Dear referee, 

The attachment displayed here seems to relate to comments made on a different paper (Marshalek et al). There seems to be a technical error of some kind. 

Tancrede Leger 

##################################

Main comment 1: ".....I think one thing that will

truly be valuable to add to this paper would be a sort of “instruction manual” for how other research

groups should structure their investigations so that the data can be easily incorporated into PaleoGrIS...."

Authors' response: Many thanks for this comment, that is a very good point and something that was still being discussed when submmitting the manuscript. We have decided to add a text file with clear instructions regarding this in the supplementary data: i.e. the PaleoGrIS 1.0 database, in the home folder, so that it cannot be missed. Anyone who downloads the database will easily see it. However, please note that this work is different from databases such as Bedmachine in a sense that the mapping of isochrones involves a lot of interpretation by the person mapping and involves mannual data filtering. This means future iteration following a similar method cannot be achieved by re-running a standard code or by using an online tool/platform that people can upload their new data to. It is more complex and requires people to re-work on the product for many months.

However, we still think there are ways by which the community can collaborate and contribute to future versions: by making the process of keeping up with new data easier for investigators whishing to produce new versions.

 

We here paste the content of these instructions:

"HOW TO CONTRIBUTE NEW DATA TO FUTURE VERSIONS OF PaleoGrIS: 

We would like new future versions of PaleoGrIS to be a wider community efforts and thus believe there are numerous opportunities for collaboration and contributions from a wide range of research teams and scientists. 

We believe the best way for new empirical investigations to actively contribute to new versions of PaleoGrIS when providing new constraints

on the retreat of the GrIS margin, as well as periphery outlet glaciers, is to download the PaleoGrIS database and to locally (offline) update 

the products (under the same format as original) with their new data (only if published peer-reviewed). We then reccomend to attach the updated products as as supplementary data/info to their new publication. The next useful step could be to contact the main authors of PaleoGrIS (Dr Tancrede Leger, Prof. Chris Clark, Dr Jeremy Ely) with the link to the newly published article which includes access to the the supplement and updated products of PaleoGrIS.  

The data that could easily be modified and improved would for instance include the Excel tables for the Geochronological database: simply by adding lines with new data at the bottom of the spreadsheets. A very useful contribution would be to modify the PaleoGrIS isochrone polyline shapefiles (with 4 different confidence levels) by updating them in the specific area studied by the investigators, provided that the same mapping methodology as original product (detailed in PaleoGrIS paper) is employed.

Alternatively, if you believe that entirely different and new Greenland-wide products resulting from your research would suit being incorporated into future versions of PaleoGrIS, please get in touch with the main authors, we would be happy to discuss! "

###################################

Main comment 2: "....The shapefiles are full of tiny triangular polygons, and they intersect in ways that will make certain kinds of GIS analyses fail. I have included a picture showing what I mean (Figure 1). If possible, I suggest finding some way to simplify the margin files so that there are fewer of these small polygons."

Authors' response: We thank the reviewer for spotting this issue. We have now modified the isochrone polygon shapefile products to remove these shapes. To do so, we re-rasterized the shapefiles to a lower resolution (75m) to simplify the margins, and re-converted to polygon shapefiles but keeping the "pixelized" aspect of margins to avoid the triangular shaped polygons to show up, which was a result of vectorization when converting from raster to polygon shapefile (Arc simplyfying things). We also removed small holes (<10 km2) in BedMachine mask product to simplify the resulting polygon shapefiles: this did not change the reconstructed ice-sheet area by more than 0.5 %. Thanks to this comment we also spotted numerous polygons that were disconnected from the main ice-sheet perimeter. These have now been removed, resulting in a cleaner product. We however do not reccommend using these polygons for model-data comparisons (see ReadME in relevant folder) and we instead encourage users to favour the isochrone polyline shapefiles, or the NETCDF isochrone or buffer products. 

###################################

Minor comment 1: "The light blue text and lines on this figure is very hard to read...."

Authors' response: Thank you for pointing this out, the colour of text and lines has now been changed to a bright yellow. 

Minor comment 2: "There is no Yang et al 2016 in the reference list"

Authors' response: This has now been added, thanks for spotting the error. 

Minor comment 3: "Google Earth has inaccurate georeferencing in Greenland...."

Authors' response: That is good to know, thanks for pointing that out. However Google Earth was not used for direct mapping of features. 

We mapped from Arc using the Arctic DEM and georeferenced imagery products, while Google Earth was exclusively used to inform the mapper when we wanted to visualize the area of concern in 3 dimensions. We therefore do not expect the innacurate georeferencing in Google earth to have any impact on our mapping. 

Minor comment 4: "In some places, MIS6/5e (and older) aged sediments have been preserved...."

Authors' response: We agree that this sentence needs to be more conservative, as although it is rare at the scale of the 

entire ice sheet periphery, it has been observed in some places, as pointed out here. 

The sentence was thus modified to that is reads: ".... and less likely to be preserved (with exceptions: i.e. Mejdahl & Funder, 1994; Kelly et al., 1999), we consider our mapped ice-marginal landforms to have been deposited during the last deglaciation"

Minor comment 5: "In some places around Greenland, there has been over 100 m of uplift since deglaciation...."

Authors' response: We agree that this will be important to re-assess once a publicly-available sea level indicator and proxy database exists.

However, we note that recent assessments of the impact of GIA-related uplift on cosmogenic exposure ages in Grenland (i.e. Jones et al., 2019) show that ages would be corrected corrected by a maximum of 7%: which, for Holocene ages (vast majority of deglacial ages in Greenland), would not represent a correction in the order of thousands of years, but rather in the order of a few hundred years. The key factor to consider is when did the uplift happen, and how quickly, along with how much. If most of the rebound had already occured prior to, or towards

the onset of the exposure period (say early Holocene), than the impact of GIA correction on the resulting exposure age would be minimal and well below analytical uncertainties.

Minor comment 6: "I find the phrase “continuous perimeters” to be confusing"

Authors' response: The term "perimeters" has now been replaced by "margins", more widely used in this specific context. 

