the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 19 Jul 2024
The Ross Sea and Amundsen Sea Ice-Sea Model (RAISE v1.0): a high-resolution ocean-sea ice-ice shelf coupling model for simulating the Dense Shelf Water and Antarctic Bottom Water in the Ross Sea, Antarctica
Abstract. The Ross Sea in the Southern Ocean is a key region for the formation of the Antarctic Bottom Water (AABW) that supplies the lower limb of the global overturning circulation, and contributes to 20–40 % of the total AABW production. AABW primarily originates from polynyas characterized by strong sea ice production and ocean convection that lead to the formation of Dense Shelf Water (DSW), the precursor of the AABW. The production and characteristics of DSW in the Ross Sea and AABW in the surrounding ocean are significantly affected by ice shelf meltwater transported from the nearby Amundsen Sea. The scarcity of long-term observations in the Ross Sea hinders the understanding of DSW and AABW variability, and numerical models are needed to explore the multi-scale variations of these water masses and the forcing mechanisms. In this work, a coupled high-resolution ocean-sea ice-ice shelf model is developed for the Ross Sea and Amundsen Sea, named RAISE (Ross-Amundsen Sea Ice-Sea Model). Detailed descriptions of the model configurations are provided. This study represents a first attempt to thoroughly evaluate the DSW properties and associated ocean-sea ice-ice shelf coupling processes among modelling studies in the Southern Ocean, using multiple datasets including satellite-based observations and hydrographic measurements from the World Ocean Database, Argo profilers and seal-tag sensors. In particular, the modelled temporal variations of DSW in polynyas and its key export passages are compared with long-term mooring observations, which are not seen in DSW studies before. RAISE demonstrates high skills in simulating the observed sea ice production rates in the Ross Sea polynyas, and the modelled spatial and temporal variability of DSW are significantly and strongly correlated with observations. RAISE can also effectively capture the observed long-term freshening trend of DSW prior to 2014 and the rebounding of DSW salinity after 2014. RAISE shows an overestimate of DSW density in the Ross Sea, which is associated with underestimate of ice shelf melting rates in the Amundsen Sea, missing ice shelf calving processes and subglacial discharge in the model. A sensitivity experiment simulating increased freshwater discharge from these processes can significantly improve the simulation of DSW properties.
- Preprint
(18079 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on gmd-2024-128', Anonymous Referee #1, 28 Aug 2024
General comments:
This study investigated the reproducibility of the sea-ice conditions (sea-ice concentration and production) and oceanic conditions in the Ross Sea using a high-resolution ocean-sea ice-ice shelf model for the Amundsen and Ross Seas, named RAISE v1.0. Additionally, the authors examined the impact of meltwater from the ice shelves in the Amundsen Sea on the water properties in the Ross Sea. Understanding the changes in the coastal water masses around Antarctica is very important due to its significant influence on deep water formation and, subsequently, global thermohaline ocean circulation. However, while reviewing this manuscript, I noticed a significant overlap with the content of the authors’ previous publication in JGR-Oceans (Xei et al. 2024, doi:10.1029/2024JC020919). Although there is a slight difference in the model integration period, it is evident that the model used is the same as that in Xie et al. (2024). Moreover, the sensitivity experiments regarding increased meltwater from the Amundsen Sea ice shelves are very similar to each other. Even the description of the model, while arranged differently, appears to be nearly the same. If my understanding is correct, this could be considered a case of duplicate publication. However, if I have misunderstood the extent of the overlap or the novelty of the current work, I would appreciate a clear rebuttal or clarification from the authors. At the very least, it is necessary to properly cite the previous work and clearly highlight the differences.Specific comments:
Figs. 3g, 10, 11
I believe it is misleading to claim that the model accurately reproduces the observations simply because the correlation coefficient is significant when seasonal variability is included. Seasonal variability has a strong cyclic pattern, which can lead to a high correlation between the model and observations, even if the model does not truly capture the underlying processes. Evaluating the model’s performance without removing the seasonal component can overestimate the model’s skill. For a more accurate assessment, the seasonal signal should be removed before calculating the correlation, or the analysis should separately address seasonal and non-seasonal/interannual variability.Fig. 7
Regarding the spatial correlation as well, there may still be residual effects from the initial conditions. Even if the model shows a good correlation with observations, it does not necessarily mean that the model accurately reproduces the underlying processes. The initial conditions can strongly influence the spatial patterns, leading to high correlations that may not truly reflect the model’s capability to simulate the key dynamics. One could imagine that the spatial correlation between the initial conditions and the observations might yield a similar correlation coefficient. It is important to demonstrate that the higher correlation is due to the high-resolution model resolving fine-scale structures that were not present in the initial conditions, rather than simply reflecting initial condition influence.Fig. 4
The interannual variability of sea ice production in the Ross polynya, which accounts for a large portion of sea-ice production in the model domain, does not reach the 95% significance level. Therefore, I do not believe the model can accurately reproduce the observed interannual variation in sea-ice production.It would be beneficial to include not only a comparison of the water mass properties but also quantitative estimates regarding DSW and AABW formation/production rate in the model.
