more useful to analyze the spatial pattern and temporal trend of observed ice flow velocities, or to compare different ice flow velocity datasets?

24. Figure 8, inset in Figure 5 is too small.

25. Lines 316-320, overestimate precipitation but underestimate melt rate? This is difficult to understand. Smith et al validated SMB models with field-measured surface runoff and showed that SMB models overestimate surface melt rate although this study was conducted during a short period.

26. Lines 327-328, this relationship is not straightforward to me.

27. Lines 334-336, this relationship is not straightforward to me.

28. Figure 8, will some metrics be useful to demonstrate their relationships? e.g., Nash–Sutcliffe model efficiency coefficient.

29. Line 344, add the penetration depth.

30. Figure 9 is not easy to follow.

31. Figure 11 (right), y axis indicates distance from XXX?

32. Lines 379-381, the explanation is too general.

33. Line 384, typo?

34. Lines 384-386, need a citation.

35. Figure 12, resolution is too low.

36. Line 441, Noël et al., 2016, TC presented 1 km SMB model.

37. Lines 427-442, you mentioned field observations are important in the main text but did not demonstrate this point in the Outlook section.

38. Lines 443-474, see my general comments. I suggest the conclusion section should be revised to: 1) better highlight the additional new contribution of this study; and 2) illustrate these new contributions properly (not too strong).

Author Response

REVIEW 2

This paper provides an overview of the current state of the mass budget of Greenland based on satellite gravimetry and remote sensing observations of surface elevation change, ice sheet velocity and calving front positions. It is well-written and organized in good format. My main concern is that the contribution of this study should be better highlighted.

General comments:

ESA CCI has done excellent work to gather together valuable datasets for Greenland, which is the basis for this paper. As such, I suggest the additional new contribution of this paper should be claimed explicitly (what new knowledge can we obtain from this paper?). Also, I suggest that the authors reconsider the strong conclusions made in this paper.

RESPONSE: We thank the reviewer for their careful reading and thoughful comments on the manuscript. We have restructured the conclusions and the Introduction to better focus the paper on the new results from the CCI project . We respond to the specific points below. 

:

Specific comments:

1.     Abstracts, line 8, is it proper to say "poorly understood ice sheet dynamical and surface mass processes", given lots of efforts have been paid on this topic?

RESPONSE: We agree that there has been a lot of work into these topics, but there still exist biases and uncertainties in models of ice sheet dynamics and surface mass balance that we shed some light on in this paper. We have therefore updated the sentence to: “We also combinethese essential climate variables with output from a regional climate model (RCM) and from an ice sheet model (ISM) to gain insight into biases in ice sheet dynamics and surface mass budget processes.

2.      line 4, explicitly describe the sensitivity. line 16, perhaps add the number before 2002 for comparison?

RESPONSE: We have added the numbers requested.

2. Key words: "Essential Climate Variables" is not a common word for me.

RESPONSE: This was included as it is a special issue focused around ESA’s essential climate variables, however a quick google scholar search shows it is indeed not a common keyword so we have removed it.

3.     Lines 33-37, therefore, the contribution of this study should be better illustrated.

RESPONSE: As part of our restructuring of the introductory paragraphs we have emphasised the added scientific value of this study.

4.     Lines 38-39, "evaluate the importance of different ice sheet processes", this is very strong...I suggest the authors reconsider this claim.

RESPONSE: We have expanded the sentence slightly to emphasise that by different ice sheet processes, we mean the relative contribution of ice sheet dynamics, surface mass budget and submarine melt to the total mass budget and to assist in explaining observations.

5.     Lines 42, "to correct and refine estimates of physical processes", also very strong description.

RESPONSE: This has been changed to “to estimate the contribution of different physical processes to mass budget”

6.     Lines 38-52, there are several strong descriptions in this paragraph. I suggest that the authors should reconsider these descriptions and evaluate the contribution of this study properly.

RESPONSE: We are not quite clear what this comment refers to but we have in any case given more detail in this paragraph explaining how the models and observations can be used in different ways to get better estimates of the state of the ice sheet

7.     Line 54, what is "the ice sheet contained on the island"?

Response: deleted for clarity.

8.     Line 62, is it necessary to evaluate the relative importance of these two components?

RESPONSE: We have added the following sentence to justify the importance of partitioning ice sheet mass loss. “The relative contribution to the total mass budget of these two components is important as it determines both the regional sea level rise fingerprint and the rate at which the ice sheet can respond to further climate change.” 

9. Line 108, make Figure 1 bigger.

RESPONSE: Figure has been enlarged.

10. Line 113, it will be interesting to add 1-2 sentences to compare the two datasets.

RESPONSE: We added the following sentence: “A pixel-by-pixel intercomparison with the latest available winter campaign map 2017/18 shows a mean bias of 0.001 m/d and a RMSE of 0.04 m/d for both easting and northing components based on more than 33 million pixels.” We have updated the reference to: Joughin, I., B. Smith, I. Howat, and T. Scambos. 2015, updated 2018. MEaSUREs Greenland Ice Sheet Velocity Map from InSAR Data, Version 2. [greenland_vel_mosaic200_2017_2018_v02.1]. Boulder, Colorado USA. NASA National Snow and Ice Data Center Distributed Active Archive Center. doi: https://doi.org/10.5067/OC7B04ZM9G6Q. [Sept 2018].

11. Lines 123-124, it may be useful to add a table showing the spatial resolutions of these images.

RESPONSE; We prefer to avoid adding an additional table as the article already has 12 figures, instead we add the following “The nominal ground resolution of the images varies between 10-30 m.”

12. Figure 2b, it is difficult to understand the y-axis (distance)...Distance from where?

RESPONSE Figure 2a & 2b are updated clearly indicating the point of reference used for the y-axis distance.

13. Line 139, need citation.

RESPONSE: Reference Schoof, 2007 added

14. Line 146, "two distinct epochs"?

RESPONSE: changed to in two different periods

15. Line 157-158, why?

RESPONSE: We think the reviewer means why there is tidal deformation visible at the grounding line. We have clarified the sentence,

16. Figure 3, add north arrow.

RESPONSE; North arrow added, see updated figure

17. Figure 5, DTU and TUDR GMB products are very different. A brief explanation may be useful.

RESPONSE: A short paragraph discussing the differences was added, the figure was also updated to include the latest release of data.

18. Lines 195-202, this paragraph does not talk about GRACE and should be removed to Section 3.6. Additionally, different SMB models perform differently for short-term surface runoff simulations, as shown in Smith et al., 2017, PNAS.

RESPONSE: This paragraph has been removed to the following section and we have emphasised further that using the CCI data allows us to evaluate the strengths and weaknesses of different models including different SMB model.

19. Line 225, perhaps add a comment to compare PISM with ISSM?

RESPONSE: As we do not use ISSM in this study we have not made an explicit sentence referring to ISSM but we have included a reference to it in the second paragraph of this section.

20. Line 266-267, I did not get the point. Perhaps highlight in Figure 6?

RESPONSE: We have restructured this paragraph to first describe the pattern of surface elevation change and then to interpret it in the light of how ice sheet processes contribute to this pattern.

21. Lines 272-273, add the numbers.

RESPONSE: We have added the numbers.

22. Line 285, how are the velocities simulated? PISM? perhaps this should be explained in the method section.

RESPONSE: A short description of the model velocity is added to the methods section. We have also added extra references to clarify our parameter choices.