Minor comment 7: "The Dalton et al. (2020) dataset is essentially unchanged from Dyke (2004) for Greenland

(aside from the Nares Strait), so I suggest using a different reference here"

Authors' response: Thanks for pointing this out. The reference in this sentence was replaced by the more appropriate review by Funder et al. (2011)

Minor comment 8: "I would propose that maybe the reason there has been no data found from prior to 14 kyr BP is because it is possible the ice sheet remained at the shelf edge until the Bølling–Allerød sea level rise event"

Authors' response: We agree with this hypothesis. Our team is currently running a numerical modelling experiment using PISM and 

producing an ensemble of numerous simulations with varying parameter configurations from 24 kyr BP to present. Despite highly different parameter configurations and despite modifying climate and ice dynamics forcings by significant amounts, all simulations produce a GrIS that remains at its maximum extent untill 16-15 kyr BP, prior to rapid collapse during the BO warming: see preliminary findings here: https://doi.org/10.13140/RG.2.2.31786.80328 

Minor comment 9: "I am curious about the choice to have the isochrone intervals be centered intermediate of the 500 year time time steps...."

Authors' response: The isochrones were imagined to be given a time range rather than one time stamp, from the early days of us thinking about the reconstruction. It was important to us to try to better acknowledge uncertainties in dating and also better seperate the geochronological uncertainties from the ones arising from spatially-variable spread in empirical constraints, when attaching time to our isochrones. Knowing this we chose to have time ranges with simple rounded values (e.g. 10 kyr BP - 9.5 kyr BP) which makes it slightly easier for both:

1- Mapping isochrones with numerical dates and labels turned on in Arc and rounded to 1 decimal: the mapper can more easily identify dates belonging to a specific time range when reading them on a map with 1 decimal rather than 2.

The likelihood to make mistakes is lower.

2- Easier for writing in the paper when referring to a specific isochrone and time range, and easier to read. We believed isochrones with time ranges such as 8.75 - 8.25 kyr BP would be more difficult and confusing to read, and would rely on thresholds between two isochrones being expressed in decades, which from a geochronological point of view would feel a bit odd given that uncertainties in the dating are in the order of centuries.  We however do acknowledge that users of the database wanting to use a single time stamp (+/- uncertainty), i.e. for comparing with model time for instance, will have to use rather odd centre values (e.g. 8.75 +/- 0.25 kyr BP) for a single isochrone. Although this might be a bit different than usual, it would not change the result or the analysis in any way, as is already pointed out in the reviewer's comment. 

Minor comment 10: "I suggest referencing England et al. (2006) for a detailed overview of the Innuitian Ice Sheet, including the glaciation of the Nares Strait...."

Authors' response: Highly relevant and incredible paper indeed ! Many thanks for the suggestion, the reference has now been added to this paragraph.

Minor comment 11: "Based on the GIA modelling by Lecavalier et al. (2014) (and my own, yet to be published study), I expect that the more expansive scenario is more likely."

Authors' response: We agree, and believe that new arising empirical data constraining Last glacial extent will likely increasingly make that argument, as this has been the trend with studies from various regions of Greenland lately.       

Minor comment 12: "The margin lines in these figures are too thin to make out the differences in the levels of confidence...."

Authors' response: Indeed the margin lines are thin, but they are visible to make out different confidence levels if the viewer zooms in to expand a single subpanel to about the size of their screen. However we tried and realized that making them thicker to the point where they could be distinguished at the whole figure scale would result in hiding most of the information regarding where the actual margin is located: which is the primary goal of this figure. We would thus be more inclined to not change the thickness of those lines, if that sounds reasonable. 

Regarding the caption: we were struggling to fit the full legend information on the figures with numerous sub-pannels filling the whole space. In order to not compromise their size, and in an attempt to address the comment, we have added a sentence in the figure captions referring the reader to the legend of the first figure to appear in this series (8-13) which all share identical styles and scales/colormaps: i.e. a reference to Figure 8, which features more white space and thus the complete legend with all labels.  

Minor comment 13: "I am not sure what is meant by “miscellaneous ice configuration”.

Authors' response: Indeed we agree the word miscellaneous is inappropriate.  We have decided to use the term "characteristic" instead and modified the sentence accordingly. 

RC3:

Overall, this a useful contribution to our overall understanding of the deglacial history of the Greenland ice Sheet (GrIS). The methods employed are tried and tested on other recent ice sheet reconstruction projects (e.g. Britice-Chrono). The mapping approach enables the broad demarcation of ice marginal positions, and in combination with pre-existing geochronological data (sourced and collated from multiple sources; e.g. Ice-D), this provides a series of isochrones with which to constrain ice retreat between 14 and 6.5 ka. This period of GrIS is already well known from multiple sectors of the GrIS, but it is useful to visual ice sheet wide retreat patterns and rates to help elucidate ice sheet response to climate forcing. The construction and provision of this database represents a serious amount of work which will make a valuable contribution to the research community.

Authors’ response:

We thank the reviewer for these positive comments.

RC3:

However, the latter parts of the paper that attempt to explore regional drivers of ice sheet change are weaker and require some additional critical discussion.

The lack of mention/appreciation of the sub-surface ocean temperatures in driving ice off the continental shelves undermines the discussion of ice sheet response to forcing. Atlantic Water ingress and the influence of the East and West Greenland Currents (different leads/lags pan-ice sheet) during deglaciation was pivotal in the response of the marine terminating margins to ocean heat flux. This needs acknowledging and discussing.

Authors’ response:

We thank the reviewer for this important point. As detailed further below in response to the relevant specific comment: we have added some text on this point in the “4.2.1  Ice-sheet-wide response” section. Regarding section 4.2.2 on regional responses, a deliberate choice was made to focus our discussion more on the ice-sheet- and sector-scale topographic heterogeneities and asymmetries, as we believe such drivers to be less mentioned by the literature than climatic and oceanic forcings, but potentially also very important. See comment directly below for more justification on this. That being said, in paragraph 1 of section 4.2.2, we still briefly introduce the importance of climatic and ocean forcings as they need to be at least mentioned. In this paragraph, in response to the reviewer’s comment, we have added some text on the importance of leads and lags in sub-surface Atlantic Water ingress across the ice-sheet periphery being responsible for different ice-retreat responses.

