10.20.26. Near final draft.

Three things this means for your daily life

Deep-ocean temperatures are the temperatures of seawater far beneath the surface.

Here, “deep” means below 2,000 meters, or about 1.2 miles; “abyssal” means roughly 4,000–6,000 meters. Warming there stores energy, expands seawater and alters the circulation that carries heat, oxygen and carbon. [2][3]

• Your coast has a long memory. Ocean warming contributes to rising seas and can influence ice loss long after emissions decline. For homes and infrastructure, think beyond the next weather season. See the consequences.
• Your food and community can be affected inland. Changes in ocean circulation and marine ecosystems can add pressure to rainfall patterns, fisheries and supply chains. See the risk inventory.
• Preparation and emissions cuts still matter. Some change is already under way; the amount of additional damage remains strongly influenced by human choices. Start with practical action.

Five key facts—and a mystery worth solving

This article explains how a largely invisible part of Earth’s climate system can shape the next 40 years and leave consequences far beyond them.

• The whole ocean absorbs about 90% of the excess heat associated with planetary warming. Most recent uptake is above 2,000 meters; the deep ocean is a smaller but persistent contributor. [1][7]
• A tiny temperature change can represent an enormous amount of energy. The explanation is the ocean’s immense volume, not an unusually dramatic thermometer reading. See the calculation.
• The best historical records differ in what they measure. Modern instruments, historical reconstructions and ancient fossil chemistry answer different questions. Follow the records.
• The mystery: how can a global deep-water warming of only hundredths of a degree accompany serious ice-sheet risks? The answer lies in where the heat goes, local temperature contrasts and the shape of the seafloor.
• There is useful hope. Reducing emissions limits added heat and future commitments; adaptation reduces exposure. The action section connects these findings with Job One for Humanity’s Plan B.

Executive summary

The deep ocean acts as a vast heat storage area and a slow-moving part of the climate system. Its warming contributes to sea-level rise and interacts with circulation, Antarctic ice, oxygen supply and carbon storage. These processes operate at different depths and on different clocks.

Observations establish recent deep and abyssal warming. A 2026 analysis finds a significant acceleration of heat uptake in the abyssal layer, but does not establish significant acceleration for the entire ocean below 2,000 meters. Reconstructions extend the historical view into the nineteenth century and, using fossils, approximately 108 million years into the past. [4][6][8]

By 2066, my evidence-based expectation is continued warming of the global deep-ocean average, generally on the scale of hundredths of a degree, with much larger changes possible locally. The most important human consequences involve persistent sea-level pressure, regional circulation changes and ecological stress. No single deep-ocean temperature provides a reliable worldwide collapse deadline. The 2066 calculations and conclusions make the assumptions explicit.

1. Why depth and location matter more than one average

“Ocean temperature” can describe the sea surface, the upper 700 meters, the 700–2,000-meter layer, or the colder deep and abyssal layers. A headline about exceptionally warm surface water does not describe water five kilometers down.

There is a second distinction: water named Circumpolar Deep Water can travel up onto Antarctica’s continental shelves. The water contacting an ice shelf may be only a few hundred meters deep. Its name describes its origin and water-mass properties, not a requirement that it always remain below 2,000 meters. [14][16]

Think of an ice shelf as a floating brace that slows the grounded ice behind it. Warm water reaching beneath the brace can thin it and reduce that restraint. Melting floating ice has little direct effect on sea level; the larger danger is faster delivery of land ice into the ocean. [2]

The energy calculation: why a small temperature rise matters

Temperature measures how warm something is. Heat content measures stored thermal energy. Scientists often report ocean heat changes in zettajoules: one zettajoule, or ZJ, equals 10²¹ joules. A watt measures the rate of energy transfer: one joule per second.

A useful bulk approximation is ΔQ ≈ ρ × c_p × V × ΔT. Here ΔQ is added energy, ρ is seawater density, c_p is its specific heat capacity, V is the volume and ΔT is the temperature change. Detailed ocean calculations allow these properties to vary.

For an approximate below-2,000-meter volume of 6.916 × 10¹⁷ cubic meters, density 1,027 kilograms per cubic meter and heat capacity 3,990 joules per kilogram per degree, a warming of 0.01°C stores about 28 ZJ. This is a calculation using rounded bulk properties; it is not a new observation. [33]

The ocean is exceptionally good at absorbing heat. It has yet to acquire an equally effective “undo” button.