A. Aschwanden and M. Truffer: Complex Greenland outlet glacier flow captured. Nature Communications 7, 10524, doi: 10.1038/ncomms10524, 2016

23. Line 283, Section 4.2, what is the point to compare modelled and observed ice flow velocities? I think it may be more useful to analyze the spatial pattern and temporal trend of observed ice flow velocities, or to compare different ice flow velocity datasets?

RESPONSE: The comparison is included as an example of possible uses of the observational data sets for model development. This has been elaborated in the revised text. A paper currently in preparation will focus more fully on the comparison between observed and modelled ice flow velocities.

24. Figure 8, inset in Figure 5 is too small.

RESPONSE: The inset and the basin numbers were enlarged, in addition Figure 8 has been updated with statistics on the match between models and GMB data.

25. Lines 316-320, overestimate precipitation but underestimate melt rate? This is difficult to understand. Smith et al validated SMB models with field-measured surface runoff and showed that SMB models overestimate surface melt rate although this study was conducted during a short period.

RESPONSE: We have restructured this paragraph to make it clearer to follow. The problem of overestimating precipitation and reducing melt rates was shown by Mauro et al in the south of Greenland. The performance of SMB models varies spatially and temporally over the Greenland ice sheet as we also discuss in relation to the basin 5 results. We have clarified this section to explain the relationships more clearly and we have in addition added in the Smith reference as it is relevant to the discussion about basin 6.

26. Lines 327-328, this relationship is not straightforward to me.

RESPONSE: See previous comment

27. Lines 334-336, this relationship is not straightforward to me.

RESPONSE: See previous comment

28. Figure 8, will some metrics be useful to demonstrate their relationships? e.g., Nash–Sutcliffe model efficiency coefficient.

RESPONSE: The Nash-Sutcliffe coefficients were added to Figure 8 and are referred to in the text.

29. Line 344, add the penetration depth.

RESPONSE: we have added the sentence:  “As highlighted by the 2012 extreme melt event on Greenland [61] radar altimetry is hampered by mapping changes in the reflective horizon of the ice sheet [62], where a sudden jump of the reflective hiorizon were seen in the interior by up-to 2 meters. Such jumps have the be kept in-mind and needs to be corrected for in the interpretation of mass balance.

30. Figure 9 is not easy to follow.

RESPONSE: We have rephrased the caption to better explain the

GrIS volume change estimates are converted into mass by assigning an appropriate density to the volume change, to give raw mass change from radar altimetry (Blue curve). The black line indicates the average derived mass change from radar altimetry during the ICESat era, whereas the red line shows the mass balance estimate of Sørensen et al (2011) from ICESat lidar altimetry alone. The diffidence between the maps of mass change from radar and lidar altimetry are used to calibrated raw radar altimetry mass change rates and drive the resulting 25 year record of mass change are shown in cyan with uncertainties. For reference the GRACE mass change rate is shown in green.

 

31. Figure 11 (right), y axis indicates distance from XXX?

RESPONSE: We have updated the y-label to clarify

32. Lines 379-381, the explanation is too general.

RESPONSE: We have added some more detail to this paragraph.

33. Line 384, typo?

RESPONSE: citation corrected

34. Lines 384-386, need a citation.

RESPONSE: Citation added

35. Figure 12, resolution is too low.

RESPONSE: We have increased resolutionof the images.

36. Line 441, Noël et al., 2016, TC presented 1 km SMB model.

RESPONSE: The Noel et al SMB model is a statistical downscaling of a regional climate model run at 11km resolution. While the approach of Noel et al significantly improved estimates of SMB from the RACMO model it does not yet support the statistical downscaling of precipitation fields. The hydrostatic limit of models in Greenland is very close to 5km and thus to get to higher resolution SMB without doing statistical downscaling a non-hydrostatic model is required.

37. Lines 427-442, you mentioned field observations are important in the main text but did not demonstrate this point in the Outlook section.

RESPONSE:This is a good and important point, we have added a few sentences emphasising the importance of field observations to assist in interpretation o f the satellite and model data.

38. Lines 443-474, see my general comments. I suggest the conclusion section should be revised to: 1) better highlight the additional new contribution of this study; and 2) illustrate these new contributions properly (not too strong).

RESPONSE: We thank the reviewer again for the constructive comments. The conclusion section has been revised to reflect this comment.

 

Author Response File: Author Response.pdf

Reviewer 3 Report

Overview
Mottram et al. synthesize a suite of observations and model outputs to provide an assessment of Greenland ice sheet change over a multitude of spatial and temporal scales.  The work aims to improve our understanding of changes in integrated mass, surface elevation, surface velocity, ice front position and grounding line position.  The paper is ambitious with its overall scope, however, ultimately it lacks focus.  Throughout the paper, I could not determine the aim of this particular work.  I could not determine if this was meant to be a methods paper displaying the ESA products for Greenland, a review paper of the “state of the science” or a detailed science paper on a particular topic.  Due to this lack of focus I felt the authors missed the bar they were trying to meet.  Overall, there are a number of issues that need to be resolved before I could recommend this manuscript for publication.


General Comments
- Several portions of the manuscript are awkwardly phrased or constructed.  Some sentences need to be split into separate coherent pieces.  Some sentences need to be simplified for clarity so that the focus is not lost.

- Overuse of acronyms

- Having an ice sheet model with the physics described (SIA+SSA) be able to attribute the mass change of the ice sheet to different processes is misleading.  Simply reproducing the observed changes in ice sheet mass in an ice sheet model is a major challenge.  No information was given on the boundary conditions (bed topography? oceanic forcing?) needed for this partitioning to be accurate.  Model assessments of calving are also presently inadequate for this work.

- I know this is the name of the ESA product, but gravimetric mass balance should really be gravimetrically-derived mass balance

- Need to separate textual (\citet) versus parenthetical (\citep) citations.  i.e. “Mottram et al. [1] do this” versus “this was done [1]”.


Line-by-line Comments
L1: average instead of estimated

L2-4: this sentence is awkwardly phrased.  Should be reversed to be something like:
“Understanding the present-day state of the Greenland ice sheet is vital for understanding the processes controlling the modern-day rates of sea level change and for making projections of sea level rise into the future.”

L9-10: the maximum rate of surface elevation change isn’t a particularly useful statistic in this context.  Perhaps use the average surface elevation change?  Or average change in low elevation areas?

L10-12: while it is true that 26 out of these 28 ice fronts are in states of retreat, it is a complicated statistic involving regional geomorphology, ocean bathymetry and conditions, and a series of other factors.  These 28 might not be representative of the mean state.  That being said, I think the authors are correct in their analysis.  I just think some caveat should be noted.

L15: what do you mean by “tie together disparate lines of evidence”

L16-20: this needs to be rewritten.

L55: no degree symbol needed for K.

L56-57: Basal melt rates are minor but exist.  “The contribution to sea level rise from the Greenland ice sheet results primarily from two major processes.”

L57-60: These sentences need to be rewritten.  “In the calculation of the surface mass budget, precipitation at the surface of the ice sheet is balanced by losses from meltwater runoff, sublimation and wind scour.  Runoff is the portion of snow and ice melt that is not refrozen in the snowpack or retained in liquid firn aquifers.  The surface mass budget, sometimes referred to as climatic mass budget or surface mass balance, is referred to in this paper as SMB.”

L63-65: Simplify this sentence.  Can list the actual primary methods for determining the ice sheet mass balance.

