Jungdal-Olesen et al. The influence of glacial landscape evolution on Scandinavian Ice Sheet dynamics and dimensions

Referee comments by David Sugden, (25/10/23)

Novelty.

This is a stimulating paper that focuses on the way ice-sheet transformation of a landscape during a series of glaciations can affect the extent, volume and dynamics of successive ice sheets.  This has implications for the extent, behaviour and timing of ice advances in different sectors of the ice sheet. It also affects interpretations of glacial deposits and our understanding of wider environmental changes.

The paper focuses on the Scandinavian Ice Sheet and sheds light on why, for example, there were icebergs floating in the North Sea during early glaciations but grounded ice during later glaciations. Further, the total ice volume was smaller in earlier ice-sheet glaciations because the ice sheet terminated in deep offshore waters closer to the Norwegian coast. The ice sheets were larger during later glaciations due to the effect of offshore glacial deposition filling in deeper zones.  The authors point out that such topographically induced ice-volume contrasts (~ 10%) could affect inferences about global ice volumes and sea-level based on marine oxygen isotope records. This is a stimulating idea, especially if one considers the possible effect of any similar topographic effects on the larger Laurentide and other ice sheets. The paper also discusses the puzzling excavation of the Norwegian trench off southern Norway and its wider regional implications.

Methodology.

The authors use a higher-order ice-sheet model and run experiments in order to test and develop hypotheses. There is a reference experiment where the model runs through a glacial cycle on the present-day topography. Mass balance relationships take into account the east-west, north-south and altitudinal climatic gradients and are constant for all experiments. Isostatic compensation in relation to changing volumes of ice and sediment is included. The reference model run yields a good representation of the observed extent of the Scandinavian ice sheet during the last glaciation and gives the reader sufficient confidence in the model. The second experiment represents ice sheets of the mid-Quaternary (~ 0.5 Ma) with the present-day offshore bathymetry filled with sediment, ie representing the topography before ice had eroded offshore troughs. The third pre-Quaternary experiment (pre-2.6 Ma) runs the model on a reconstructed pre-existing topography and bathymetry. In this case glacially eroded basins and fjords are filled in.

The parameters used in the model are tabulated and easily accessible.  The limitations of the model are outlined, especially the coarse resolution of the 10 x 10 km grid that cannot represent local  effects such as that of individual fjord systems, and the generalized approach to climate that does not replicate any local variability.

Presentation.

The text is generally clear and accompanied by a suite of effective diagrams illustrating the differences between the three models at various stages of the glacial cycle. The reader is shown the differences in ice thickness, mass balance, deformation velocity and sliding velocity at such stages.  There are also maps of the differences in topography used in the three experiments and ice volume changes through a glacial cycle.

Other points.

Line 53. Paxman not in references.

Lines 74-79. I had to read this paragraph several times to follow it. Perhaps less dynamic words such as: 1) absence of glacial deposits offshore, ii) infill of fjords and glacial valleys onshore, and iii) replacement of a wedge……?

Table 1. Excellent to have this clearly presented.

Fig 5. Caption. You are asking the reader to flip back to Fig 3 to follow the diagram. Why not label the columns and the two rows?

Fig 6. Boost caption a little to explain what it shows? I know it is in the text but the reader needs more help when looking at the diagram.

Line 419. It might be clearer to start a new paragraph when moving on to retreat.

Fig 7. Caption. You ask us to go back to Fig 3 to understand the colour scale. Why not add a colour scale to this diagram?

Line 557/8 Great that you offer access to the code and data.

References. Details of many of the references in my pdf are messed up. Check before publication?

This paper runs interesting and novel experiments, using a numerical ice sheet model in a new way to investigate how topography affects the well-studied ice sheet as the Scandinavian ice sheet. The authors outline the point of the experiment clearly and explain what the study is not (i.e. not giving a new reconstruction), which is especially important for this type of work. The gap in current research knowledge is identified and the need for the paper is clear. I recommend publication for this paper as it is robust, and the findings are justified by the approach. I only found the minor points below to amend.

Minor changes:

Please change the background used in the figures so that the colours used to display the data are not also used within the background.

A table summarising the different experiments may be useful, even if just in the Supplementary Information.

Figure 1: The graticule labels are hard to read, the white labels work better.

Line 74: Quaternary, not Quaternar

Line 117: A_flow, not A

Table 1: This is a very useful table. I would recommend including all the notation used in the paper in this table. It would also be more usable if the rows were in alphabetical order (as much as possible) to find the parameters easier. I would also change the range of values to [min, max], as opposed to [max:min], and label the middle column something such as ‘Description of parameter’.

Line 165/166: Define T_positive

Line 192: Represents, not represent

Line 198: Between geothermal heat flux, not between of geothermal heat flux

Line 198/199: Include the parameter symbols inline, e.g. geothermal heat flux from the bed, q_b, and the heat flux from the temperature gradient in the basal ice, q_c.

Figure 2: The c and d panels are unclear.

Figure 3: The panels g-i might benefit from a non-linear scale to make the differences clearer. The colour gradient for g-i and j-l should be different if they are different outputs, and this particular colour choice is hard to see.

Section 3.1: This section could benefit from being split up into more than one paragraph. Suggest new paragraphs on lines 302 and 311.

Line 332: 10,000, not 10.000

Line 431: Capitalise figure

Figure 7: This figure would benefit from including the colour bar so it is easy to interpret without the previous figure.

