the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 29 Jun 2026
Southern Ocean dominance and basin asymmetry in centennial-scale bottom-water return
Abstract. Return of abyssal bottom waters toward the upper ocean is a key component of overturning closure because it reconnects the deep ocean with surface-influenced layers and helps regulate the redistribution of heat, carbon, and nutrients. Yet the large-scale organization and timing of this return remain incompletely quantified. Here we combine the 1/12° GLORYS12 global ocean reanalysis with Constituent-oriented Age and Residence time Theory to diagnose bottom-water concentration, ventilation thickness, concentration-weighted diapycnal transport, and water age within a single framework. The results reveal a globally connected but strongly basin-asymmetric return structure. The Southern Ocean forms the dominant return hub, where strong cross-isopycnal exchange and broad outcropping pathways provide the principal route through which abyssal waters reach lighter, upper-ocean-connected density classes within the centennial diagnostic window. Beyond this circumpolar branch, the Atlantic exhibits broad and comparatively efficient upward penetration, the Pacific retains older bottom waters and weaker ascent through much of the basin, and the Indian Ocean displays more spatially focused subtropical and ridge-associated return corridors. These results show that abyssal return is not a single diffuse background upwelling branch, but a basin-dependent structure shaped by Southern Ocean upwelling, topographic mixing, equatorial exchange, and basin-scale overturning geometry. Together, the diagnostics provide a global view of where bottom-water influence rises, how strongly it fills the water column, and how renewal timescales differ among basins.
- Preprint
(2658 KB) - Metadata XML
-
Supplement
(1554 KB) - BibTeX
- EndNote
Status: open (until 24 Aug 2026)
- RC1: 'Comment on egusphere-2026-3511', Anonymous Referee #1, 23 Jul 2026 reply
-
RC2: 'Comment on egusphere-2026-3511', Mian Liu, 04 Aug 2026
reply
General Comments
This study applies the CART framework to the GLORYS12 reanalysis to diagnose global bottom-water return pathways, transformation, and ages. The topic is significant and the diagnostic framework is well suited to the problem. However, the manuscript in its current form contains several conceptual and methodological gaps that need to be addressed. My comments focus on: (1) the robustness of the CART implementation, particularly the treatment of diffusion and the numerical stability of the age computation; (2) the inconsistency between the conceptual schematic and the actual hotspot-based results; (3) the oversimplification of the forcing strategy and its implications for interpreting centennial-scale ages; and (4) the lack of attention to boundary and marginal-sea influences. I recommend major revision.
Specific Comments
1. The CART diffusion tensor and its implementation are insufficiently specified
The governing equations (Eq. 1) include an eddy-diffusivity term KTKT, which the authors state is estimated using the K-profile parameterization (Large et al., 1994). However, the manuscript does not specify how this diffusivity tensor is applied in the CART tracer equations:
- Is KTKT applied as a purely vertical (diapycnal) diffusivity, or does it include horizontal/isopycnal components?
- If the latter, what is the orientation of the diffusion tensor relative to the neutral density surfaces? This matters critically because diffusion along isopycnals (rather than across them) does not contribute to water-mass transformation but can still spread the bottom-water concentration signal, potentially biasing the diagnosed pathway structure.
- Does the CART implementation use the same KPP mixing coefficients as the GLORYS12 native mixing scheme, or is it a separate offline diagnosis? This distinction affects whether the diapycnal exchange diagnosed in the tracer equations is dynamically consistent with the velocity field.
The authors should clarify these implementation details. If the diffusion tensor is not aligned with isopycnal surfaces, the diagnosed upward penetration of bottom-water signals could be artificially enhanced by horizontal/numerical diffusion rather than true diapycnal transformation.
2. Numerical stability of the water age computation in low-concentration regions
The water age is defined as A=T/C. This ratio becomes numerically unstable and physically meaningless where C approaches zero (e.g., in the upper ocean or far from source regions). The authors mention a concentration threshold of 0.05 to "identify regions effectively reached by bottom-water signals," but it is unclear how this threshold is applied to the age field:
- Is the age masked where C<0.05, or is it still computed but flagged?
- If masked, how does the masking affect the zonal-mean age distributions shown in Figure 6, particularly in the upper density classes where C is generally low?