L65-71: Rework this paragraph.  
“In the first Ice Sheet Balance Inter-comparison Exercise (IMBIE), the average Greenland mass budget from four independent methods was -237 Gt/yr between 2000 and 2011, which is a contribution of 0.65 mm/year to mean sea level rise; and -263 Gt/yr between 2005 and 2011, which is a contribution of 0.72 mm/year to mean sea level rise [12].  This accounts for a little less than a third of the average 3.1 mm/year observed sea level rise between 1993 and 2017 [12,13]. As each measurement technique is sensitive to different processes and there is high inter-annual variability in precipitation, melt, calving fluxes and other processes, the total mass budget of the Greenland ice sheet is sensitive to time period chosen and methods used [12,14].”

L72-75: Rework this sentence. “There has been a significant increase in the amount of data available from in-situ, airborne and satellite observations of the Greenland ice sheet.”

L75-76: Rework this sentence. “The ESA CCI project has consolidated, standardised and integrated satellite remote sensing data to create high quality datasets that are accessible to scientists and policymakers.”

L76-81: Simplify this sentence. “This data has been used to identify significant trends and changes in ice dynamics in Greenland including identifying significant changes in ice velocity at large outlet glaciers [16], assessing the causes surface elevation change across the ice sheet [2,4,5,7] as well as assessing the importance of different processes controlling seasonal velocity changes on the ice sheet [17].”

L94-97: Rewrite this sentence

L102-104: “Ice velocity maps with 250 m grid spacing are produced at 6 to 12 day intervals and are annually combined and averaged to create more spatially-complete ice sheet-wide mean-velocity mosaics.”

L105: Don’t use mimicking

L106-107: Why mention the production of the 2017-2018 map?

L107: snapshots

L109-113: “Quality assessment of the ice velocity retrieval algorithms are performed through internal consistency checks (for example, checking how the algorithm performs on bedrock outcrops that are assumed stationary), and inter-comparisons with other algorithms through dedicated round-robins [18]. The velocity data is validated against independent datasets from ground-based in-situ GPS, higher resolution sensors (such as TerraSAR-X and COSMO-SkyMed, and publicly-available published datasets (MEaSUREs, eg. [8,15]).”

L121: remove “expert”

L125: change this to that the data is “in reference to a stationary geoid rather than topographic height from a DEM.”  Which geoid model was used?

L137: Could mention the vulnerability of the ice shelf at Zachariae Isstrøm (Mouginot et al. 2015)

L160: List the radar altimetry missions provided in the product.  Combining radar altimetry from different missions is non-trivial.  While the references are provided, more detail would be helpful.  This is one particular aspect of my point about the lack of clarity on the intent of this manuscript.  In other sections, the method is detailed to a much greater degree than what is provided here.  If this is a review article, as listed in the conclusions, then this section should be much more detailed.

L164-165: “More detail into the specific methods is provided in the method review paper [5] and in mission specific papers [2,4,22-25]”.

L166: Here the figure is listed as the average rate between 2012 and 2016 but the caption for the figure lists 2016-2017.  There is an inconsistency of acronyms.  Here surface elevation change is SEC, but in the figure it is EC.  I would err against using any of these acronyms personally.

L174-176: Rewrite this sentence.

L177-179: Different GRACE processing centers provide different levels of spherical harmonic truncation.

Figure 5: While 4 different solutions are listed as being available from the ESA, only two are shown here.  No errors are listed on the GRACE data.  Also the basin numbers are very small, which makes it difficult for later sections.  The numbers were impossible to read when printed.

L197: Remove “apparently”

L197-198: Citation for calculating ice fluxes from GRACE and SMB?

L198-199: Isn’t RACMO2 partially based on HIRLAM?
https://www.projects.science.uu.nl/iceclimate/models/racmo.php

L211-213: Awkward construction

L216-218: Rewrite sentence.
“Because surface melt can refreeze or be retained within the snowpack and firn layers, the total melt generated does not necessarily lead to  surface runoff or a mass loss from the system.”

L226-229: Split this sentence.
“226 Given the vast amount of feedback mechanisms and interactions involving ice sheets in the climate system, ice sheet modelling and estimating future rates of ice loss is a major challenge.  Changes in ice sheet dynamics have been identified as a major source of uncertainty in sea level rise projections by the IPCC in the fifth assessment report [8].”

L229-231: could actually enumerate the different ice sheet models (shallow ice, shallow shelf, Blatter-Pattyn, Full-Stokes).  Descriptions listed here are a mix of generic (basic zero order models, hybrid models, second order models) and specific (Full-Stokes).

L233: Slight edits to sentence
“Regardless of model complexity, proper boundary and initial conditions are necessary, which requires high-quality observational data to be available.”

L236.5: “PISM is an open-source thermodynamically coupled, polythermal hybrid stress balance ice sheet model [47,48], which combines the Shallow Ice Approximation (SIA) [49] and the Shallow Shelf Approximation [50]”

L246: grid resolution is uniform spatially?  Is 2km appropriate for dynamically changing areas?

L251: is mass conserved as the grid resolution changes?

L251-253: why not just state the range of years used to generate the average? (1980-1995?).  Also, some observed changes in the state of the Greenland ice sheet start occurring in the 1990’s (e.g. Shepherd et al., 2012).  Would this affect your results?

Figure 6: is the SMB estimate here from the firn model or calculated using an assumed density?  Model does not reproduce changes observed at major glaciers (Jakobshavn Isbrae, Helheim Glacier and Kangerlussuaq Glacier) and smaller outlets in NW Greenland.  Possible to reproduce these changes with more realistic model physics?

L265-268: uncertainty in conversion from volume to mass may complicate this comparison.   Also radar altimeters can miss changes in accumulation due to penetration of the radar signal into the surface.  should add some caveats.

L276-279: here a simple set of physics was used at a relatively coarse resolution.   what bed topography was used?  what is the oceanic forcing (i.e. from OMG?)?  

L296-297: basal friction is commonly inverted in ice sheet models with present-day ice velocities.  is that the case here?

L304-305: could reference the general location of each basin number (i.e. basin 5 is in south Greenland).

L310-312: this seems out of place (talking about mass in this section).

L325: be specific please.
“Overestimating precipitation in the high topographic relief coastal areas can lead to an underestimate in precipitation in the ice sheet interior due to orographic effects”  

L334: might be overstepping the attribution.  This is another section supporting my point about the lack of clarity on the intent of this manuscript.  Is this meant to be a topical science article?

L340-342: might need to mention uncertainties in the GRACE-derived estimates (Wahr et al., 2006), uncertainties in the SMB estimate (van den Broeke et al., 2009) and possibly seasonal fluctuations in ice dynamics (Moon et al., 2015)

L352: no detail is given for the ICESat processing

L353-265: combining laser altimetry with radar altimetry is a non-trivial task.  They can measure different surfaces due to sampling footprints and radar signal penetration. Radar altimeters have been historically poorer at capturing changes in coastal areas.  A simple scaling may not take into account a number of different effect, such as changes in surface processes.

L363: should be “high-slope”

L363-364: statistically out of balance after 2004?

Figure 9: No errors on GRACE or ICESat data.

Figure 10: This figure is very small and cut in half in the PDF.  Might as well have a table versus this bar graph visual of calving front retreat.  Or as a map of all the glaciers with circles of various diameter to denote the retreat distance.

L377: the episodic behavior isn’t out of the ordinary for many tidewater glaciers

Section 4.6: this section seems completely out of place in terms of detail compared with other sections.

L384: fix citation (broken)

L384-386: Sentence seems awkward.  Maybe something like “Floating ice shelves similar to those in Antarctica are found in Greenland, however are typically confined within fjords”

L386-388: Why mention the technical challenge without specifying why?  Available data?