 “Moving from ice-sheet wide to a focus on individual regions (the ice catchments of Rignot & Mouginot, 2012), we find heterogeneous retreat patterns with apparent differences in both the magnitude of regional areal extent loss and the speed of ice margin retreat (Figs. 16, 17). Spatially variable climate and ocean forcings (Fig. 20) may have caused some of these differences. For instance, the saline Irminger Current transporting warm Atlantic waters offshore South Greenland (Fig. 1), likely caused this region to experience early, more intense oceanic and atmospheric warming (Levy et al., 2020; Fig. 20c). The inflow of relatively warmer, more saline sub-surface water originating from the North Atlantic Current was likely crucial in causing sub-shelf melt near the grounding lines of marine-terminating ice-sheet margins (Knutz et al., 2011; Jennings et al., 2014). It has been argued significant leads and lags in early deglaciation across Greenland may be partly related to the complex advection patterns of this water mass enabled by the Irminger and West Greenland current (Fig. 1), influencing ice retreat in SE, SW and CW Greenland, and the distinct intermediate Return Atlantic currents influencing ice-margin retreat across the NE and CE Greenland shelves (Hebbeln et al., 1994; Nørgaard‐Pedersen et al., 2003; Hansen et al., 2022).”  

RC3

 The role of sea-level is partly mentioned but not explored in detail. The switch from a full glacial sea, up rapid uplift (and falling RSL) being particularly pivotal in grounding line instability as the ice retreated to the coast at the opening of the Holocene. The loss of ice shelves and sea-ice conditions also need a mention, though the role of both in influencing rates of deglaciation are much difficult to assess.

Authors’ response:

We agree that RSL change is important and likely plays a significant role in driving Ice-margin response by acting on the spatio-temporal variability of ice-sheet margins transitioning from marine to fully grounded terrestrial system. Paragraph 3 of discussion section on “Influence of heterogeneous onshore bed topography” touches on that and we have added a sentence to make it clearer that spatially-heterogeneous RSL and GIA-induced uplift are important drivers in this case:



“Due to the added ablation effect of calving, topographic heterogeneities between Fjords and inter-Fjord sectors can generate different modes of deglaciation for adjacent outlet glaciers, as observed in North Greenland (Larsen et al., 2020). Therefore, the transition from marine to land-terminating margins, which occurred at different times in different regions due to variabilities in glacial isostatic adjustment uplift and thus in relative sea level change (Long et al., 2011), is an essential factor influencing the dynamics of former ice sheet margin retreat in Greenland.”

However, our discussion on “regional responses” is mainly focused on the heterogeneous magnitude and rate of retreat between different sectors of the ice-sheet, with a particular focus on potential topographic drivers of such heterogeneities related to asymmetries in subglacial topographies. We have decided to focus the discussion on this particular point as we believe it is more rarely mentioned than climatic and oceanic drivers, arguably more important drivers at the scale of individual outlet glaciers, which are the subjects of the majority of Greenland studies. Because our study investigates ice-sheet wide, or sector-wide responses quantitatively, we thought focusing on these discussion points would be interesting. Our limited discussion is certainly not aiming to be a large review of all possible drivers of ice-sheet margin retreat during the deglaciation. We also would like to point out this is not what we set out to do or say we will do in the paper title nor in the abstract. That would most likely require a separate paper focused more on the controls. This paper is not designed to be a review paper, but rather the release of a database, ice-sheet wide reconstruction, and also some additional thoughts and hypotheses that arise from the quantitative analyses conducted with this reconstruction. This distinction is important and the confusion between the two is possibly why the reviewer is feeling that many important discussion points are not developed enough. However, we agree that a large review paper that explores all these different controls would certainly be interesting and needed.

We have modified the sentence leading to the section of the discussion on regional responses, to make this point clearer:

“However, at the regional to valley scales, we believe non-climatic and oceanic factors also help explain the variability. We discuss and hypothesise some of these mechanisms in the following sections, with a particular focus on ice-sheet- and sector-scale topographic heterogeneities and asymmetries. Whilst numerous other non-climatic factors, such as spatial heterogeneities in relative sea level and sea-ice cover change, for instance, may also play roles in controlling such regional heterogeneities, we here only focus on what we believe to be main or first-order drivers.”

Specifically on RSL change: heterogeneous RSL responses between regions is mostly related to the magnitude of GIA-related uplift post deglaciation, which itself is a consequence of 1) the magnitude of ice build-up in the specific region during advance to full-glacial conditions, and 2) the magnitude and speed of ice-margin retreat and thinning during deglaciation, and 3) upper and lower mantle viscosity, and 4) the magnitude of far-field RSL signals . Factors 1 and 2 are likely the most potent, and we believe regional heterogeneities in RSL change are thus more likely to be a consequence of differential ice-margin evolution (itself more caused by climatic, oceanic, and topographic feedbacks), than the other way around: the cause. Although we agree subsequent RSL change could then enhance or instead dampen the trend in ice-sheet margin retreat signals, via feedbacks related to grounding line instabilities etc, but these would have occurred after deglaciation of continental shelves and major re-arrangements in ice-sheet geometries, during the early Holocene, and we suspect these feedbacks to be less impactful on sector-wide ice sheet areal extent change than pre-conditioning topographic settings causing asymmetries in ice-sheet geometries.  

Furthermore, we are not sure how the loss of ice shelves and sea ice conditions would have impacted the ice-sheet margin retreat magnitude and rates differently between different sectors of the ice sheet, and we have insufficient information on these to conduct any analyses. We think this is an interesting hypothesis that awaits further work.



