The Atlantic Meridional Overturning Circulation (AMOC) is widely treated as highly sensitive to freshwater forcing: hosing experiments weaken it, past abrupt climate events are attributed to meltwater pulses, and CMIP-class (Coupled Model Intercomparison Project) models project a substantial twenty-first-century decline. He & Clark (2022) challenged this picture: proxy records indicate only a muted AMOC response to early-to-middle-Holocene meltwater forcing equivalent to ~50 m of sea-level rise, and a transient Holocene simulation that included this muted response improved agreement with proxy temperatures of the past 21,000 years, implying the AMOC 'may not be as sensitive to FW fluxes and Arctic freshening as is currently projected'. The open problem is to determine, quantitatively, how sensitive the AMOC actually is to realistic freshwater forcing — and whether current climate models over- or under-estimate that sensitivity — through the linked sub-questions, including coastal meltwater routing to deep-water formation sites, the uncertain magnitude/timing/routing of deglacial freshwater forcing, proxy constraints on deglacial–Holocene AMOC, and the state dependence of sensitivity between deglacial and interglacial climates. The literature through 2026 leaves each strand open: the meltwater paradox persists across the PMIP4 (Palaeoclimate Modelling Intercomparison Project Phase 4) deglacial ensemble, routing/resolution effects are demonstrated only in single-model pairs, the first quantitative Holocene AMOC reconstruction remains model-dependent, and an active counter-thread argues models are too stable rather than too sensitive.

Background

The Atlantic Meridional Overturning Circulation (AMOC) is the Atlantic branch of the global ocean conveyor: warm, salty upper-ocean water flows northward, releases heat to the atmosphere, and returns southward at depth as North Atlantic Deep Water. Because this circulation carries a large share of the poleward heat transport of the climate system, changes in the AMOC have outsized climate impacts, and it is listed as one of the principal climate tipping elements. Deep-water formation — the winter convection by which surface water is made dense enough to sink — is concentrated in the Nordic and Labrador seas, and it is suppressed when the surface becomes too fresh or too light.

Palaeoclimate records show the AMOC has changed abruptly in the past: millennial cold stadials such as Heinrich Stadial 1 (~19–14.6 thousand years before present) and the Younger Dryas (~12.9–11.7 thousand years before present) have long been attributed to meltwater pulses from the decaying Northern Hemisphere ice sheets inhibiting deep-water formation. The standard tool for studying this is the 'hosing experiment': a coupled atmosphere–ocean climate model is forced with an imposed freshwater flux over the North Atlantic. Such experiments can push the AMOC past a threshold beyond which it collapses into a weak state and may not recover even after the freshwater input stops ('hysteresis'), and CMIP6 (Coupled Model Intercomparison Project Phase 6) models project a significant decline of the AMOC over the twenty-first century under greenhouse warming. Direct AMOC observations cover only the last two decades, so the deglacial record is the only long-term testbed for how sensitive the circulation really is to freshwater.

He & Clark (2022) revisited this testbed with a transient simulation of the Holocene in a coupled climate model. The final deglaciation of the Northern Hemisphere ice sheets between ~11.7 and 6 thousand years ago delivered meltwater equivalent to roughly 50 m of global sea-level rise — yet proxy records indicate only small AMOC changes over that interval. In their transient simulation, including this muted AMOC response — prescribing the reconstructed AMOC instead of the reconstructed freshwater forcing, with no freshwater flux imposed throughout the Holocene — improved agreement between simulated and proxy temperatures over the past 21,000 years, leading the authors to argue that the AMOC may be considerably less sensitive to freshwater than end-of-century projections assume. They identified the weak points that make the question unresolved: (i) the magnitude, timing, and especially the spatial routing of deglacial freshwater forcing are poorly constrained; (ii) meltwater enters the ocean along thousands of kilometers of coastline, and how much of it is trapped and mixed on the shelf by narrow boundary currents (which CMIP-class ocean grids of ~1° cannot resolve) before reaching deep-convection sites is unknown — earlier work (Condron & Winsor 2012) had already shown that whether deglacial meltwater reached the North Atlantic via the Gulf of Mexico, the St. Lawrence, or the Arctic changes the climate response qualitatively; (iii) the workhorse AMOC proxy, the sedimentary ratio of the isotopes protactinium-231 to thorium-230 (231Pa/230Th^{231}\mathrm{Pa}/^{230}\mathrm{Th}), is influenced by particle flux and water-mass geometry as well as by overturning strength, so reconstructed AMOC cannot yet serve as a definitive target for model prescribing; (iv) it is unknown whether AMOC sensitivity to freshwater under deglacial boundary conditions differed from interglacial (present-day-like) sensitivity; and (v) a systematic assessment of whether models overestimate freshwater sensitivity — through sensitivity tests of routing, resolution, and the decoupling between overturning strength and water-mass properties — had not been performed. These strands define one coherent objective: calibrating the freshwater response of the AMOC across past and future climate states.