L390: each of these glaciers

L391: their collapse could

L396-397: awkward sentence construction

L406: reach where?  the base?

L410: summer (spelling)

L411: why mention Zachariah Isstrøm here?

L438: This sentence needs to be broken up into coherent pieces.  Something like:
“The successful launch of the GRACE-Follow On (GRACE-FO) mission in May 2018 will help extend the time series of ice sheet mass balance and allow for better assessments of regional variability”
“The continued developments in regional climate models, such as the non-hydrostatic HARMONIE-AROME model, and reanalysis products, such as MERRA-2 Replay Analyses, has great potential for deriving more accurate SMB reconstructions.”

Author Response

Overview
Mottram et al. synthesize a suite of observations and model outputs to provide an assessment of Greenland ice sheet change over a multitude of spatial and temporal scales.  The work aims to improve our understanding of changes in integrated mass, surface elevation, surface velocity, ice front position and grounding line position.  The paper is ambitious with its overall scope, however, ultimately it lacks focus.  Throughout the paper, I could not determine the aim of this particular work.  I could not determine if this was meant to be a methods paper displaying the ESA products for Greenland, a review paper of the “state of the science” or a detailed science paper on a particular topic.  Due to this lack of focus I felt the authors missed the bar they were trying to meet.  Overall, there are a number of issues that need to be resolved before I could recommend this manuscript for publication. RUTH


General Comments
- Several portions of the manuscript are awkwardly phrased or constructed.  Some sentences need to be split into separate coherent pieces.  Some sentences need to be simplified for clarity so that the focus is not lost.

- Overuse of acronyms

- Having an ice sheet model with the physics described (SIA+SSA) be able to attribute the mass change of the ice sheet to different processes is misleading.  Simply reproducing the observed changes in ice sheet mass in an ice sheet model is a major challenge.  No information was given on the boundary conditions (bed topography? oceanic forcing?) needed for this partitioning to be accurate.  Model assessments of calving are also presently inadequate for this work.

- I know this is the name of the ESA product, but gravimetric mass balance should really be gravimetrically-derived mass balance

- Need to separate textual (\citet) versus parenthetical (\citep) citations.  i.e. “Mottram et al. [1] do this” versus “this was done [1]”.

RESPONSE: We thank the reviewer for the thorough and thoughtful comments. We have restructured the introduction and background  as well as conclusions and outlook sections to better focus the paper. We have also thoroughly revised the overly long sentences and added more detail on the modelling aspects, in particular the ice sheet model PISM to better focus the discussion. We respond inline to the specific comments below

Line-by-line Comments
L1: average instead of estimated

Response: changed

L2-4: this sentence is awkwardly phrased.  Should be reversed to be something like:
“Understanding the present-day state of the Greenland ice sheet is vital for understanding the processes controlling the modern-day rates of sea level change and for making projections of sea level rise into the future.”

RESPONSE; This is a much better way of expressing this sentence. It has been changed.

L9-10: the maximum rate of surface elevation change isn’t a particularly useful statistic in this context.  Perhaps use the average surface elevation change?  Or average change in low elevation areas?

RESPONSE: As part of the work focusing the paper we have significantly rewritten the abstract to clarify the highlights of the paper and direct the focus towards the response of the ice shete to a warming climate.

L10-12: while it is true that 26 out of these 28 ice fronts are in states of retreat, it is a complicated statistic involving regional geomorphology, ocean bathymetry and conditions, and a series of other factors.  These 28 might not be representative of the mean state.  That being said, I think the authors are correct in their analysis.  I just think some caveat should be noted.

RESPONSE: See previous response

L15: what do you mean by “tie together disparate lines of evidence”

RESPONSE: This part of the abstract has been simplified and the phrase removed

L16-20: this needs to be rewritten.

RESPONSE: See previous response.

L55: no degree symbol needed for K.

RESPONSE: degree sign removed

L56-57: Basal melt rates are minor but exist.  “The contribution to sea level rise from the Greenland ice sheet results primarily from two major processes.” 

RESPONSE: We have added the sentence” More minor contributions to the mass budget come from basal melting and sublimation at the surface but these processes are not considered in detail here.”

L57-60: These sentences need to be rewritten.  “In the calculation of the surface mass budget, precipitation at the surface of the ice sheet is balanced by losses from meltwater runoff, sublimation and wind scour.  Runoff is the portion of snow and ice melt that is not refrozen in the snowpack or retained in liquid firn aquifers.  The surface mass budget, sometimes referred to as climatic mass budget or surface mass balance, is referred to in this paper as SMB.” 

RESPONSE: This section has been rewritten to clarify the meaning as well as to refocus the paper.

L63-65: Simplify this sentence.  Can list the actual primary methods for determining the ice sheet mass balance.

RESPONSE: the whole section has been reworked, we have included a list of the primary methods of determining ice sheet mass balance.

L65-71: Rework this paragraph.  
“In the first Ice Sheet Balance Inter-comparison Exercise (IMBIE), the average Greenland mass budget from four independent methods was -237 Gt/yr between 2000 and 2011, which is a contribution of 0.65 mm/year to mean sea level rise; and -263 Gt/yr between 2005 and 2011, which is a contribution of 0.72 mm/year to mean sea level rise [12].  This accounts for a little less than a third of the average 3.1 mm/year observed sea level rise between 1993 and 2017 [12,13]. As each measurement technique is sensitive to different processes and there is high inter-annual variability in precipitation, melt, calving fluxes and other processes, the total mass budget of the Greenland ice sheet is sensitive to time period chosen and methods used [12,14].”

RESPONSE: see previous comment.

L72-75: Rework this sentence. “There has been a significant increase in the amount of data available from in-situ, airborne and satellite observations of the Greenland ice sheet.”

RESPONSE: see previous comment.

L75-76: Rework this sentence. “The ESA CCI project has consolidated, standardised and integrated satellite remote sensing data to create high quality datasets that are accessible to scientists and policymakers.”

RESPONSE: see previous comment.

L76-81: Simplify this sentence. “This data has been used to identify significant trends and changes in ice dynamics in Greenland including identifying significant changes in ice velocity at large outlet glaciers [16], assessing the causes surface elevation change across the ice sheet [2,4,5,7] as well as assessing the importance of different processes controlling seasonal velocity changes on the ice sheet [17].”

RESPONSE: see previous comment.

L94-97: Rewrite this sentence

RESPONSE: Sentence has been simplified.

L102-104: “Ice velocity maps with 250 m grid spacing are produced at 6 to 12 day intervals and are annually combined and averaged to create more spatially-complete ice sheet-wide mean-velocity mosaics.”

RESPONSE: Sentence has been simplified.

L105: Don’t use mimicking

RESPONSE: replaced with: “equivalent to”

L106-107: Why mention the production of the 2017-2018 map? 

RESPONSE: This map is a product that is foreseen to be available by the time or shortly after the paper will be published. The reader is likely interested in this product and for the sake of completeness it is mentioned here as well.

L107: snapshots

RESPONSE: changed

L109-113: “Quality assessment of the ice velocity retrieval algorithms are performed through internal consistency checks (for example, checking how the algorithm performs on bedrock outcrops that are assumed stationary), and inter-comparisons with other algorithms through dedicated round-robins [18]. The velocity data is validated against independent datasets from ground-based in-situ GPS, higher resolution sensors (such as TerraSAR-X and COSMO-SkyMed, and publicly-available published datasets (MEaSUREs, eg. [8,15]).”