RC3

The section on inertia requires additional thought. It touches on some relevant concepts but is too brief. This is particularly important in relation to the rate of Early Holocene ice loss and should be placed into context when considering the last 4500 yrs as Neoglacial conditions have prevailed. How do ice sheet-wide response rates to forcing differ during pronounced periods of -ve  (Early Holocene) and +ve (NeoGl) mass balance?  Have a look at the Young et al paper et al. (2013) paper for the Early Holocene and consider what are the implications of the Neoglacial for future GrIS mass balance change?

Authors’ response:

Indeed the observed lag in areal extent response is an ice-sheet wide signal, but the story is likely to be much more complex at the outlet glacier scale. We have added a sentence highlighting the findings of Young et al. (2013) for the Jacobshavn Isbrae outlet, to show that some were rather dynamic and well coupled to Holocene temperature changes. We have also added some details on the delay observed, at the ice-sheet scale, in the response to positive mass balances during Neoglacial cooling. We are grateful to the reviewer for pushing us on this point, and helping improve our argument and being clearer on the local vs. regional distinction.

“Therefore, our ice-sheet-wide empirical reconstruction agrees with previous observations that during the Holocene, the ice-extent response of the Greenland Ice Sheet lagged the cessation of warming, potentially by thousands of years. This delay in turn caused a lag in the response of the ice sheet to positive mass balances during mid-to-late Holocene cooling. Indeed, while both summer insolation (Berger & Loutre, 1991) and temperature proxy records (e.g. Erb et al., 2022) reveal atmospheric cooling across Greenland from 8-6 kyr BP to the pre-industrial era, significant neoglacial glacier expansion did not occur until 2.5-1.7 kyr BP (Kjær et al., 2022). Such lagged response observed at the ice sheet scale was however not ubiquitous across the entire ice-sheet margin, as certain more dynamic outlet glaciers were well coupled with short-lived early Holocene temperature changes (e.g. Jakobshavn Isbræ; Young et al., 2013b). Centennial to millennial-scale inertia of the Greenland Ice Sheet following warming has been previously suggested (e.g. Yang et al., 2022), and has implications for ongoing climate warming generating committed future mass losses, and thus sea level rise contributions, that may last for centuries to millennia (e.g. Greve & Chambers, 2022).”

RC3

Finally, the discussion Greenland ice streams with respect to their dynamic behaviour though time. The designation of an ‘isbrae’ type ice stream originated in Greenland. Selective linear erosion and topographic over-deepening have controlled valley/fjord evolution of Quaternary timescales. Topographic control of ice stream position and behaviour through time is particularly critical as the ice steps back from mid/inner shelf positions into the coastal fjords. So, topography becomes pivotal in controlling ice sheet/ice stream dynamics in the window 12-8ka. Caution is required when linking ice stream behaviour to climate forcing. On the mid/outer continental shelves, ‘soft-beds’ and low gradient seafloors may have provided better conditions for ice stream switching/re-arrangement earlier in the glacial cycle but we know little of such history as yet.

Authors’ response:

we thank the reviewer for this interesting point. The hypothesis of potential ice stream reconfigurations during deglaciation is only briefly mentioned in the discussion: where we focus more on the idea of ice divide migrations as a result of possibly different ice sheet geometries to present during full glacial extent, which may in turn have modified the configuration of ice streams when comparing ice sheet velocity fields at full glacial extent (e.g. 17 kyr BP) with their configurations towards the early Holocene, when indeed coastal topographies likely became more important in controlling ice stream position by constraining ice flow. There is actually a scale issue hidden away in these arguments in that whilst individual ice streams may respond at different times and appear highly influential, in actual fact they might be stealing or giving ice flux away to adjacent catchments (whose boundaries migrate), such that it all gets smoothed out a regional or ice sheet scale. This was a key finding for example for ice streams of the Laurentide Ice Sheet (Stokes et al. 2016) where ice stream activity was found to scale to ice sheet volume.

With regards to the increasing importance of pre-existing topographies: we have added a sentence to the third paragraph in the discussion section on “Influence of heterogeneous continental-shelf topographies”:

“Consequently, during the Late-Glacial and early Holocene periods, a significant ice divide migration may have occurred in NE Greenland. The recent findings of a possible re-configuration of NEGIS during the early Holocene (Franke et al., 2022) could perhaps be linked to such regional adjustments in ice-flow. Around Greenland more generally, the ~12 to ~8 kyr BP window may have been critical for ice-stream re-adjustments, as the ice sheet stepped back from mid/inner shelf positions into coastal fjords and thinned towards its margins, and as pre-existing overdeepenings became increasingly important in channelising ice flow.”    

However, we do not see where in the text we are linking ice stream behaviour to climate forcing, since we decided to explicitly focus on non-climatic drivers in our discussion.

RC3 : SPECIFIC COMMENTS

RC3

Abstract:

This is a reconstruction based on pre-existing geochronological  and geomorphic data with  new ice margin positions mapped.

Authors’ response:

Although we are not entirely sure, we assume this comment refers to the statement in the abstract about PaleoGrIS being: ”the first Greenland-wide isochrone reconstruction of ice-sheet extent evolution through the Late-Glacial and early-to-mid Holocene informed by both geomorphological and geochronological markers”. This sentence was to highlight that at the scale of the entire ice sheet, no other previous investigations had combined geochronological markers, and mapped ice-marginal landforms, to produce a detailed reconstruction of past ice margin evolution through the studied time period. However we agree this can become confusing to the reader this early in the manuscript. We have modified the sentence so it reads: “we here present PaleoGrIS 1.0, a new Greenland-wide isochrone reconstruction of ice-sheet extent evolution through the Late-Glacial and early-to-mid Holocene informed by both geomorphological and geochronological markers.” 

RC3

Li 78: starting from the wrong state – reword

Authors’ response:

The sentence was reworded to: “As few models include the committed response to late-Quaternary environmental changes, current projection simulations might effectively be starting from a possibly unrealistic state, whose influence is thus far unquantified.”

RC3

Li 85: and pattern of ice margin retreat during the Late-Glacial and early-to-mid Holocene periods (between ~14 and ~6 kyr BP) remains poorly constrained. This is a misleading statement. The central west coast (Disko and Uummannaq regions) has a well constrained deglacial history. As do many areas in the S, SE, SW, E, NO and NE.

