the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Possible provenance of IRD by tracing late Eocene Antarctic iceberg melting using a high-resolution ocean model
Abstract. The Eocene-Oligocene Transition is characterised by the inception of the large-scale Antarctic ice sheet. However, evidence of earlier glaciation during the Eocene has been found, including the presence of ice-rafted debris at Ocean Drilling Program Leg 113 Site 696 on the South Orkney Microcontinent (Carter et al., 2017). This suggests marine-terminating glaciers should have been present in the southern Weddell Sea region during the late Eocene, generating sufficiently large icebergs to South Orkney to survive the high Eocene ocean temperatures. Here, we use Lagrangian iceberg tracing in a high-resolution eddy-resolving ocean model of the late Eocene (Nooteboom et al., 2022) to show that icebergs released from offshore the present-day Filchner Ice Shelf region and Dronning Maud Land could reach the South Orkney Microcontinent during the late Eocene. The high melt rates under the Eocene warm climate require a minimum initial iceberg mass on the order of 100 Mt and an iceberg thickness of several tens of metres to be able to reach the South Orkney Microcontinent. Although these sizes are at the larger end of the present-day range of common iceberg sizes around Antarctica, the minimum estimates are not unfeasible and, hence, the present study confirms previous findings suggesting glaciation and iceberg calving were possible in the late Eocene.
- Preprint
(8909 KB) - Metadata XML
-
Supplement
(9308 KB) - BibTeX
- EndNote
Status: open (until 12 Sep 2024)
-
RC1: 'Comment on egusphere-2024-1596', Anonymous Referee #1, 02 Aug 2024
reply
Review of egusphere 2024-1596 (late Eocene Antarctic IRD)
This paper uses an iceberg trajectory and melting model, embedded within a reasonable late Eocene ocean model to see if it is possible for icebergs from the southern Weddell Sea to have reached an ODP site with Eocene IRD in the South Orkney Islands area. The background and Introduction is well written and argued; it sets the scene well as to what needs to be done to reconcile observations and climate ideas of the time period. The end of the Introduction sets this out well but doesn’t strictly say how this paper is tackling the question. Perhaps a synopsis of the paper’s structure here would assist the reader? This would be particularly useful as there is very extensive Supplementary Information material.
The methodology section is well and carefully explained. There are a lot of choices made in the model simulations, but the authors argue for their choices well, and are up front about the likely errors in melting rates and trajectories. They do return to the simulation assumptions a number of times, appropriately and sensibly. Overall, there look to be sufficient simulations for the conclusions reached in the paper to be robust and not particularly dependent on the model parameter choices.
Specific questions
l c. 45: What was the palaeo-depth of ODP site 696 during the Eocene? And what is its current depth? It looks shallow in several images for the Eocene – how reliable is the reconstruction?
Results section
Do they really need the backward trajectory simulations? Surely one can seed the forward simulations with sufficient size ranges to find the minimum sizes needed to be released from the likely origin sites?
Section 3.1: I think the paper underestimates the likely number of simulations that might provide IRD at the ODP site. There are significant numbers of C4 and C5 iceberg simulations that are shaded grey, reach the South Atlantic, yet don’t “score”. Given the restrictions on things like advection using monthly mean winds, and the uncertainty from a short ocean simulation window I think the authors are being unreasonably restrictive in forcing only simulations going within 1, or at most 2, gridpoints from the ODP site to contribute to a positive score. To my mind, anything going in the narrow flow near the Orkney sub-continent is a feasible contributor. I encourage the authors to think more flexibly about the errors they allow. The authors do turn to this question in the Discussion section, so it is appreciated by them.
Section 3.3: I don’t quite understand the backward trajectory studies. Is the iceberg size (ie C1-C3) what size the iceberg is at the ODP site on back release? Does each backward modelled iceberg “grow” through reverse melting processes along its route? This could be explained better in the methodology and probably it is good to remind the reader in section 3.2 as well. Why is the thickness restricted for backward C1 and C2 icebergs? Surely the most likely scenario is for a thick iceberg from a tidewater glacier, or an ice shelf calving and eventually reducing to a few tens of metres when the IRD drops onto the Orkney sub-continent? I can only see forward icebergs being allowed to behave like this, or have I missed something?
- 510-515: the statement that iceberg thicknesses of a few tens of metres is at the high end of modern estimates for Antarctica must be wrong. Most icebergs today are released from iceshelves hundreds of metres thick. While the authors can argue that Eocene glaciation would probably not have been of this type, even modern tidewater glaciers in the Northern Hemisphere typically produce thicker icebergs than the C1-C2 definitions. The authors need to revisit their wording in places.
Table B1 legend needs to be clearer. What is the datapoint variable being shown?
A test that is missing, although present implicitly within the various runs, is an attempt to use the geological record for ODP696’s IRD to isolate the most likely seed points for the icebergs. This approach might be most revealing and help with the paper’s argument. There is currently a lot of detail and simulations that obscures the main message somewhat.