RESPONSE: changed

L121: remove “expert”
RESPONSE: changed

L125: change this to that the data is “in reference to a stationary geoid rather than topographic height from a DEM.”  Which geoid model was used?

RESPONSE: Adjusted, the geoid is EGM96 (this is now mentioned)

L137: Could mention the vulnerability of the ice shelf at Zachariae Isstrøm (Mouginot et al. 2015)

RESPONSE: We have inserterd this reference, it is also discussed in the results sections

L160: List the radar altimetry missions provided in the product.  Combining radar altimetry from different missions is non-trivial.  While the references are provided, more detail would be helpful.  This is one particular aspect of my point about the lack of clarity on the intent of this manuscript.  In other sections, the method is detailed to a much greater degree than what is provided here.  If this is a review article, as listed in the conclusions, then this section should be much more detailed. 

RESPONSE: The timeseries is explaind in much detail in Sørensen et al. 2018. Howver we add a little more information to the section:

 

The surface elevation change (SEC) is based on satellite radar altimeter observations from 1992-2017 using data from ESA radar altimeters (ERS-1, ERS-2, ENVISat and CryoSat-2). The time series of observations is averaged as 5-year running mean estimates, to reflect climate variability and not weather. The SEC estimation uses a mission specific combination of cross-over-, repeat-track- and least-squares-methods to estimate the temporal evolution of surface elevation at a common 5 km uniformed grid for the entire GrIS. Full details in the mission specific set-up for deriving surface elvation change are given the review of Sørensen et al. [5]. If the reader is interested in more mission specific papers we refer to [2,4,22–25]. In the following the observed SEC is compared to the results of RCM and ISM modelling (see section 3.6 and 3.7) to infer  the importance of surface mass balance versus ice sheet dynamics

L164-165: “More detail into the specific methods is provided in the method review paper [5] and in mission specific papers [2,4,22-25]”. 

RESPONSE: see previous comment

L166: Here the figure is listed as the average rate between 2012 and 2016 but the caption for the figure lists 2016-2017.  There is an inconsistency of acronyms.  Here surface elevation change is SEC, but in the figure it is EC.  I would err against using any of these acronyms personally.

RESPONSE: We have fixed the caption and 2016-17 was a typo.

L174-176: Rewrite this sentence. 

RESPONSE: sentence has been simplified

L177-179: Different GRACE processing centers provide different levels of spherical harmonic truncation. 

RESPONSE: This is now explicitly mentioned.


Figure 5: While 4 different solutions are listed as being available from the ESA, only two are shown here.  No errors are listed on the GRACE data.  Also the basin numbers are very small, which makes it difficult for later sections.  The numbers were impossible to read when printed.

RESPONSE: Error bars as well as mass balance estimates including uncertainties were added to Figure 5. The products shown in Figure 5 are now based on the latest CSR RL06 data. We compare the mass balance estimates for these products to those from the ITSG product. This justifies our decision to just illustrate products from one processing centre. Finally, the inset in Figure 5 and the basin numbers were enlarged.

L197: Remove “apparently” 

RESPONSE: Removed

L197-198: Citation for calculating ice fluxes from GRACE and SMB?

RESPONSE: We have moved this paragraph to the following section and have clarified that we are referring to published studies that have used SMB – discharge rather than directly comparing GMB with SMB to derive ice fluxes which in principle is possible.

L198-199: Isn’t RACMO2 partially based on HIRLAM?
https://www.projects.science.uu.nl/iceclimate/models/racmo.php

RESPONSE: HIRHAM and RACMO are regional climate models that both use the dynamical scheme from the HIRLAM numerical weather prediction model, though slightly different versions. They do however have different physics schemes and in particular different SMB models. We have added this to the section.

L211-213: Awkward construction

RESPONSE: Sentences broken up and simplified

L216-218: Rewrite sentence.
“Because surface melt can refreeze or be retained within the snowpack and firn layers, the total melt generated does not necessarily lead to  surface runoff or a mass loss from the system.”

RESPONSE: Sentences broken up and simplified

L226-229: Split this sentence.
“226 Given the vast amount of feedback mechanisms and interactions involving ice sheets in the climate system, ice sheet modelling and estimating future rates of ice loss is a major challenge.  Changes in ice sheet dynamics have been identified as a major source of uncertainty in sea level rise projections by the IPCC in the fifth assessment report [8].”

RESPONSE: Sentences broken up and simplified

L229-231: could actually enumerate the different ice sheet models (shallow ice, shallow shelf, Blatter-Pattyn, Full-Stokes).  Descriptions listed here are a mix of generic (basic zero order models, hybrid models, second order models) and specific (Full-Stokes). 

RESPONSE; updated

L233: Slight edits to sentence
“Regardless of model complexity, proper boundary and initial conditions are necessary, which requires high-quality observational data to be available.”

RESPONSE; updated

L236.5: “PISM is an open-source thermodynamically coupled, polythermal hybrid stress balance ice sheet model [47,48], which combines the Shallow Ice Approximation (SIA) [49] and the Shallow Shelf Approximation [50]”

RESPONSE; updated

L246: grid resolution is uniform spatially?  Is 2km appropriate for dynamically changing areas?

RESPONSE; Grid resolution is spatially uniform. This has been added to the text. Aschwanden and Truffer (2016) analysed the effect of increasing model resolution on improving ice velocities in major outlets in Greenland and found that most of the investigated outlets improved velocities with increasing model resolution and that a grid resolution of a minimum of 2km was needed in order to obtain correct orders of magnitude for the fastest-moving outlets. A smaller resolution than 2km may improve the flow but also increases the computational load significantly.

 

L251: is mass conserved as the grid resolution changes?

RESPONSE; Test runs with constant forcing changing only resolution of model (and forcing fields) and allowing bed topography to reflect the resolution change typically show a short transient with a quite steep decrease in ice mass when resolution changes, reflecting an increase in the number of resolved outlets. During spinup, the model is run until the ice mass is stable for each resolution.

L251-253: why not just state the range of years used to generate the average? (1980-1995?).  Also, some observed changes in the state of the Greenland ice sheet start occurring in the 1990’s (e.g. Shepherd et al., 2012).  Would this affect your results?

RESPONSE: We have simplified the sentence to make it clear that we mean the period 1980 to 1995. We use this period as it is closer to the average of the 20th century and the spin-up aims to bring the ice sheet as close as possible to equilibrium with climate before running the model up to the present day. As there has been an acceleration in melt, associated with warmer average temperatures since around 2000, though, as the reveiewer rightly points out, likely starting even earlier in the 1990s, we made the choice to use the first part of the regional climate model output for spin-up and then start the run using the remaining22 years.

Figure 6: is the SMB estimate here from the firn model or calculated using an assumed density?  Model does not reproduce changes observed at major glaciers (Jakobshavn Isbrae, Helheim Glacier and Kangerlussuaq Glacier) and smaller outlets in NW Greenland.  Possible to reproduce these changes with more realistic model physics? 

RESPONSE; The SMB estimate here is calculated using a model that includes firn processes such as refreezing and retention of melt water as well as densification of the snow layers as described in Langen et al 2017. We have expanded the discussion to clarify this and to explain in more detail how we can use this analysis to determine what is driving observed surface elevation change.

L265-268: uncertainty in conversion from volume to mass may complicate this comparison.   Also radar altimeters can miss changes in accumulation due to penetration of the radar signal into the surface.  should add some caveats. 