- More fundamentally, the ratio A=T/C is sensitive to small fluctuations in C; did the authors test the sensitivity of the age field to the choice of the lower concentration cutoff (e.g., 0.02 vs. 0.10)? A brief discussion or sensitivity test would be helpful.
3. The conceptual schematic (Figure 1) is at odds with the hotspot-oriented results
Figure 1 presents a schematic of bottom-water upward ventilation that appears to emphasize broad, diffuse diapycnal mixing and gradual upward spreading. However, the results in Figures 2–5 consistently emphasize that abyssal return is concentrated in localized topographic hotspots (Drake Passage, Kerguelen, Campbell/ridge sectors, equatorial Pacific) rather than being spatially homogeneous.
The schematic does not reflect this key finding. It shows relatively uniform isopycnal lifting and gradual lateral spreading, which aligns more with the classical Munk/Stommel view than with the "basin-dependent, hotspot-driven" structure the authors claim to have discovered. I recommend that the authors revise Figure 1 to explicitly illustrate the spatial heterogeneity of diapycnal mixing, perhaps by showing concentrated upwelling zones alongside broader along-isopycnal spreading, to better match their own diagnostic results.
4. The use of monthly climatological forcing and the interpretation of "centennial" ages
The CART simulation was forced with monthly climatological forcing derived from GLORYS12 (Line 112). This means that interannual-to-decadal variability (e.g., Southern Annular Mode, ENSO, decadal changes in AABW formation) is excluded. While this is a common simplification for tracer-age studies, it has important implications for interpreting the age field:
- The ages shown in Figure 6 represent the climatological-mean renewal timescale, not the actual transient age that would result from the real ocean's varying forcing. The authors should state this clearly and avoid over-interpreting the absolute age values (e.g., "70–80 years along the longer pathways") as if they were true calendar ages.
- Does the 100-year integration length suffice for the Pacific abyssal regions, where ages appear to exceed 80–100 years in the northern interior? The age field near the end of the integration may still reflect the initial conditions rather than a true steady state. The authors mention a quasi-equilibrium criterion (∂C/∂t<10−3), but for regions with very slow renewal (e.g., North Pacific abyss), even this criterion may not guarantee full equilibration within 100 years. A more transparent discussion of this limitation is warranted.
5. The role of marginal seas and overflows is neglected
The study focuses on the three major ocean basins, but bottom-water influence is also mediated by marginal seas (e.g., the Weddell Sea, Ross Sea, Nordic Seas) and by overflow pathways (e.g., through the Drake Passage, Scotia Sea, and various ridge gaps). The GLORYS12 reanalysis resolves these features, yet the paper does not discuss their contribution to the diagnosed return structure. For instance:
- Does the ventilation thickness in the Weddell Sea sector (Figure 3) reflect local bottom-water formation/recirculation, or cross-isopycnal export?
- Are the overflow pathways through the Romanche Fracture Zone or the Vema Channel in the Atlantic resolved in the concentration fields, and do they contribute to the Atlantic's "efficient upward penetration"?
Adding a brief discussion of these boundary and overflow influences would enrich the basin-comparative analysis and demonstrate that the authors have considered the full spectrum of return routes.
6. The classification thresholds in the regime analysis (Figure 5c–d) appear arbitrary
The authors define four dynamical-tracer regimes using a concentration threshold of C=0.25 and a diapycnal velocity threshold of ∣ωd∣=10−6 m /s. However, no justification is provided for these specific values:
- Why 0.25 rather than 0.2 or 0.3? Does this threshold reflect a physical separation (e.g., a distinct mode in the joint distribution of C and ωd), or is it chosen for convenience?
- The velocity threshold of 10−6 m/s (~3 cm/day) is relatively high for abyssal diapycnal motions. What fraction of grid points is excluded by this threshold? If the authors have tested alternative thresholds, they should report the sensitivity. If not, they should at least acknowledge the arbitrary nature of the classification and interpret the percentages (e.g., Region 1 occupies 44.8% of the area) with appropriate caution.