Citation: https://doi.org/10.5194/gmd-2024-128-RC1 -
AC1: 'Reply on RC1', Zhaoru Zhang, 28 Aug 2024
We thank the reviewer for making efforts in our manuscript review, and proposed many constructive comments. We will respond to these comments and make the corresponding analysis and revisions carefully later on. But we would like to respond to the 1st comment now to clarify the novelty of this study, as we believe this is crucial to prevent any potential confusion regarding the originality of our research. Our arguments are as follows.
First, the model developed in this manuscript is an updated version of the one used in Xie et al. (JGR: Oceans, 2024). The difference is that surface temperature and salinity are nudged to a monthly mean climatology provided by the World Ocean Atlas 2018 in this model, while Xie et al. (2024) did not apply any nudging. The two versions are developed at the same time. We actually compared the simulations from the two models, and found that the version employing nudging performs better in sea surface temperature and sea ice concentration assessed against satellite observations. Additionally, it better captures the interannual variation of DSW compared with CTD data (collected in austral summer). In the manuscript, we did not mention Xie et al. (2024) as it was still under review at the time we submitted this manuscript, and we think it might be inappropriate to cite it without a DOI assigned. We will add such comparisons and emphasize the difference of this model in the revised manuscript. In addition, Xie et al. (2024) is focused on scientific problems relevant to the influence of enhanced ice shelf melting in the Amundsen Sea on the Ross Sea water properties, rather than model development and validations. They provided 4 figures for model validation (in the main text and supplementary materials), and while the 3 figures for validating sea ice production, hydrographic variables along the Ross Sea cross-shelf transect and spatial distributions of sea ice concentration are plotted in the same way as those in this manuscript, as mentioned above, these results are based on different model versions and are not duplicate results.
Second, as for the sensitivity experiments regarding increased meltwater from the Amundsen Sea ice shelves, the scientific motivations and configurations in Xie et al. (2024) and this work are quite different. Xie et al. (2024) is focused on the impacts of accelerated ice shelf basal melting in the future on DSW formation and CDW intrusion in the Ross Sea, and they increased the melting rates based on future projections of ice shelf melting in the Amundsen Sea from CMIP6 scenarios. In this work, we designed ice shelf sensitivity experiments to address the missing and underrepresented ice shelf melting processes in the model. This adjustment aims to mitigate the overestimation of DSW salinity observed in the model compared to mooring observations. We conducted these experiments by artificially increasing the ice shelf melting rates to match the values estimated from satellite data.
We hope misunderstandings of the novelty of this work can be avoided by addressing the points above, and we’ll provided more detailed responses during the revision stage. We thank the reviewer for highlighting these issues, as other reviewers or readers may have similar concerns.
Citation: https://doi.org/10.5194/gmd-2024-128-AC1 - AC2: 'Reply on RC1', Zhaoru Zhang, 02 Nov 2024
-
AC1: 'Reply on RC1', Zhaoru Zhang, 28 Aug 2024
-
RC2: 'Comment on gmd-2024-128', Anonymous Referee #2, 22 Sep 2024
This manuscript presents an Ocean – Sea Ice – Ice Shelf model for Ross Sea and Amundsen Sea (called RAISE). The ocean and sea ice components are primarily based on the ROMS configurations; a static ice shelf is added to the model, which allows inclusion of ice shelf melting. The manuscript presents most technical details for model implementation and validation with available data for key ocean physical processes relevant to production of Dense Shelf Water and Antarctic Bottom Water.
Most of the implementation seems straightforward and reasonable in most parts. The simulation spans 17 years (2003-2019) with a prior 5-year spin-up. Numerical experiments with an inert tracer (dye) are run, which show the transport and dispersion patterns of Circumpolar Deep Water (CDW) and ice shelf water (ISW). Another experiment is also run by increasing melting rates of ice shelf from Amundsen Sea ice shelves (Melt+). The results, as compared with available in situ observations or remote sensing, seem reasonable in general with one significant exception (explained below). The general functioning of the system (e.g. transport, mixing, ice production/melting) also appears to be consistent with our knowledge. The main finding is that ice melting from Amundsen Sea plays a key role in determining salinity (and hence density) of dense shelf water (DSW) in this region. It could be worth expanding this topic a bit by explaining, for example, does this affect DSW properties in southwestern Ross Sea shelf? Or does it affect Antarctic Bottom Water (ABW) properties or production?
Based on one transect, the model has under-predicted the salinity in the top 150 m by 0.1-0.2 psu (model temperature is slightly, ~1oC, higher than observed) (Figure 6). In contrast, modeled subsurface salinity is than observed (by ~0.05 psu). This seems consistent with lower ice concentration than observed (Figure 3). The logic seems to be: Warmer temperature in surface layer leading to less sea ice formation, which in turn leads to insufficient sea brine formation and under-prediction of DSW production. However, this seems contradicting to the sea ice production comparison (Figure 4), which shows model over-predicts the sea ice production. The scatter plot between modeled vs WOD salinity for shelf water also shows a <1 slope, i.e. surface salinity is higher than WOD salinity and vice versa for subsurface water (red dots in Figure 7d)? Perhaps there are spatial mismatch (the transect is on western Ross Sea) or perhaps ice thickness is an issue? In addition, the Melt+ experiment with enhanced ice shelf melting significantly reduces DSW density (salinity), bringing the density much closer to observed. Does this mean now we are getting a smaller slope if we plot a new scatter plot? Regardless, it seems some more clarification is needed to reconcile these.