RESPONSE: See comment above.  We discuss elsewhere the problem of penetration of the radar signal into the snowpack and have added some caveats about the uncertainties in this analysis

L276-279: here a simple set of physics was used at a relatively coarse resolution.   what bed topography was used?  what is the oceanic forcing (i.e. from OMG?)?

RESPONSE: Reference to the bedrock data set has been given earlier in the text. The oceanic forcing is given by a constant heat flux. Information on this has been added to the text.  

L296-297: basal friction is commonly inverted in ice sheet models with present-day ice velocities.  is that the case here?

RESPONSE: This is not the case, PISM uses till friction angles to determine yield stress and we have stated this is the text. This is however a possibel direction of future work that we also mention in the outlook section.

L304-305: could reference the general location of each basin number (i.e. basin 5 is in south Greenland).

RESPONSE: We have revised this section to make it easier for the reader to follow by referencing the different locations.

L310-312: this seems out of place (talking about mass in this section).

RESPONSE: The surface elevation change data is compared with the predicted surface elevation change from surface mass balance model as discussed in the responses above. Here we examine if the GMB data also supports the hypothesis that the regional climate models have a low biased precipitation in the interior of Greenland.

L325: be specific please.
“Overestimating precipitation in the high topographic relief coastal areas can lead to an underestimate in precipitation in the ice sheet interior due to orographic effects”  

RESPONSE: The section has been rewritten to clarify the orographic effects.

L334: might be overstepping the attribution.  This is another section supporting my point about the lack of clarity on the intent of this manuscript.  Is this meant to be a topical science article?

RESPONSE: The importance of albedo in determing seasonal surface mass balance rates is well known in the glaciological literature, we have added some references to support this interpretation of the difference between the SMB models in northern Greenland. In this section we also aim to point to some of the open science questions that need further work to conclude.

L340-342: might need to mention uncertainties in the GRACE-derived estimates (Wahr et al., 2006), uncertainties in the SMB estimate (van den Broeke et al., 2009) and possibly seasonal fluctuations in ice dynamics (Moon et al., 2015)

RESPONSE: We have added a paragraph on the uncertainties in the GMB, SMB and ice dynamics.

L352: no detail is given for the ICESat processing

Response: We have added some more detail. “This correction by-passes the need for models of; firn changes, GIA and appropriate density of the snow and ice involved in the volume change, in addition to correcting for the marginal parts of the ice sheet not observed by the radar altimeter. Especially the marginal parts are the main driver in mass loss and may be the main reason for the higer mass balance of the raw radar estimate (blue curve in figure 9) compared to the scaled estimate (cyan curve)”


L353-265: combining laser altimetry with radar altimetry is a non-trivial task.  They can measure different surfaces due to sampling footprints and radar signal penetration. Radar altimeters have been historically poorer at capturing changes in coastal areas.  A simple scaling may not take into account a number of different effect, such as changes in surface processes.

RESPONSE; See comment above.

L363: should be “high-slope”

RESPONSE: Corrected

L363-364: statistically out of balance after 2004?

RESPONSE:

Figure 9: No errors on GRACE or ICESat data.SEBASTIAN

RESPONSE: This is now added to the figure.

Figure 10: This figure is very small and cut in half in the PDF.  Might as well have a table versus this bar graph visual of calving front retreat.  Or as a map of all the glaciers with circles of various diameter to denote the retreat distance.

RESPONSE: We have changed the right panel to a tableto more clearly show the calving front retreat rates

L377: the episodic behavior isn’t out of the ordinary for many tidewater glaciers

RESPONSE: That this episodic behaviour is normal is the point of this paragraph. We have clarified the sentence to state this explicitly.

Section 4.6: this section seems completely out of place in terms of detail compared with other sections.

RESPONSE: As part of the restructuring of the manuscript we have significantly increased the length and detail in the other sections and section 4.6 is now more in proportion

L384: fix citation (broken)

RESPONSE: Fixed

L384-386: Sentence seems awkward.  Maybe something like “Floating ice shelves similar to those in Antarctica are found in Greenland, however are typically confined within fjords”

RESPONSE: Fixed

L386-388: Why mention the technical challenge without specifying why?  Available data?

RESPONSE: As this was not relevant to the discussion in the section we have removed this part of the sentence, the technical difficulties are in any case mentioned in the earlier methods section.

L390: each of these glaciers

RESPONSE: We have rewritten parts of this section as part of our refocusing of the paper and this has been removed.

L391: their collapse could

RESPONSE: See previous comments

L396-397: awkward sentence construction

RESPONSE: See previous comment.

L406: reach where?  the base?

RESPONSE: See previous comment.

L410: summer (spelling)

RESPONSE: fixed

L411: why mention Zachariah Isstrøm here?

RESPONSE: We have rewritten parts of this section as part of our refocusing of the paper and this has been removed.

L438: This sentence needs to be broken up into coherent pieces.  Something like:
“The successful launch of the GRACE-Follow On (GRACE-FO) mission in May 2018 will help extend the time series of ice sheet mass balance and allow for better assessments of regional variability”
“The continued developments in regional climate models, such as the non-hydrostatic HARMONIE-AROME model, and reanalysis products, such as MERRA-2 Replay Analyses, has great potential for deriving more accurate SMB reconstructions.”
RESPONSE: fixed

 

Author Response File: Author Response.pdf

Reviewer 4 Report

Mottram et al provide a detailed analysis of Greenland Ice Sheet Essential Climate Variables data product results.  The explanations of the methods are sufficient and effective.  The validation and comparison with other independent data sets is a best practice and an essential step for ongoing improvement to the ECV data products. Each data product reported has the same signal indicative of the impact of a warming climate on the GIS. I have reviewed 38 papers this year, and this is the best initial submission draft I have seen. Most of the comments below are questions that will either clarify an item in the paper or some cases this reviewer who the clarification will help.  

7: “These” refers to what variables or just list as ECV? Surface elevation change, ice velocity, ice mass change and grounding line location?

28: List the ECV’s here.

33:  Are all of these data products ECV’s?

57: Suggested rewrite “Surface mass balance (SMB) accounts for precipitation at the surface of the ice sheet, surface ice melt and runoff, and surface melt that refreezes in the snowpack [10] or is retained in liquid firn aquifers. The latter subsurface properties complicate SMB determination, sometimes call climate mass budget.”

83:  What is “them”? The data products?

127:  Is there any pattern to the color coding of ice front positions, or is just varied for visual clarity?

133: Add reference for the few pixels accuracy comment.

139:  Given the limited number of these glaciers is this a sensitive indicator of outlet glacier stability, which can impact ice sheet stability, instead of ice sheet stability.

148:  A range of GLL precision must be given along with a reference including the precision after April 2016.

236: “Shallow Shelf Approximation (SSA) [50].

240: Does this mean calving front is static but not calving flux?

241: “in the”

270:  Given the short five year period here what mechanism could dynamic processes generate the observed elevation change?

281:  Should be noted that may not just be dynamic thinning from calving but also from increased frontal and basal melt. Ie. (Porter et al 2018)  https://doi.org/10.3389/feart.2018.00090

292: Explain why the model could not be initialized with the NEGIS observed velocity? This may not fit in paper, but would like to see how briefly it can be addressed.