Problem Statement

Determine, quantitatively and with stated confidence, how sensitive the Atlantic Meridional Overturning Circulation (AMOC) is to freshwater forcing, and whether current climate models over- or under-estimate that sensitivity — the question raised by He & Clark (2022) from the muted AMOC response to early-to-middle-Holocene freshwater forcing (equivalent to ~50 m of global sea-level rise). Here, freshwater sensitivity means an explicitly stated quantitative relation: the change in AMOC strength — the Atlantic meridional overturning streamfunction in Sv (1 Sv = 106^6 m3^3 s−1^{-1}) at a stated latitude/depth, or the transport through a standard monitoring section such as 26.5°N — per unit freshwater forcing of stated magnitude, rate, duration, and spatial routing, evaluated for a specified background climate state (deglacial, Holocene, present-day, or twenty-first century) and reported with quantified uncertainty; whether current climate models over- or under-estimate that sensitivity is judged against the freshwater-forced component of the twenty-first-century AMOC weakening projected by CMIP-class coupled models, distinguished from the heat-forced component. The question spans the following linked sub-questions: (1) how meltwater discharged along thousands of kilometers of coastline — deglacial ice-sheet melt in the past, Greenland and Arctic freshwater today — is routed by boundary currents, shelf trapping, and mixing to the regions of North Atlantic deep-water formation, and on what scales it is diluted before reaching them; (2) the magnitude, timing, and spatial routing of deglacial and Holocene freshwater forcing, constrained by sea-level and ice-sheet records, well enough to attribute AMOC changes to freshwater versus other forcings; (3) AMOC strength and variability over the last deglaciation and Holocene from proxy reconstructions whose non-AMOC influences (particle-flux effects on 231Pa/230Th^{231}\mathrm{Pa}/^{230}\mathrm{Th}, hydrographic effects on the oxygen-isotope ratio δ18O\delta^{18}\mathrm{O}, water-mass geometry) are quantified, so that reconstructed AMOC can serve as a target for model prescribing; (4) whether AMOC sensitivity to freshwater differs between deglacial and interglacial background states; and (5) whether, and in which direction, the AMOC weakening projected for the twenty-first century is biased by how models represent freshwater routing, horizontal resolution (boundary currents and eddies), and overturning–property decoupling, together with the evaluation of alternative mechanisms for past AMOC disruptions. These sub-questions serve one objective: resolving the Holocene meltwater–AMOC paradox (a large deglacial meltwater load coexisting with only a muted AMOC response) and thereby calibrating the freshwater sensitivity of AMOC projections.