7. Minor inconsistencies in terminology and presentation
- Line 191 and Section 3.1: The authors refer to "30-year mean" diapycnal velocity in Figures 2b–e, but the CART simulation was integrated for 100 years under monthly climatological forcing. If the forcing is climatological (i.e., repeating annual cycle), what does a "30-year mean" refer to? Does it mean years 30–60 of the integration, or is it a time average over the last 30 years of the 100-year run? This needs clarification for all time-averaged fields (Figures 4, 5, and 6).
- Line 240: The phrase "initial stages (years 1 and 10)" is used to describe ventilation thickness evolution. Since the forcing is climatological, "years" here refer to integration steps rather than calendar years. The authors should explicitly state this to avoid confusion with actual temporal evolution in the real ocean.
- Line 415: The statement "the Pacific acts as the principal reservoir of old bottom waters" is supported by the age field, but the concentration field (Figure 4c) shows that bottom-water influence in the Pacific is largely confined to the deepest density classes. This does not necessarily mean the Pacific is a reservoir—it could simply reflect weak diapycnal mixing there. The authors might consider a more nuanced wording that distinguishes "storage" from "lack of transformation."
Recommendation
The manuscript presents a valuable global diagnostic exercise, but the issues outlined above—particularly the lack of clarity in the CART implementation, the mismatch between the conceptual model and the results, and the over-interpretation of climatologically forced age fields—need to be addressed. I recommend major revision. I encourage the authors to provide a more transparent description of their numerical methods and to temper their claims about the absolute timescales in light of the forcing simplifications.
Citation: https://doi.org/10.5194/egusphere-2026-3511-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 49 | 25 | 12 | 86 | 11 | 7 | 13 |
- HTML: 49
- PDF: 25
- XML: 12
- Total: 86
- Supplement: 11
- BibTeX: 7
- EndNote: 13
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
General Comments:
The subject of this manuscript is essentially the Antarctic cell of the Meridional Overturning Circulation (MOC) which studied on the basis of Glorys12 reanalysis using Constituent-oriented Age and Residence time Theory. The aim of the work is to establish the asymmetry of the return flow coming from the Pacific, Indian and Atlantic oceans within this cell to the Southern Ocean. The idea behind the study seems quite reasonable. However, it lacks quantitative results that would allow us to verify both the quality of the calculations performed and the adequacy of the tools and data used. The tools used in the task are not fully described. The provided illustrations contain obvious errors. The manuscript requires major revision.
Specific Comments:
1. The authors attempt to confirm that the primary upward transport within this cell, from the lower to the upper branch, is concentrated directly in the Southern Ocean. In this situation, the authors can provide estimates of vertical cross-isopycnal transport specifically for the Southern Ocean, defining its northern boundary, for example, at 40°S, as well as for each of the remaining parts of the three oceans separately. The resulting estimates can be compared with existing estimates of the Antarctic cell transport. Such a comparison can serve as a criterion for the quality of the calculations performed and the adequacy of the tools used to solve the problem, estimates for which are lacking in the current version of the manuscript. In this regard, it would be appropriate to include in the Introduction the intensities of both the Antarctic MOC cell and the Atlantic MOC cell, associated with deep water lowering in the North Atlantic. Here are a few references:
Atlantic cell (14-18 Sv):
1. Inverse box model of the World Ocean (A. Ganachaud // J. Geophys. Res., 2003, 108(C7), 3213. DOI: 0.1029/2002JC001565)
2. The transport balance across the Subantarctic Front in the Southern Ocean (!) between three components: the Ekman transport in the surface layer of the ocean, the eddy transport across the front, and also the transverse transport to the front by geostrophic currents below the crests of the ridges crossing the Antarctic Circumpolar Current (M.N. Koshlyakov, R.Yu. Tarakanov // Oceanology, 2011, 51(5), 721. DOI: 10.1134/S0001437011050110).
3. Direct mathematical modeling of ocean water circulation (R. Marsh, A.J.G. Nurser, A.P. Megann, A.L. New // J. Phys. Oceanogr., 2000, 30(5), 1013. DOI: 10.1175/1520-0485(2000)030<1013:WMTITS>2.0.CO;2).
4. Generalization of hydrographic measurement data (W.J. Schmitz, Jr. // On the World Ocean Circulation: Volume II, The Pacific and Indian Oceans. A Global Update, Woods Hole Oceanog. Inst. Tech. Rept., WHOI–96–08, 1996, 237 pp).
