the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 04 Mar 2026
Current status of ocean observation, ensemble reanalysis and CMIP6 models in describing Antarctic Bottom Water
Abstract. Antarctic Bottom Water (AABW), a key component of the global meridional overturning circulation, forms in distinct regions around Antarctic coasts. The properties of AABW from different sources can be distinguished by their temperature and salinity properties. This study proposes a classification scheme to divide AABW into three subtypes based on their properties and geographic regions: Weddell Sea and Prydz Bay-originated AABW (WPBW), Ross Sea-originated AABW (RSBW), and Adélie Land-originated AABW (ALBW). These three subtypes are clearly identified in the World Ocean Atlas 2023 (WOA23) dataset. Taking WOA23 as a benchmark, we then systematically evaluate the ability of other datasets, including ensemble ocean reanalysis and CMIP6 models, to represent the spatial distribution and thermohaline properties of these AABW subtypes. Results indicate that the ensemble reanalysis performs well, albeit with an overestimation of ALBW. Nearly all CMIP6 models fail to capture these AABW subtypes, except after the correction of systematic temperature and salinity biases. Notably, CESM1-CAM5-SE-LR is the only model reproducing the three AABW subtypes without the correction, likely due to that the implemented overflow parameterization fits tuned physical processes. Comparison with its high resolution counterpart indicates that increased model resolution may not necessarily improve AABW simulations. This refined classification framework, along with the evaluation of various datasets, provides a foundation for more detailed investigations into the AABW formation, variability, and trends under climate change.
- Preprint
(10071 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 29 Apr 2026)
- RC1: 'Comment on egusphere-2026-978', Anonymous Referee #1, 09 Apr 2026 reply
-
RC2: 'Comment on egusphere-2026-978', Anonymous Referee #2, 16 Apr 2026
reply
General comments
This manuscript proposes a classification of AABW into regional subtypes based on salinity ranges and geographic sectors, and evaluates how well observational climatologies, reanalysis products, and CMIP6 models reproduce these subtypes. The main contribution, i.e. a systematic comparison of AABW between a wide range of products, is potentially useful to the community in my opinion.
However, I have several concerns regarding the physical basis and novelty of the proposed classification, as well as aspects of the methodology and interpretation. Please see my comments below. I recommend major revisions.
Major comments
1. Novelty and physical basis of the classification
The manuscript presents the classification as a “new framework”. However, the idea that AABW can be separated into regional source types (Weddell, Ross, Adélie/Prydz) based on thermohaline properties is already well established in the literature (e.g. Talley, Pardo et al., and many other studies). The present approach appears to formalise this using fixed salinity ranges and geographic masks, but it is not clear that this constitutes a genuinely new framework.
More importantly, the classification is defined using very narrow salinity ranges applied far downstream of formation regions. It is unclear whether such narrow thresholds can robustly distinguish source contributions in the interior ocean. The decision not to include conservative temperature also weakens the constraints. This raises the question of whether the method is identifying dynamically meaningful water masses, or simply segmenting the field based on arbitrary thresholds. As the previous reviewer noted, sensitivity analysis would help here.
2. Use of endmember properties without a mixing framework (OMP issue)
This is probably my biggest concern. The salinity ranges are taken from Pardo et al. (2012), who use Optimum Multiparameter (OMP) analysis. Therefore the quoted values represent end members, rather than mean properties. OMP explicitly accounts for mixing and derives bulk water mass fractions based on unmixed source waters. In the present manuscript, these endmember values are used directly as classification thresholds without performing any mixing analysis. It is unclear to me whether this is appropriate.
3. Exclusion of temperature
The manuscript explicitly excludes potential temperature from the classification on the basis that it is “too variable”. This is difficult to justify physically, particularly as the manuscript also uses density (and therefore implicitly temperature).
Also - excluding temperature while later discussing temperature patterns extensively (e.g. Figures 1–2 and throughout the Results) leads to an inconsistency between the definition and the interpretation.
4. Bias correction
The bias correction (Section 3.4) subtracts regional mean T/S differences between models and reanalysis. This approach effectively forces the models closer to the reference state and artificially improves agreement. To me, it is not clear that this procedure provides meaningful insight into model skill.
Minor comments
Introduction
- L47: repeat transects remain the primary constraint on AABW properties and underpin climatologies such as WOA.