Authors’ response: 

Apologies, in fact this sentence initially included an extra statement at the end to highlight that the deglacial history: “remains poorly constrained in numerous regions*”. It is indeed misleading in its current form. The take home message being empirical constrains are by nature highly heterogeneous. We have modified the sentence to:  “remains heterogeneously constrained in numerous regions.”

RC3

Li 88: Was the GriS influenced significantly Meltwater Pulse 1A? Where is the evidence? Elaborate.

Authors’ response: 

This sentence: “During that time, the ice sheet was responding to rapid and high-amplitude fluctuations in atmospheric and oceanic boundary conditions (e.g. the Bølling-Allerød Interstadial; Buizert et al., 2018), whilst also adjusting to fast rates of relative sea level change following the demise of other ice sheets (e.g. during Meltwater Pulse 1A; Lin et al., 2021).” was merely meant to highlight the main types of significant environmental changes that occurred during the Late Glacial and early-to-mid Holocene periods, changes which were likely to cause re-adjustments in Greenland ice sheet geometry and dynamics at that time. Namely, changes associated to 1) variations in atmospheric conditions, 2) variations in ocean conditions (temp, salinity, etc) ; 3) variations in relative sea level, which necessarily induced changes in the proportions of floating versus grounded ice towards the ice sheet margins, causing important grounding lines migrations. The example of Meltwater Pulse 1A was merely cited to mention one particular episode with rapid Relative Sea Level change, within the time studied here. However, we agree this can confuse the reader as one might expect a causation between this specific event and ice-sheet margin evolution, when reading this. This would not suit an introduction as is too speculative. We have thus removed this cited example from the sentence. 

RC3

Li 92: A key question concerns how far the ice sheet retreated behind its contemporary margin in response to the warming of the Holocene Thermal Maximum (10-5 kyr BP; Cartapanis et al., 2022), You need to add more references here. This is a question that has been asked way before 2022, by many, many authors.

Authors’ response: 

Indeed, and a 2022 citation is also relevant to show that, importantly, it remains an open question to this day. We have added Weidick et al. (2004) to this bracket.  

RC3

Li 97: Again, a lot of references missing a lack of appreciation of many studies from the last 50yrs that have built our understanding of GrIS retreat over the last 15,000yrs. Acknowledge some of the pioneers, not just recent review papers!

Authors’ response: 

We agree this specific bracket could feature older investigations as examples. We have added Ten Brink & Weidick, 1974; Hjort, 1979; England, 1985 to the bracket. 

RC3:

Li 103/4: facilitate targeting of understudied regions for future fieldwork.  Why do you think there are understudied parts of Greenland? I understand the statement, but the inference is ‘the community’ have left holes, when of course this is an issue related to access and logistics. This does the Greenland research community a bit of dis-service.

Authors’ response: 

We fully agree with the reviewer that the main reason for certain regions being less studied is naturally access and logistics, and we are certainly not arguing that point here. However, this does not mean it is the end of fieldwork in Greenland, there are plenty of opportunities for exploring new field sites. With this sentence, we are only referring to an important challenge when choosing an area to study, which is knowing precisely what work has been done in the past, where people have worked and collected data, in order to make sure not to duplicate work. This can be tricky and requires conducting a vast review of a vast literature which can be time-consuming. Not every research team willing to work in Greenland and answer new research questions has necessarily been through a thorough reviewing process, nor has worked there for several years prior. This would also be the case for new researchers in the community such as PhD candidates for instance. The point we are making here is merely that databases and reviews can be useful to help members of the empirical community to efficiently assess current empirical knowledge and its geographical spread, and quality. For instance, after presenting PaleoGrIS 1.0 at 2 conferences in 2023, we already have colleagues in Spain who have asked us to share data from our reconstruction and database, in order to help target new unstudied field sites next summer for cosmogenic dating, in places where they had identified good geomorphological records. With this work, ourselves at Sheffield (PALGLAC team) have identified a site near Nuuk (in fact quite easy to access by boat), which currently features a lack of geochronological constraints. This resulted in us going there in August 2023 for a field and sampling campaign: and the analysis is now ongoing. Although we fully agree the Greenland empirical community has constructed a fantastic library of empirical data over many decades (and if anything, our compilation is a good and easy way to visualise that for anyone), we still believe there are plenty of opportunities for more data collection. We believe the former production of open-access databases and reconstructions with quality assessments, and data-filtering conducted for the specific purpose of tracking ice-sheet margin change through time (e.g. Britice, PATICE, DATED, Dalton et al), has undoubtedly been useful for many research teams willing to assess the state of knowledge for other palaeo- ice sheets. 

RC3

Li 175: Contradicts your earlier statement about a lack of work.

Authors’ response: 

we have modified the prior statement which we agree was misleading. See response to comment above.

RC3:

Li 225: and unlikely to be preserved, we consider our mapped ice-marginal landforms to have been deposited during the last deglaciation…. Why make this assumption? This is a total unknown in Greenland. If you ran a similar exercise in Iceland, you would find that overridden moraines are extremely common in the geological record. In Greenland, there may not be as many examples of overridden moraines, but there are places where moraine spacing gets condensed due to offshore/onshore transition, topographic  and climate control. So, the rates of ice margin change are extremely complex. Have a look at the extensive literature on the fjord stade moraines in Disko Bugt (See Young et al. 2013). This is a clear example of both ice stream and interstream areas feeling a rapid climate downturn (9.3. and 8.2 events) but also being sensitive to ice flux and topography.

Authors’ response: 

We thank the reviewer for this point, and we agree that this assumption, based only on the low quantity of observations of preserved pre-LGM ice-marginal deposits in Greenland, is likely too confidently asserted in this sentence. More realistic would be to assume such deposits are “less likely to be preserved” than the overwhelming presence of ice-marginal deposits related to the last deglaciation, which we fully agree can be more or less condensed depending on how sensitive a specific outlet glacier was to changing climate and ocean conditions. We have modified the sentence in that way. 