Citation: https://doi.org/10.5194/egusphere-2024-1596-RC1 -
RC2: 'Comment on egusphere-2024-1596', R. Marsh, 29 Aug 2024
reply
General Comments
The authors present intriguing model evidence for late Eocene Antarctic glaciation, via impressive detective work using forward and backward interactive virtual iceberg trajectories. This is achieved through offline use of the Parcels Lagrangian framework extended for icebergs, with data from a simulation of the past ocean circulation that resolves swift narrow boundary currents and some eddying. Considerable thought is given to iceberg size, melting and provenance, with a specific focus on evidence for contemporaneous IRD at ODP Site 696 near South Orkney in the early (narrower) Drake Passage. The Abstract provides a clear and concise summary of findings. Overall presentation is well-structured throughout, figures are informative, and the manuscript is well written. The Supplementary Materials provide extensive technical details, explaining how icebergs are modelled with Parcels, covering specifics of Parcels coding, key equations and parameters, and experimental design. In summary, the manuscript should be suitable for publication, subject to minor revisions in response to the specific comments outlined below.
Specific Comments
- 9, lines 201-205: Please clarify whether icebergs 'un-melt' (or ‘re-freeze’) along backward trajectories; this is implicit and perhaps obvious given later results and discussion, but it would be helpful to emphasise this here.
- 11, Fig. 4: Its appears that C1-C3 (symbols and lines in the legend) are not shown, but I suspect it is simply the case that these symbols and lines are overplotted by identical equivalents for C5; perhaps clarify this in the caption.
- 15, Sect. 4.1: Can the authors further remark on how swift/different were the Eocene currents, relative to today, with quantitative evidence from the simulations used here? Any differences in speed and coherence of the Antarctic Coastal Current (ACoC) will impact travel times to Site 696. Around most of Antarctic, the present-day ACoC is largely due to strong (sometimes katabatic) easterly winds which drive a strongly barotropic slope current, modified by buoyancy forcing that introduces some baroclinic shear, all subject to seasonality (the ACoC is stronger in winter). In the region of interest (Weddell Sea), the ACoC is further part of a larger and stronger subpolar gyre, driven by large-scale wind and buoyancy forcing. How might all this be different in the late Eocene simulation, of consequence for iceberg trajectories to Site 696?
- Further to this, it appears in Figs. S4.1 and S4.2b, and in the animation, that the proto-ACC is narrow and perhaps weaker than today (associated with a narrower Drake Passage and weaker westerlies prior to full Antarctic glaciation?) Is a weakened ACC relative to a ‘similar’ ACoC (and Weddell gyre) of consequence for icebergs reaching Site 696?
- 18, line 420: Can the authors clarify the statement ‘However, note that the Eocene model was run without ice in Antarctica’; this statement, along with ‘no direct information on sea-ice cover exists in the Eocene model used’ (line 145), indicates that the POP model was implemented without a sea ice scheme (this seems odd, given the Los Alamos provenance of POP, shared with the CICE sea ice model); while sea ice may be absent for much of the year, it may be expected to form during mid-winter at high latitudes around Antarctica, when insolation is near-zero – unless substantial heat is transported southward throughout the year. In summary, some more detail on this aspect/assumption of late Eocene high-latitude climate would be appropriate here and elsewhere (see next point).
- 19, around line 447: Regarding melting, what about the seasonal cycle? As icebergs are released ‘daily at all respective release locations for the five-year model period’ (line 206), and that lifetimes vary from months (C3) to years (C4), one might expect that the chance of a given iceberg reaching Site 696 may be sensitive to time of release (through the year). There are some hints at seasonality, e.g., in S4.3 (right middle panel – buoyant convection melt rate) and Fig. S3.5b (wave erosion), so it seems that this may be a factor (e.g., iceberg mass loss by wave erosion is minimal in late summer, so virtual icebergs released in spring will drift further before much melt ensues). How different was late Eocene seasonality (SST, sea ice extent, winds) compared to the present day? Further to this point and related to frequent statements regarding ice (presumably sea ice), it may be relevant that sea ice traps icebergs in winter, which (if winter sea ice were a factor in the late Eocene) may delay multi-year drift time to Site 696.
- p.19, lines 459-460: Wave erosion is indeed likely to play a dominant role in iceberg mass loss; given the parameterization, Equation (S2.11), and the seasonal cycle (Fig. S3.5b), wind speed is likely to be key to this term. How confident are the authors in the late Eocene winds, irrespective of these being only available from monthly wind stress? How different are late Eocene winds compared to the present day? As explained (lines 165-167), the POP model is ‘forced at the surface by a fixed atmosphere of the fully coupled simulations of the Community Earth System Model, or CESM, version 1.0.5 (Baatsen et al., 2020) under a 2 x preindustrial CO2 forcing to simulate the late Eocene’; given this forcing and the late Eocene geometry, bathymetry and orography, it is likely that the winds are quite different. Can the authors elaborate on this, of relevance to the study?
Citation: https://doi.org/10.5194/egusphere-2024-1596-RC2
Data sets
Data from: Possible provenance of IRD by tracing late Eocene Antarctic iceberg melting using a high-resolution ocean model Mark V. Elbertsen, Erik van Sebille, and Peter K. Bijl https://doi.org/10.5281/zenodo.11146355
Model code and software
MeltingIcebergs Mark V. Elbertsen, Erik van Sebille, and Peter K. Bijl https://doi.org/10.5281/zenodo.12628424
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
202 | 74 | 18 | 294 | 14 | 10 | 8 |
- HTML: 202
- PDF: 74
- XML: 18
- Total: 294
- Supplement: 14
- BibTeX: 10
- EndNote: 8
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1