311: Do the model results indicate that SMB mismatch with SEC is from precipitation underestimation because RCM is too low or is there any room for the issue to be from the SMB multi-layer firn model for refreezing and retention

323:  The RACMO model is not really concerned with snow density in its precipitation forecast. I am just wondering if the work of Fausto et al (2018) on snow density provides any insight that helps explanations here vs basin discrepancy with GRACE. https://doi.org/10.3389/feart.2018.00051

358: Appropriate density does for firn, snow and ice? Does this also reflect findings of Fausto et al (2018)

363: slope

374:  Is it worth a key listing of key factors that may lead to stability or instability, calving front geometry, fjord geometry, mélange persistence, subglacial melt rates, and subglacial meltwater outflow?

384:  Fix reference

405: Are rifts distinguished from crevasses in this process or are they just the end member of full thickness penetration crevasse?

410: summer

411: The lubrication hypothesis does make sense, on Petermann there is also exceptional basal melting particularly in summer that would lead to thinning and uplift. Does Figure 12 A and B suggest basal melting must be the key?

415:  Stress intensity factor is use in Figure 12 C and D, but is note explained.  Please briefly do so near here.

417:  The persistence of Petermann ice tongue in an environment rich in surface water suggests the Petermann is not as vulnerable to this as other ice shelves that experience limited surface melt. MacDonald et al (2018) Annals of Glaciology DOI: 10.1017/aog.2018.9.  NEGIS also has had much surface melt ponding and streams for sometime.  

420: For Petermann ice thickness increases considerably in last 25 km of flotation, does fracture depth, Figure 12D, have less ability to impact a much thicker tongue.  The floating shelf would of course not be as thick at the time terminus retreat reached that point, but would still be thicker than the generally thin limited thickness change in the region calved to date.

423: The increasing vulnerability is of course going to occur with more melt, but the above reference suggests caution in applying this to a rapid collapse for an ice shelf that has existed in an environment of considerable melt for a long time. Such an ice shelf has developed a drainage system. This does not make it less vulnerable to additional melt driven thinning and the warming of the ice itself and increased water availability for crevasse penetration enhancement. Petermann already receives plenty of attention in the paper, so I am not advocating for increased coverage.

452:  The conclusion should refer back to the ECV’s since the paper started with this. Something along the lines of “Continued independent validation and comparison, along with focused field work and enhanced modelling of calving will lead to more accurate ECV products. All of the data products reported here have a signal indicative of the significant impact and response of the GIS to climate change.”

Author Response

Review 4

Mottram et al provide a detailed analysis of Greenland Ice Sheet Essential Climate Variables data product results.  The explanations of the methods are sufficient and effective.  The validation and comparison with other independent data sets is a best practice and an essential step for ongoing improvement to the ECV data products. Each data product reported has the same signal indicative of the impact of a warming climate on the GIS. I have reviewed 38 papers this year, and this is the best initial submission draft I have seen. Most of the comments below are questions that will either clarify an item in the paper or some cases this reviewer who the clarification will help.  

RESPONSE: We thank the reviewer for their very kind comments and thorough review. As some of the other reviewers requested we focus the paper more we have rewritten and clarified several sections to show not just the ECV data products but how we can build an integrated picture of the impact of a warming climate on the ice sheet.  We respond to the specific comments and questions in line below.

7: “These” refers to what variables or just list as ECV? Surface elevation change, ice velocity, ice mass change and grounding line location?

RESPONSE: We have clarified this section and listed the ECVs

28: List the ECV’s here.

RESPONSE: See previous comment

33:  Are all of these data products ECV’s?

RESPONSE: Yes they are. See previous comment

57: Suggested rewrite “Surface mass balance (SMB) accounts for precipitation at the surface of the ice sheet, surface ice melt and runoff, and surface melt that refreezes in the snowpack [10] or is retained in liquid firn aquifers. The latter subsurface properties complicate SMB determination, sometimes call climate mass budget.”

RESPONSE: We have rewritten this section to simplify the sentences and clarify the meaning

83:  What is “them”? The data products?

RESPONSE: Yes they are. See previous comment

127:  Is there any pattern to the color coding of ice front positions, or is just varied for visual clarity?

RESPONSE: the colour coding is just for visual clarity, there is no patern behind it.

133: Add reference for the few pixels accuracy comment.

RESPONSE: Reference added.

Khvorostovsky, et al., Algorithm Theoretical Baseline Document (ATBD) for the Greenland_Ice_Sheet_cci project of ESA's Climate Change Initiative, version 3.2, 28 Jun 2018 (http://www.esa-icesheets-cci.org/).

REMEMBER to do this

139:  Given the limited number of these glaciers is this a sensitive indicator of outlet glacier stability, which can impact ice sheet stability, instead of ice sheet stability.

RESPONSE: This is a good point, our formulation here was unclear. We have clarified the section

148:  A range of GLL precision must be given along with a reference including the precision after April 2016.

RESPONSE: The InSAR double differencing method (DInSAR) is the most accurate technique for detection of the grounding line, nevertheless it can only detect the tidal flexure zone which is a proxy of the real GLL. We are not aware of any study which compared actual GLL’s against DInSAR derived GLL’s to quantify this. Instead we report here the uncertainty of (manual) GLL delineation from digitizing errors, which varies based on coherence and type of features. We adjusted the following sentence and added a reference:

“The accuracy depends primarily on grounding zone geometry (slope), tidal amplitude, ice flow velocity and the quality of the interferogram (SNR, Coherence, feature clarity) and uncertainties vary from 200m to locally more than 1.5 km (Nagler et al., 2018).”

Nagler;T., et al., Comprehensive Error Characterisation Report (CECR). Antarctic_Ice_Sheet_cci project, ESA's Climate Change Initiative, version 3.0, 28 June 2018. Available from: http://www.esa-icesheets-antarctica-cci.org/

236: “Shallow Shelf Approximation (SSA) 

RESPONSE: fixed

240: Does this mean calving front is static but not calving flux?

RESPONSE: Yes that is correct. The PISM version here does not have a dynamic feedback but uses a form of ocean kill to determine where calving should occur.

241: “in the”

RESPONSE: fixed

270:  Given the short five year period here what mechanism could dynamic processes generate the observed elevation change?

RESPONSE: This is especially seen in the Jakobshavn area, where the topography is controlling the shape of the elevation change and there appears also to be significant dynamic thining due to high velocity gradients. We have expanded this section to discuss the ice dynamics further.

281:  Should be noted that may not just be dynamic thinning from calving but also from increased frontal and basal melt. Ie. (Porter et al 2018)  https://doi.org/10.3389/feart.2018.00090

RESPONSE: Noted and added to this paragraph

292: Explain why the model could not be initialized with the NEGIS observed velocity? This may not fit in paper, but would like to see how briefly it can be addressed.

RESPONSE: One could do an inversion based on surface velocities to obtain basal friction, and then use this for the simulations. However, it would require a consistent surface velocity covering the entire GIS, not just a single area. In general, PISM does not require surface velocities as an initial condition, only bed topography and ice thickness (along with assumptions regarding till angles). We have added a brief overview of this to the section.

311: Do the model results indicate that SMB mismatch with SEC is from precipitation underestimation because RCM is too low or is there any room for the issue to be from the SMB multi-layer firn model for refreezing and retention

RESPONSE: It is possible that some of the mismatch results from firn model parameterisations and we have added this caveat to the discussion. However, there is evidence from shallow cores that the model is understimating precipitation in some of the higher elevation areas and overestimating precipitation at lower elevations. The areas with the largest inconsistency also correspond to some of the area with relatively high precipitation too. We have clarified this point as the surface elevation change data also uses a firn model to account for penetration of radar into the snowpack.