RC3:

Li 337: which given the young nature of the ages (<14 kyr BP) and the relatively low magnitude of surface elevation change related to glacial isostatic adjustment during the Holocene, This is misleading. In many parts of the ice sheet there has been 50-120m of uplift postglacially. The authors need to engage with the sea-level literature.

Authors’ response: 

We fully agree and are aware of the magnitude of GIA-related uplift post-glaciation in Greenland. The important word here is “relatively”. In fact, what was meant here is that this magnitude of uplift (50-120) is relatively low when compared to many other deglaciated regions of the world (e.g. Canada, West Antarctica, Norway, Sweden, Finland…: where uplift has been > 400 m in numerous places), and is low specifically when analysing the sensitivity of TCN exposure ages to surface elevation change through time. We here copy the response to reviewer 1 on this same point:

“We agree that this [the impact of GIA uplift on TCN age correction] will be important to re-assess once a publicly-available sea level indicator and proxy database exists. However, we note that recent assessments of the impact of GIA-related uplift on cosmogenic exposure ages in Greenland (i.e. Jones et al., 2019) show that ages would have to be corrected at maximum (moreover only in SW Greenland) by ~7%, which, for Holocene ages (vast majority of deglacial ages in Greenland), would not represent a correction in the order of thousands of years, but rather in the order of a few hundred years. The more important factor to consider for TCN ages is: when did the uplift happen, and how quickly? This can be harder to determine. If most of the rebound had already occurred prior to, or towards the onset of the exposure period (say early Holocene), then the impact of GIA correction on the resulting exposure age would be minimal and well below analytical uncertainties.”

In order to make our point clearer here, we have added the range of magnitude of uplift in Greenland to the sentence to better indicate that indeed, such magnitude of surface elevation change is relatively low when specifically assessing its impact on TCN ages:

“, which given the young nature of the ages (<14 kyr BP) and the relatively low magnitude of surface elevation change (50-120 m) related to glacial isostatic adjustment during the Holocene, is not thought to cause age offsets greater than analytical uncertainties in most Greenland regions (Jones et al., 2019).”

RC3:

Li 884: drop significantly between 6.5 kyr BP and 2-1 kyr BP, when the ice sheet was responding to the Holocene. This needs rewording. Post the HTM the ice sheet reached its minima sometime between 7 and 5 ka probably. By 5 – 4.5 ka we then we see the onset of the Neoglacial and ice sheet re-expansion culminating in the LIA.

Authors’ response: 

We have modified this part of the paragraph to reword it like so: “Following the Holocene Thermal Maximum, between 6.5 kyr BP and 2 kyr BP, the number of available dates drop significantly (Fig. 7). This coincides with the ice sheet reaching its minimum Holocene extent (between ~7 and ~4.5 kyr BP), which in most Greenland regions was either as extensive, or more retreated than today’s margin position (Larsen et al., 2015; Briner et al., 2016). After ~5-4 kyr BP, the onset of Neoglacial cooling caused ice-sheet re-advances culminating in the Little Ice Age, which explains the small relative increase in the number of dates in our compilation after 2 kyr BP (Fig. 7).”.  

Please note we use slightly more conservative time ranges as suggested given that we are referring to the entire ice sheet margin and that the precise timing of minimum ice-sheet extent during the Holocene is both mostly unknown, and likely spatially heterogeneous.

RC3:

or more retreated around most of its perimeter – reword

Authors’ response: 

We have done so accordingly, see response directly above. 

RC3:

Li 780: For TCN exposure dating, misleading ages can result from nuclide inheritance causing too-old apparent ages, postdepositional disturbance causing too-young apparent ages, or laboratory contamination/errors potentially causing both (Dunai, 2010). But more often than not these TCN ages have already been discussed as part of the original research publication.

Authors’ response: 

We completely agree with that statement, which is why our identification of outliers in the database only occurs when described as such by original authors, without further interpretation, as is mentioned in the methods section (see line 361).

This specific sentence only re-iterates the potential reasons for which TCN exposure ages can be misleading, which are inherently different to other dating techniques such as Radiocarbon dating. This is simply a reminder of the theory for readers that aren’t familiar with this technique, as scientists with different backgrounds might be interested to read this work.

RC3:

Li 1640. There is no mapped reconstructions for the Uummannaq region – why - too complicated ?

Authors’ response: 

The PaleoGrIS 1.0 reconstuction covers the entire ice-sheet periphery, including the Uummannaq region. Our reconstruction for that region is described in section 3.2.7., with ~600 words allocated. However, it is indeed not shown in a figure of the style of figures 8-13. This is solely because these figures only show a sub-selection of the reconstruction, in order to provide a first order visualisation of our work. Producing such figures for all Greenland regions would take a lot more space in a paper that already features a total of 21 figures. There are no other reasons for not presenting the reconstruction in this area with such a figure. Our reconstruction in numerous other regions in NW, NO, NE, and SE Greenland are also not shown with such figures. 

RC3:

Li 2100: I think the community already know where the holes in the data are!

Authors’ response: 

We agree this must certainly be the case for researchers and teams with ample experience in investigating Greenland’s past evolution. However, we believe it remains useful to clearly present and highlight areas with knowledge gaps, in particular to researchers new to this field, or with little experience in Greenland. Figure 19, for instance, was presented at two distinct conferences in 2023. Both times, researchers engaged specifically with that figure, and wanted to have a closer look at it after the presentations. Clearly, having a quick look at where these gaps are broadly located, is of interest to members of the empirical community.  

RC3:

Li 2115: "It must be noted that such data gaps are most likely related to the increased logistical, safety, access, and financial constraints associated with field investigations in more remote and problematic regions. Furthermore, a great source of uncertainty impeding a better understanding of the deglacial dynamics originates from the ice sheet being generally more extensive today than between ~6 and ~2 kyr BP, when responding to the Holocene Thermal Maximum". This is just stating the obvious.