- L86: exhibits
- L87–90: please provide a reference for the stated claim - it does not obviously follow in my opinion.
Methods
- L155: Pardo et al. (2012) is based on OMP analysis. Are the quoted values intended as endmembers? This should be clarified explicitly.
- L163: please justify the exclusion of temperature more clearly.
- L174–176: the analysis appears to focus only on two depth levels (4000 m and near-bottom). Please justify this choice and discuss its limitations. Is it possible that the properties of AABW remain similar, but the models just form less? This distinction is needed and is entirely missed by focusing on 2 discrete depth levels.
Results
- L211: capitalisation issue.
- Figures 1–2 and throughout: there is extensive discussion of potential temperature despite it not being part of the classification.
- L239–241: The relevance of this discussion to AABW formation is unclear.
- Figures 3–6: These figures are difficult to read due to small size and dense layout. Consider enlarging panels or splitting figures.
- please quantify how well or otherwise each product describes AABW relative to the baseline
- some references appear in inconsistent fonts throughout.
Citation: https://doi.org/10.5194/egusphere-2026-978-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 123 | 59 | 13 | 195 | 27 | 19 |
- HTML: 123
- PDF: 59
- XML: 13
- Total: 195
- BibTeX: 27
- EndNote: 19
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This manuscript divides Antarctic Bottom Water (AABW) into three subtypes, WPBW, RSBW, and ALBW, and evaluates their representation in WOA23, an ensemble reanalysis product, and 16 CMIP6 models. The attempt to treat AABW as distinct subtypes is potentially useful, as it addresses the different characteristics of AABW originating from different source regions. The manuscript also brings together a broad range of datasets, which is a strength. However, I do not think the paper is ready for publication in its present form without major revision. My main concerns are that the proposed subtype framework has not yet been shown to be sufficiently robust, the model evaluation remains largely qualitative, and several of the key figures are overloaded and difficult to interpret.
Major comments
1. The proposed subtype framework needs a clearer demonstration of robustness.
The whole paper is built on the subdivision of AABW into WPBW, RSBW, and ALBW using a neutral density threshold and fixed salinity ranges. Since all later comparisons depend on this framework, I think the manuscript needs at least a basic sensitivity analysis showing whether the main subtype patterns remain stable under modest changes in the thresholds. At present, the classification is introduced and then used throughout the paper, but its robustness is not really demonstrated.
2. The evaluation of the reanalysis and CMIP6 models remains too qualitative.
Much of Section 3 relies on visual inspection of maps and uses fairly strong language such as models performing “relatively well”, “capturing” the AABW subtypes, or “failing to capture” them (e.g., Lines 248–267, 343–370). These judgments may be reasonable in a broad sense, but they are not supported by clear quantitative metrics. The paper would be much stronger if the authors added a more objective comparison, for example through a simple quantitative metric. In the present version, some of the model-ranking statements feel more assertive than the evidence shown in the figures.
3. Figures 3–6 are difficult to read.
The multi-panel layout makes each individual panel very small, and it is hard to assess subtype extent or compare models systematically. This is a real problem because these figures are used to support a substantial fraction of the interpretation in Section 3. I would encourage the authors to simplify the presentation, for example, by moving the full model gallery to the Supplement and retaining only a smaller number of representative cases in the main text, ideally together with a summary figure or table that presents the main model differences more efficiently.
Minor comments
1. Line 229: Please clarify Figure 2. The caption states that the subtype distribution is shown for 1993–1998 based on the reanalysis dataset, whereas Section 2 describes the reanalysis product as spanning 1993–2022 (e.g., Line 125). Please explain why only 1993–1998 was used here rather than the full reanalysis period. By contrast, in Figure 1, the WOA23 time period (1991–2020) appears to be consistent with the description in the data section.
2. Line 205: The terminology here should be WOA23 rather than WOA2023.
3. The manuscript would benefit from another careful round of language editing and formatting cleanup before publication. Examples include:
Line 25: “high resolution”→“high-resolution”
Line 40: “under the background of”→“in the context of”
Line 172: “Here”→“here”
Line 211: “Consistent”→“consistent”
Line 279: “reasonably”→“reasonable”
Line 341: “although its spatial”→“although their spatial”
Line 350: “reasonably well position”→“a reasonably good position”
Line 400: “section”→“sections”
Lines 436 and 438: “captures”→“capture”
Line 437: “WAO23”→“WOA23”