References

  1. He, F. & Clark, P. U. (2022). Freshwater forcing of the Atlantic Meridional Overturning Circulation revisited. Nature Climate Change 12(5), 449–454. https://doi.org/10.1038/s41558-022-01328-2
  2. Bakker, P. (2022). Ocean sensitivity to freshwater. Nature Climate Change 12(5), 419–420. https://doi.org/10.1038/s41558-022-01324-6
  3. Condron, A. & Winsor, P. (2012). Meltwater routing and the Younger Dryas. Proceedings of the National Academy of Sciences 109(49), 19928–19933. https://doi.org/10.1073/pnas.1207381109
  4. Snoll, B., Ivanović, R., Gregoire, L., Sherriff-Tadano, S., Menviel, L., Obase, T., Abe-Ouchi, A., Bouttes, N., He, C., He, F., Kapsch, M., Mikolajewicz, U., Muglia, J. & Valdes, P. (2024). A multi-model assessment of the early last deglaciation (PMIP4 LDv1): a meltwater perspective. Climate of the Past 20(4), 789–815. https://doi.org/10.5194/cp-20-789-2024
  5. Shan, X., Sun, S., Wu, L. & Spall, M. (2024). Role of the Labrador Current in the Atlantic Meridional Overturning Circulation response to greenhouse warming. Nature Communications 15, 7361. https://doi.org/10.1038/s41467-024-51449-9
  6. Ma, Q., Shi, X., Scholz, P., Sidorenko, D., Lohmann, G. & Ionita, M. (2024). Revisiting climate impacts of an AMOC slowdown: dependence on freshwater locations in the North Atlantic. Science Advances 10(47), eadr3243. https://doi.org/10.1126/sciadv.adr3243
  7. Pöppelmeier, F., Jeltsch-Thömmes, A., Lippold, J., Joos, F. & Stocker, T. F. (2023). Multi-proxy constraints on Atlantic circulation dynamics since the last ice age. Nature Geoscience 16(4), 349–356. https://doi.org/10.1038/s41561-023-01140-3
  8. Gerber, L., Lippold, J., Süfke, F., Valk, O., Testorf, P., Ehnis, M., Tautenhahn, S., Max, L., Chiessi, C. M., Regelous, M., Szidat, S., Friedrich, O. & Pöppelmeier, F. (2025). Low variability of the Atlantic Meridional Overturning Circulation throughout the Holocene. Nature Communications 16, 6748. https://doi.org/10.1038/s41467-025-61793-z
  9. Jiang, Z., Brierley, C. M., Bader, J., Braconnot, P., Erb, M., Hopcroft, P. O., Jiang, D., Jungclaus, J., Khon, V., Lohmann, G., Marti, O., Osman, M. B., Otto-Bliesner, B., Schneider, B., Shi, X., Thornalley, D. J. R., Tian, Z. & Zhang, Q. (2023). No Consistent Simulated Trends in the Atlantic Meridional Overturning Circulation for the Past 6,000 Years. Geophysical Research Letters 50(10), e2023GL103078. https://doi.org/10.1029/2023GL103078
  10. Jackson, L. C., Alastrué de Asenjo, E., Bellomo, K., Danabasoglu, G., Haak, H., Hu, A., Jungclaus, J., Lee, W., Meccia, V. L., Saenko, O., Shao, A. & Swingedouw, D. (2023). Understanding AMOC stability: the North Atlantic Hosing Model Intercomparison Project. Geoscientific Model Development 16(6), 1975–1995. https://doi.org/10.5194/gmd-16-1975-2023
  11. Devilliers, M., Yang, S., Drews, A., Schmith, T. & Olsen, S. M. (2024). Ocean response to a century of observation-based freshwater forcing around Greenland in EC-Earth3. Climate Dynamics 62, 4905–4923. https://doi.org/10.1007/s00382-024-07142-0
  12. Wei, X. & Zhang, R. (2024). Weakening of the AMOC and strengthening of Labrador Sea deep convection in response to external freshwater forcing. Nature Communications 15, 10357. https://doi.org/10.1038/s41467-024-54756-3
  13. van Westen, R. M., Kliphuis, M. & Dijkstra, H. A. (2024). Physics-based early warning signal shows that AMOC is on tipping course. Science Advances 10(6), eadk1189. https://doi.org/10.1126/sciadv.adk1189
  14. Mehling, O., Vanderborght, E. & Dijkstra, H. A. (2026). Critical freshwater forcing for AMOC tipping in climate models – compensation matters. Earth System Dynamics 17(3), 563–579. https://doi.org/10.5194/esd-17-563-2026
  15. Boot, A. A. & Dijkstra, H. A. (2025). Physics of AMOC multistable regime shifts due to freshwater biases in an EMIC. Earth System Dynamics 16(4), 1221–1235. https://doi.org/10.5194/esd-16-1221-2025
  16. Lu, W., Oppo, D. W., Liu, Z., Zhu, C., Condron, A., Lynch-Stieglitz, J., Guo, W., Hess, A. V. & Wang, S. (2025). Warmer shallow Atlantic during deglaciation and early Holocene due to weaker overturning circulation. Nature Geoscience 18(8), 787–792. https://doi.org/10.1038/s41561-025-01751-y
  17. Wen, Q., Zhu, C., Chen, D., Liu, M., Ning, L., Yan, M., Liu, J. & Liu, Z. (2023). Separating Direct Heat Flux Forcing and Freshwater Feedback on AMOC Change Under Global Warming. Geophysical Research Letters 50(46), e2023GL105478. https://doi.org/10.1029/2023GL105478