Authors’ response: 

We agree, that is definitely the case for someone with ample knowledge of the past Greenland Ice Sheet evolution. However, it is safe to say this is not necessarily the case for all potential readers of this work. Scientists mostly studying other ice sheets and glaciated regions might not necessarily be aware of the former Holocene dynamics specific to the Greenland ice sheet. We believe it doesn’t hurt to re-iterate. 

Furthermore, we also wanted to write the detail about complex access and logistics so that readers don’t make quick conclusion on why such regions appear to be understudied in Greenland. This is actually related to a previous comment made by the reviewer on “doing the Greenland research community a bit of dis-service” (see earlier). That sentence (“it must be noted that … “) was introduced here to make sure too quick conclusions would not be made, and to re-iterate that the empirical community has clearly done an impressive job at producing widespread and remarkable data so far in Greenland, to reconstruct past glacial history.  

RC3:

Li 2161: of a more-retreated-than present – reword

Authors’ response: 

The sentence was reworded to: “The findings of former ice-sheet margins likely located behind present-day ones during and following the Holocene Thermal Maximum (Briner et al., 2014),”

RC3:

2145 4.2.1 Ice-sheet-wide response  -  This section is dealt with very briefly and lacks a critical appraisal of ice/atmos/ocean interactions and feedbacks.  The linear relationships between air and sea surface temp need exploring in more detail, but the complete lack of recognition of the role of sub-surface ocean temps in driving ice off the continental shelves really undermines this discussion. Atlantic Water ingress and the behaviour of the East and West Greenland Currents during deglaciation and the onset of Holocene was pivotal in the response of the marine terminating margins to ocean heat flux. This point needs to be addressed You touch on this lines 2245 – 2255 but it’s very brief.

Authors’ response: 

This section was meant to be short and rather simple as it focuses exclusively on the ice-sheet-wide response and thus was not meant to address regional ocean forcings and their different magnitudes based on the roles of different Greenland current and water masses in driving ice retreat from specific sectors or continental shelves etc. Such detail would rather suit the next discussion section more, see response to comments below. However, to add more details concerning the essential role of ocean warming and subsurface temperature increases in driving mass loss and ice retreat during deglaciation and the early Holocene, we have added a few sentences to this paragraph:

“….These correlations suggest climate and ocean forcing were the dominating agents of former ice-sheet extent change. For instance, the overall acceleration of outlet-glacier retreat that we observe between ~11.5 and ~9.5 kyr BP is coeval with the potent atmospheric warming that characterised the Younger-Dryas-to-early-Holocene transition in the North Atlantic and Greenland regions (Grootes et al., 1993; Fig. 20). This faster retreat also coincides with significant ocean warming, with marine records indicating increasing sea surface temperatures offshore most Greenland regions from ~16 kyr BP, and reaching maximum warmth at ~10-8 kyr BP (e.g. Williams, 1993; Jennings et al., 2006; 2017; Osman et al., 2021: Fig. 20). More particularly, offshore SE, SW, and CW Greenland regions, warmer and saline sub-surface Atlantic water ingress from the Irminger current (Fig. 1) is thought to have caused rapid mass loss and initial retreat of grounded ice margins from the outer continental shelves (Knutz et al., 2011; Ó Cofaigh et al., 2013; Jennings et al., 2017). Similar forcing is thought to have occurred in NE and CE Greenland, with warmer saline Atlantic water advected southward from Fram Strait, along the Greenland coast, and across continental shelves to marine terminating margins, with the establishment of the West Spitsbergen, Return Atlantic and East Greenland currents during deglaciation (Hopkins, 1991; Hebbeln et al., 1994). Furthermore, …”

RC3:

Li 2172: The section on inertia doesn’t really say or prove anything. On a Holocene timescale how does the mass balance regime of the GrIS change with respect to long term inertia and ice sheet response times?

Authors’ response: 

With this paragraph, we are simply using our reconstruction and ice-sheet wide areal extent analysis to make the observation that minimum Greenland Ice Sheet extent was reached centuries to millennia after peak ocean and atmospheric warmth. This notion of a centennial to millennial-scale lag in ice sheet response to the cessation of warming and onset of cooling during the Holocene is indeed not a new idea, as we mention in the paragraph. We are thus not trying to provide a proof for this concept, rather simply confirming this observation with a new, and more accurate quantitative assessment of Greenland Ice Sheet areal extent change than was produced before. This inertia is however an important behaviour to mention and we don’t necessarily feel like this paragraph is “not saying anything”. This inertia potentially has implications for future GrIS evolution and response to Anthropogenic climate change, and is thus important to quantify and understand better in future work. Understanding more in depth the SMB feedbacks and ice-dynamical reasons for this inertia requires the use of a coupled thermomechanical ice sheet model with good paleoclimate forcing and a more specific set of investigations. Our work does not produce the data to answer the question asked by the reviewer with this comment. This is well beyond the scope of this study. On the other hand, the type of work that would address the controls on this inertia would undoubtedly need to conduct model-data comparisons, which we believe will be eased with the PaleoGrIS database. 

However, some text was added to this section in response to the main reviewer comment on this point, see above for more details.  

RC3:

Li 2182: You need to mention Kjaer et al. (2022) in relation to the Neoglacial. The Neoglacial was an ice sheet wide response to N. hemisphere cooling.

Authors’ response: 

We thank the reviewer for this comment and have added this reference to two locations in this paragraph.

RC3:

Li 2258 – 2267: Ice advance across the continental shelves 18 – 15ka due to H1? This completely unsubstantiated and needs to be amended.  Most regions were retreating from the shelf edge in this window. By all means doube-check the literature (e.g. Ó Cofaigh et al., (2013), but is definitely not true for the NE. The GriS may have experienced marginal oscillation during overall retreat (e.g. GZW formation) but not climate driven advances.