323:  The RACMO model is not really concerned with snow density in its precipitation forecast. I am just wondering if the work of Fausto et al (2018) on snow density provides any insight that helps explanations here vs basin discrepancy with GRACE. https://doi.org/10.3389/feart.2018.00051

RESPONSE: The RCM data we show here is Surfae mass balance, for both models this includes  densification processes, retention and refreezing and so forht. It would definitely be interesting to investigate further if there are systematic biases in the models for different regions of Greenland that can explain some of these features, but we feel that that is beyond the scope of this paper.

358: Appropriate density does for firn, snow and ice? Does this also reflect findings of Fausto et al (2018)

RESPONSE: See above comment

363: slope RUTH

RESPONSE: fixed

374:  Is it worth a key listing of key factors that may lead to stability or instability, calving front geometry, fjord geometry, mélange persistence, subglacial melt rates, and subglacial meltwater outflow? RUTH

384:  Fix reference

RESPONSE: fixed

405: Are rifts distinguished from crevasses in this process or are they just the end member of full thickness penetration crevasse?

RESPONSE: Here only vertical crevasses are considered. 

410: summer

RESPONSE: fixed

411: The lubrication hypothesis does make sense, on Petermann there is also exceptional basal melting particularly in summer that would lead to thinning and uplift. Does Figure 12 A and B suggest basal melting must be the key? RUTH

RESPONSE: We have not examined the submarine environment at Petermann glacier in this paper but thinning due to basal melt is undoubtedly significant as shown by a lot of the literature on this glacier. It is significant in that the stress intensity factor and fracture propagation is sensitive to ice thickness

415:  Stress intensity factor is use in Figure 12 C and D, but is note explained.  Please briefly do so near here.

RESPONSE: We have added the following explanation Stress intensity factor is a quantitative measure that shows the stress state of a fracture considering the applied loading and fracture geometry. When the stress intensity factor reaches a certain threshold, also known as fracture toughness, the fracture propagation is considered unstable allowing a fracture to propagate the entire thickness of the ice shelf rapidly.

417:  The persistence of Petermann ice tongue in an environment rich in surface water suggests the Petermann is not as vulnerable to this as other ice shelves that experience limited surface melt. MacDonald et al (2018) Annals of Glaciology DOI: 10.1017/aog.2018.9.  NEGIS also has had much surface melt ponding and streams for sometime. RUTH/JR  

RESPONSE: From MacDonald et al (2018) it follows that with predicted increasing air temperatures a higher density of lakes can form increasing the volume of water available for hydro fracturing and cause an increase in the ice tongue instability. However this effect can be mitigated by drainage of excess water through surface rivers. Figure 12 therefore only shows the potential effect increasing runoff can have on the stability of the Petermann ice shelf.    

420: For Petermann ice thickness increases considerably in last 25 km of flotation, does fracture depth, Figure 12D, have less ability to impact a much thicker tongue.  The floating shelf would of course not be as thick at the time terminus retreat reached that point, but would still be thicker than the generally thin limited thickness change in the region calved to date. RUTH/JR

RESPONSE: We have added the following paragraph to explain the sensitivity of the stress intensity factor to thickness. ”The final fracture depth is dependent on the ice thickness by the ratio of crevasse depth over ice thickness. A thicker ice shelf means that the ratio is smaller and, all other things considered equal, would reduce the final fracture depth. This is however only the case in the situation where there is enough water available for hydro fracturing, in the case no water is present the final fracture depth would increase slightly relative to the thinner ice shelf. “

423: The increasing vulnerability is of course going to occur with more melt, but the above reference suggests caution in applying this to a rapid collapse for an ice shelf that has existed in an environment of considerable melt for a long time. Such an ice shelf has developed a drainage system. This does not make it less vulnerable to additional melt driven thinning and the warming of the ice itself and increased water availability for crevasse penetration enhancement. Petermann already receives plenty of attention in the paper, so I am not advocating for increased coverage. RUTH/JR

RESPONSE: we agree entirely and have clarified this and added the McDonald et al  and Dow et al 2018 references to the discussion

452:  The conclusion should refer back to the ECV’s since the paper started with this. Something along the lines of “Continued independent validation and comparison, along with focused field work and enhanced modelling of calving will lead to more accurate ECV products. All of the data products reported here have a signal indicative of the significant impact and response of the GIS to climate change.”

RESPONSE: We have significantly revised the conclusions and added the following “Continued independent validation and comparison, along with focused field work and enhanced modelling of calving and ice dynamics will lead to more accurate ECV products. All of the data products reported here have a signal indicative of the significant impact and response of the Greenland ice sheet to climate change. This allows us to define the current changes in Greenland as a baseline for the important processes likely to drive future ice sheet change and consequent sea level rise.”

 

Round 2

Reviewer 3 Report

The changes made to the Mottram et al. manuscript are substantial and greatly improve the presentation of this work.  However, there are still a number of issues that should be resolved before the publication of this manuscript.

General comments

- In the comparison of RCM outputs of SMB with GRACE, is the SMB data processed to be "GRACE-like" similar to Alexander et al., TC (2016) or Velicogna et al. GRL (2014)?

- Section 4.4 needs to be rewritten for clarity.  These neglections are non-trivial (firn processes, GIA, column density, compensating for signal loss at the ice margins).  Was the radar altimetry data scaled using a laser altimetry estimate with firn processes removed?  Why was Glacial Isostatic Adjustment neglected?  While GIA uplift/subsidence is a minor correction, neglecting the process seems strange.  How different are the laser and radar altimetry estimates over coincident areas away from the coasts?  What do you mean by an appropriate density?

- Several references (L364, 379, 467, 534, 537) have spacing issues with the references

Line-by-line comments

L78: SAR interferometry is a piece of the input-output-method or mass-budget-method

L86-87: I would mention Glacial Isostatic Adjustment (GIA) here was well

L94: Perhaps remove "then" before "propagate"

L96: Maybe don't use "though key".  Perhaps "necessary" instead

L97-100: split this sentence for coherency

L100-103: Perhaps "in and around Greenland". Also should be "Operation IceBridge".  Fix the citation.

L103: I would replace with "The ESA CCI project has consolidated, standardised and integrated satellite remote sensing data to create high quality datasets that are accessible to scientists and policymakers."

L145: There's a problem with the citation here

L175: Need a space after "loss"

L201: Envisat (here and elsewhere)

L205: elevation (misspelled)

L237: Do both of these estimates use the same GIA correction?  Which correction is used?

L248: need a space after "e.g."

L268-270: awkwardly constructed sentence

L287-288: could be rewritten for clarity

L329: sentence could be rewritten for clarity

L330: sentence is awkwardly constructed

L332-335: should cite Krabill et al., "Greenland Ice Sheet: Increased coastal thinning", GRL (2004) here.

L345: should split this sentence after Colgan

L373: there are modeling groups investigating the effects of geothermal heat fluxes.  could say that these are not included or investigated here.

L401-403: differences could also be related to signal leakage in the GRACE estimate.

L405-407: awkwardly constructed sentence

L407-413: sentence is awkwardly constructed

L451: oddly placed comma

L456: remove "On the other hand"

Figure 8: labels are very small

Equation 1: citation problem in caption (unlinked).

L546-548: sentence is awkwardly constructed

L568: I would remove "but is beyond the scope of the current work"

L580: GRACE Follow-on

L599: GRACE-era

L602: RCM-derived

L616: lowercase H in however.  or split the sentence

L623: Grounding line position is one metric of ice shelf condition but you could mention some other metrics (ice shelf thickness, areal extent of meltwater ponding, basal melt channelization)

L630: sentence is awkwardly constructed