The manuscript can also benefit from some polishing and clarifying a few details. For example,
- It is unclear, however, how exactly the melting rate under the ice shelf is calculated and applied. It is not clear how Melt+ experiment is being implemented other than that heat/salt transfer coefficients are modulated as done by Nakayama et al. (2020).
- Figure 10 (TNB mooring) & Figure 12 (Ross Island CTD) suggest modeled salinity was also higher than observed on the western Ross shelf and even areas close to Ross Sea Ice Shelf. Does Melt+ improve salinity simulation in those area too?
- Based on dye experiment, it takes 5 years for those dyes released from Amundsen Sea to reach western Ross Sea, suggesting a delay of 5-10 year in the impacts. This may help to explain why salinity improvements at CA1 and CA2 in Melt+ only takes place after mid-2008? Figure 11 did not show salinity changes before 2008, so it is clear if this is truly the case.
- It seems higher model SIP than observed (Figure 4) is contradictory to the lower ice concentration than observed (Figure 3). But this may also mean the model over-estimates the ice thickness. So perhaps some clarifications about how model performs on sea ice thickness will be helpful.
Some editorial comments/suggestions
Page 2, line 31-34, Is this true there were no modeling studies examining this issue?
Page 2, line 35, suggest changing “DSW” -> “DSW properties”?
Page 2, line 36, “which are not seen in DSW studies before”. It is unclear what have not been seen, temporal variations or observations?
Page 3, line 41, “under-estimate” -> “an under-estimate”
Page 5, figure 1 lower panel. It is worth noting the model grid is not conforming to orthogonality in those two southern corners. This maybe intentional and shall not affect model results since they are land points.
Page 6, entire paragraph, so some aspects of the interactions between ice shelf melting and DSW may have been examined.
Page 7, line 117, “finite-volume” should be “finite-difference”
Page, 8, line 133, maybe spell out “EN4” and add either a link or a reference?
Page 8, line 135, “Below 1000 m (the isobath at the shelfbreak)”. Is 500m a better delineator for shelf-break. 1000 m may be more like upper slope.
Page 10, line 153, so does the release only take place once, or is it released continuously? If the latter, for how long?
Page 10-11, section 2.3, how is ice shelf melting simulated? Since this is one of most critical factors in this study, perhaps spell out some details how this is being implemented?
Page 11, line 179, Melt+, increasing ice melt rate by how much?
Page 11, line 187, add WOD behind “World Ocean Database”
Page 12, line 197. Add period behind 2016.
Page 12, line 205, define SIC here
Page 13, line 216, remove “sea ice concentration” now
Page 13, section 4.1, how about sea ice thickness, any data to compare?
Page 14, line 234-240. Not sure about this – so I assume satellite estimates of sea ice production are based on observed (perhaps also estimated) ice thickness and ice concentration? But does this statement suggest that these estimates did not take into accounts the change in ice thickness due to the bottom portion of sea ice?
Page 16, line 262, add “underestimates” to read like “overestimates temperature and underestimates salinity in the surface layer…”. Also modeled subsurface temperature on top of the bank appears to warmer than observed. Overall, the model predicts a deeper surface mixed layer.
Page 19, line 293, suggest change “with temperature” to “where water temperature is above oC”? And after this, maybe start another sentence, sth like “As seen in Fig. 8a …”?
Page 19, line 298, suggest deleting “the role of”
Page 19, line 301-303, this sentence is confusing. Maybe revise this to sth. like, “Compared to the western portion of the RIS, there is more ISW beneath the eastern portion, indicating stronger influence on the hydrography over the Ross Sea shelf.
Page 20, line 312, I assume here dye concentration (and Figure 9) is from bottom layer?
Page 21, line 321-322, do tides play any role in spreading dyes?
Page 22, line 333, “in both middle and bottom” -> “in both layers”
Page 22, line 335, Section 4.1?
Page 22, Figure 10 caption, “… (b) Same as (a) but density”?
Page 23, line 345, again, this should add “properties” so it reads like “… variations of DSW properties…”
Page 25, line 384, suggest changing “falls in the range of” to “is in line with” or “is on the upper end of”
Page 25, line 388, “the satellite estimates” - > “all the satellite estimates”
Page 29, line 455, again it seems higher model SIP is contradictory to the less ice concentration
Citation: https://doi.org/10.5194/gmd-2024-128-RC2 - AC3: 'Reply on RC2', Zhaoru Zhang, 02 Nov 2024
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 470 | 137 | 27 | 634 | 12 | 14 |
- HTML: 470
- PDF: 137
- XML: 27
- Total: 634
- BibTeX: 12
- EndNote: 14
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1