Authors’ response: 

We agree with the reviewer, this sentence is misleading and needs changing, as the timing of advance and retreat was in fact very different in different regions. The evidence in central West Greenland indicates that retreat of outlet glaciers did not start until after 15 kyr BP (Ó Cofaigh et al., 2013; Jennings et al., 2017): however, such retreat likely started earlier in NE Greenland. Moreover, little evidence exists to determine the timing of pre- maximum extent advances. We have thus changed the sentence to:

“When the Greenland Ice Sheet reached its maximum extent (between 24 and 16 kyr BP; Funder et al., 2011), and when local atmospheric- and more importantly sea surface- temperatures were lowest (Grootes et al., 1993; Osman et al., 2021), the ice sheet’s margins had advanced across continental shelves (e.g. CW Greenland: Ó Cofaigh et al., 2013).”

RC3:

Li 2280: Warmings – reword to warm periods.

Authors’ response: 

The sentence was modified accordingly.

RC3:

Li 2285: Such a response would be an example of geographically-induced ice sheet hysteresis? This needs a fuller explanation if you want to push this idea. Topographic and geological influences come into this equation, plus ice stream v interstream areas. You mention ice divides and ice stream switch on and off. All relevant, but all fairly conjectural. The Uummannaq Ice stream ‘system’ is a particularly good example of the complexity of a hysteresis across one catchment.

Authors’ response: 

We agree with the reviewer that the mention of the term hysteresis, which arguably requires more justification here, is not necessary. We have removed this sentence from the paragraph and from paragraph 4 of section “Influence of heterogeneous onshore bed topography”.

 

We also fully agree that all concepts mentioned in this part of the discussion are conjectural and hypothetical. Hence the sentence:  “ We discuss and hypothesise some of these mechanisms in the following sections.” In section 4.2.2. One could argue that is sometimes one of the main point of a discussion, and we believe it still makes interesting arguments that are perhaps worth investigating further in future work. Indeed, whilst many studies focus on the behaviour of individual outlet glaciers, where the roles of ocean, SMB forcings are predominant, fewer studies attempt to analyse the former behaviour of the Greenland Ice sheet at the ice sheet scale, and thus hypothesise what might be controlling heterogeneities in margin response to deglaciation at the scale of entire ice sheet sectors.

RC3:

Li 2297: The Holocene minima in W Greenland was largely a response to surface mass balance change and a low gradient ice surface. Perceived wisdom has not related this to the ice sheet being below sea-level. Particularly the interstream regions.

Authors’ response: 

In order to stress that lower topographies may not necessarily be the strongest control here: but rather may have contributed: we have changed the sentence to: “This may have contributed to the CW Greenland Ice Sheet region experiencing more retreat behind contemporary margins than other regions in response to the Holocene Thermal Maximum (Simpson et al., 2009; Larsen et al., 2015), although this remains a hypothesis.”. 

We do not, however, believe that there is currently enough (if any?) data-based evidence to refute this hypothesis. We invite the reviewer to direct us to such evidence in case we are unaware of it. We believe it is an interesting observation that the W Greenland region, which seems to have experienced more retreat behind present-day margins during the mid-Holocene than other regions, also displays a higher proportion of low subglacial topographies, shown to be an important factor for grounding-line instability and the quick demise of ice sheet margins following atmospheric and oceanic warming, in the context of other ice sheets (e.g. West Antarctica at present).  

RC3:

Li 2330: Moreover, current empirical reconstructions of relative sea level change in Greenland are too uncertain to establish an accurate map of paleo sea levels at the ice sheet scale. You need to rethink this. There are several areas of the GriS that have superb sea-level records. The W and SW for example. There is an overview paper by Long et al. (2011) that can shed some light.

Authors’ response: 

We thank the reviewer for this comment, and agree that this sentence is misleading as such. To better articulate what was meant, i.e. that knowing where the past sea level was through time, at the valley scale, to assess the precise timing of transition from marine to terrestrial ice-sheet margins throughout Greenland, remains a challenge, we have reworded to:

“Moreover, whilst numerous regions feature robust paleo sea-level records (e.g. the Disko Bugt area; Long et al., 2011), it remains a challenge to establish an accurate, time-dependent, and Greenland-wide map of paleo sea levels at the valley scale.”

RC3:

Li 2355: Ice stream switching/re-arrangements. The new work by Frenke et al. in NE Greenland has raised the possibility of recent ice stream migration/shutdown, but I wonder how important that is during deglaciation when ice thins and becomes trapped within over-deepened topography. West Greenland in particular is underlain by a ‘hard’ bed with valley networks established and over-deepened on million yr timescales. Ice flow corridors are, hence, largely controlled by topography – ‘isbrae’ type is ice streams being typical of Greenland. During deglaciation topographic control of ice dynamics is very important (e.g. Lane et al, 2014).   That’s not to say ice stream switching/hosing never happens. There is evidence for this on the continental shelves (particularly during the Miocene)

Authors’ response: 

We thank the reviewer for this comment and interesting point, and agree that near the ice-sheet margins, ice streaming is very likely to be controlled by subglacial topography given the high magnitude of overdeepening in most Greenland Fjords and coastal topographies. However, as we move upflow and away from the ice margin, ice thicknesses increase rather rapidly, and ice flow can become unconstrained by topographies. Towards the onset of deglaciation, when the ice sheet was still predominantly marine-terminating with margins towards the outer continental shelves, present-day coastal topographies would have been covered by thick ice in numerous regions. It has been hypothesized that during the full glacial extent, many contemporary ice streams and outlet glaciers currently nested within the fjord heads of the present coast, may have converged and flowed on to the continental shelf as larger, composite ice streams (Roberts and Long, 2005; Weidick and Bennike, 2007). We believe it is possible to have had different ice stream configurations than present in several cases then, and subsequent reconfigurations when thinning near the margins occurred during deglaciation, as subglacial topographies increasingly constrained ice flow. We believe both our arguments could be saying the same thing here possibly, see reply to main comment above, and see the sentences added to discussion paragraph on the increasingly important role of topography for ice-stream re-adjustments during the 12-8 kyr BP window. Moreover, it is important to re-iterate this part of the discussion is highly hypothetical. To introduce more nuance, we have also modified the bullet point discussion phrase slightly and rephrased it to:

“…may be responsible for:

 -Re-arrangements of certain ice streams and of surface velocity fields during retreat”