Lines 470/471/472/506: Swap the years round, e.g. (477-429 ka)

Line 476: 0.5 Ma, not Ma ka

Line 500: Build-up

In this study, Jungdal-Olesen et al utilise modelling experiments of the Scandinavian ice sheet to investigate how changes in bed morphology can impact the development of the ice sheet and its architecture. Landscape evolution across the Quaternary (<2.6 Ma) is considered, and the authors find a significant impact of the pre-glacial bathymetry on limiting south/westward expansion of the ice sheet. Specifically, the filling of the preglacial North Sea Depression and the building out of the Norwegian Shelf were major factors facilitating this ice-sheet expansion during the Mid Quaternary. Insights into the late-stage development of the Norwegian Channel (potentially <0.5 Ma) are also given, with modelling experiments demonstrating a composite development of the trough related to distinct phases of ice streaming during these most recent glacials. The deepening of the Norwegian Channel during the Late Quaternary is shown to once again slow the pace of ice-sheet expansion, though to a lesser extent compared to the older North Sea Depression.

While the potential impact of landscape evolution on ice sheet dynamics and volumes is a widely studied topic, the authors identify a clear knowledge gap whereby the large-scale impacts utilising realistic landscapes and modelled ice configurations has remained understudied. The modelling of ice systems across the North Sea is a particularly complex challenge, and the experiments here provide useful insight into the relationship between shelf geometry and ice-sheet evolution through the Quaternary.

More specific comments: 

32: The focus of this citation was on chronological data, rather than geomorphology. Stroeven et al 2016 would probably be more appropriate here.

Sec 2.1.1. 

- Is there a eustatic sea-level forcing? Are the relative sea level changes observed in Fig 7, for example, purely isostatically driven? Presumably, any SL forcing is also consistent between experiments. Mention of calving in the model too would be useful as it seems to be an important regulator of ice sheet extent in the North Sea. The model and parameters used are otherwise well described.

3.1.

321: There are additional ice-divide reconstructions over the North Sea that could be used to contrast with here: see Clark et al. 2022 and the BriticeChrono reconstruction.

4.1.

- It's not critical but it could be interesting (and within scope) in this section to discuss differences in the inception of the ice sheet, rather than just the maximum extents. For example, the similarity of your PREQ-onshore and reference volume trajectories seems to indicate that the uplift of plateau areas/changing hypsometry during the Quaternary has not introduced any major ice-elevation feedbacks that may have enhanced ice-sheet growth through time. Or maybe that they are countered by other feedbacks such as enhanced ice drawdown through fjord outlets?

- The volumes of Hughes et al. do not provide a particularly robust validation in my opinion, based on their over-simplified derivation. For example, their total EISC volume is >30% larger than the SLE value stated earlier on line 37. A more effective comparison might be to compare patterns and magnitude of loading with previous studies e.g., Vachon et al. 2022 Fig 7A, or similar.

- Though I understand you aim to isolate the impact of landscape geometry, there are knock-on effects on the grounded ice by not accounting for ice shelves. For example, I imagine buttressing effects resulting from a relatively larger ice shelf across the PREQ North Sea would reduce the volume differences in this sector (e.g., Fig 5C). Gasson et al. 2018 is a palaeo example to compare with, and think it would be a useful caveat to mention E.g., '10% is a potential maximum relative volume reduction during PREQ, though second-order ice dynamics resulting from processes such as x and x could reduce this value.'

- While I agree that landscape evolution is one of many factors complicating the use of a consistent proxy for ice volume, I don't think it's a fair assumption that the proportional differences between the global LR04 record relate directly to proportional volume differences of this one ice sheet. Take the volume evolution of the Barents Sea ice sheet through the Weichselian, which does not track d18O. You do argue this later in the paragraph so feel this LR04 comparison is not particularly useful. I would suggest instead removing this sentence starting on 491, and emphasise this added uncertainty for Quaternary ice-sheet reconstructions e.g., Batchelor 2019; Knies et al 2009, particularly in marine sectors. For example, the insights here are particularly relevant for the Barents Sea domain too which experienced an opposite transition from terrestrial to marine-based dominated.

4.2.

514: Boulton and Hagdorn were in fact explicitly unable to reproduce an 'ice stream funnelling ice along the entire length of the Norwegian Channel', and were highly sceptical of the idea, demonstrating similar time-transgressive zones of streaming as in your experiments.

526: I think there are many other major reasons why you cannot rule out continuous ice streaming in the channel (besides the sliding limitations given), which mainly revolve around how the ice saddle and regional ice divides developed. This is not a critique though - it is a well-known modelling challenge, so a wider context here would be relevant. That the saddle persists through to late-stage deglaciation may be down to more fundamental aspects e.g., the presence of the 'ice wall', or that the lake drainage event was crucial here. Gandy et al. 2020 and their approach using a negative SMB anomaly in the southern North Sea would be useful to contrast with here in this respect. 

 

529: In terms of the main discussion point here, being the formation of the Norwegian Channel, I think it would be useful to touch a bit deeper and be more explicit on the ice dynamics between the experiments. My takeaway is that the streaming patterns needed to erode the trough (shown in MLQ onwards) only appear when a saddle forms. The limiting factor for this based on PREQ seems to be the central North Sea depression, which effectively restricted the margin through calving? What are the reconstructed water depths here for example? Maybe there are other mass balance feedbacks onshore though? Just think it would be useful to be a bit more explicit on what's driving the differences between experiments. It could be illustrated simply in a figure showing calving/melt anomalies from the ref figure, like in Fig 6B.

I appreciate the choice of the 'scientific' colour ramps on the figures. 

The details in Fig2 are not particularly obvious at 100% scale, but the resolution in the pdf was sufficient when zooming in.

Fig 5/7: Suggest to include all the necessary labels/legends in the figure.

There are a fair number of typos/grammatical mistakes that should be checked throughout e.g., the very last line 555: ...*the North Sea. The manuscript otherwise reads very well.

Some citations have unusual formatting of the page numbers.