Antarctic cell (transport 15-20 Sv based on the balance of various tracers, i.e. freons, radiocarbon, quasi-conservative phosphate complex, etc., directly in the Southern Ocean, which gives the most reliable estimates):
1. W.S. Broecker, S.L. Peacock, S. Walker, R. Weiss, E. Fahrbach, M. Schroeder, U. Mikolajevich, E. Heinze, R. Key, T.-H. Peng, S. Rubin // J. Geophys. Res., 1998, 103(C8), 15833. DOI: 10.1029/98JC00248.
2. S. Peacock, M. Visbeck, W.S. Broecker In: Inverse Methods in Global Biogeochemical Cycles, Geophys. Monogr. Ser., 114, Eds P. Kasibhatla, M. Heimann, P. Rayner et al.,USA, D.C., Washington, AGU Publ., 2000, pp. 185–195. DOI: 10.1029/GM114p0185.
3. A.H. Orsi, W.M. Smethie, J.L. Bullister // J. Geophys. Res., 2002, 107(C8). DOI: 10.1029/2001JC000976. Other methods yield a much wider range of transports for the Antarctic cell. The list of papers providing transport estimates for both cells is much longer. If necessary, more recent studies can be easily found using the list provided.
2. The main problem of this manuscript becomes apparent when examining Figure S3 in the Supplement. In particular, the presence of neutral density inversions raises questions. These inversions may be related to the zonal averaging that these figures correspond to. However, the entire reanalysis dataset may contain a large number of such inversions, which calls into question the appropriateness of using this dataset to address the problem addressed by the authors. The dataset should be checked for the presence of such inversions. The topography of surface 28.1 should be presented in the Supplement, and it should be confirmed that it is the only one. In Figure S3, the presence of a neutral density isopycnal 28.2 up to 40°N in the Atlantic looks strange. In reality, this isopycnal inclines to the bottom when crossing abyssal sills near the equator. In the Indian Ocean, the inclination to the bottom begins already at 10°S, and in the Pacific Ocean, at 15°S. The authors can easily verify this using, for example, the WOD23 database. This discrepancy between Fig. S3 and the observational data indicates either a data processing error on the part of the authors or a discrepancy between the reanalysis data and the actual ocean state. In either case, the authors need to address this problem.
3. Lines 102-103. The system of partial differential equations is incomplete; it lacks initial/boundary conditions. For example, the initial C and Γ distributions are not specified. What 100 years are we talking about? Is this a forecast or a solution under stationary boundary conditions that simulate only seasonal variations?
Technical corrections:
Line 89. Add the abbreviation CART to subheading 2.1.
Captions to Figure 2 and throughout the text and in the Supplementary Appendix: The highlighted framework in the Pacific Ocean is not the Campbell Plateau, but a section of the mid-ocean ridge at its intersection with the Udintsev and Eltanin fracture zones. The Campbell Plateau is located south of New Zealand and is the shelf of its microcontinent.
Line 147. The reference to Marshall et al. 1999 is incorrect. Potential density was used in that work. Please reformulate the link.
Line 184. Hotspots in Figure 2a are located not only in the equatorial Pacific, but also in the Indian and Atlantic Oceans.
Line 191. Since the authors are simulating seasonal variation, the question arises: to what averaging period does Figure 2a refer? For which 30 years out of 100 were the averaging performed to construct Figures 2b-e?
To what averaging period does Figure 4 refer?
In Figure 5a and its caption, the transport cannot be in Sv; it must be expressed per unit area. In Figure 5b, the transport can be expressed in Sv, but then it is necessary to indicate which density step it refers to. Otherwise, the transport must be expressed in transport/density units.
Line 395. «This old-age structure is consistent with the basin's strong stratification and comparatively weak deep overturning, which limit upward renewal once bottom waters leave the Southern Ocean source sectors.» This sentence is unclear, since the abyssal and deep density stratification of the Pacific Ocean north of the ACC band is weakest compared to the Indian and Atlantic Oceans. And weakening stratification should facilitate overturning.
Line 452. What is "first-order asymmetry"? Restate or expand on the concept.