The ocean has several clocks

Water carries heat as it circulates; mixing transfers heat between layers. Some deep waters last contacted the surface centuries ago. Meanwhile, waves and pressure adjustments can transmit changes across basins in decades, before the same water parcels have traveled that far. Slow ventilation therefore does not mean that every consequence must wait a thousand years. [6][11]

Back to contents

2. Deep-ocean temperatures since the Industrial Revolution

For orientation, the charts begin the industrial era at about 1750. There is no globally complete thermometer record below 2,000 meters running continuously from 1750 to 2026. Early measurements are sparse; full-depth modern profiling and repeated global ship surveys developed much later. [6][36]

The longest practical industrial-era reconstruction used here

Zanna and colleagues reconstructed ocean heat storage using surface temperature history and an ocean-transport model. The downloaded archive provides annual estimates from 1870 to 2018, including water below 2,000 meters. Figure 1 converts that deep heat-content anomaly into an approximate average temperature equivalent using the bulk calculation above. [4][5]

Figure 1. Industrial-era reconstruction below 2,000 meters. Gray areas mark periods without a plotted global series. The blue band shows the archive’s supplied standard deviation after conversion; it does not include all model uncertainty.

The archive’s below-2,000-meter heat anomaly reaches about 22.5 ZJ in 2018 relative to 1870, equivalent here to approximately 0.008°C. This product is one historical reconstruction, not an agreed total observed industrial-era warming. Its fixed-transport assumptions can miss circulation-related changes. [4][5]

What repeated temperature measurements show

An independent observation-based analysis estimated a below-2,000-meter global trend of 0.00039 ± 0.00017°C per year for 1991–2010. The deep 2,000–4,000-meter layer warmed at 0.00034 ± 0.00019°C per year, and the 4,000–6,000-meter layer at 0.00053 ± 0.00011°C per year. These changes are measurable with carefully calibrated ocean instruments. [3]

Figure 2. Measured temperature trends expressed as cumulative changes from a common starting value. The straight lines summarize fitted trends; they do not invent annual observations.

Figures 1 and 2 should be read separately. They use different methods and baselines, and their recent rates differ. Joining them into a smooth 1750–2026 curve would hide that disagreement.

The important 2026 update

Johnson’s 2026 study fitted changes in warming rates to ship and Deep Argo observations. Its estimated abyssal heat uptake rose from 5.4 ± 4.9 trillion watts around 1988 to 20.2 ± 3.9 trillion watts around 2018. That is about 3.7 times the earlier central estimate. The shallower deep layer showed no statistically significant rate change. [6]

Figure 3. The significant acceleration applies to the abyssal layer. The study does not establish a quadrupling of total ocean warming, deep-ocean temperature, or the entire below-2,000-meter heat uptake rate.

Back to contents

3. Before industry: approximately 108 million years of evidence

Ocean sediments contain the shells of organisms that lived near the seabed. Their chemistry preserves temperature clues. Oxygen isotopes also respond to global ice volume; magnesium-to-calcium and clumped-isotope methods provide additional constraints. Scientists must separate those influences before interpreting an ancient thermometer.

Cramer and colleagues reconstructed deep-ocean temperature trends over approximately 108 million years. Their record resolves broad changes on roughly two-million-year and longer scales. The blue curve below was extracted from the published figure; gaps were retained. It is an approximate digitization, not a download of the numerical supplement. [8]

Figure 4. Ancient deep-water temperatures. Blue: digitized central reconstruction from Cramer et al. Orange: Meckler et al. deep North Atlantic sample temperatures and reported 95% intervals. The records use different methods and should not be stitched into one uniform global series.

The broad picture is a much warmer ancient deep ocean followed by major long-term cooling. The independent North Atlantic reconstruction includes estimates around 15–20°C at some early Paleogene sampling times, with substantial uncertainty. Those estimates illustrate why a single familiar curve should not be treated as the final word. [9][10]

Neither record resolves a thermometer reading for 1750. “Near the present” on a million-year chart can still mean hundreds of thousands of years ago. Ancient geography, ocean gateways, ice cover and carbon dioxide differed greatly; a warm ancient state is evidence of possibility, not a timetable for this century.

On shorter historical scales, Gebbie and Huybers showed how parts of the deep Pacific could still be responding to Little Ice Age cooling while the surface and deep Atlantic warmed. Regional cooling can therefore coexist with net global ocean heat gain. [11]

Back to contents

4. Four terms that must stay distinct

A threshold-crossing date, the time needed for a transition, and the date of its full consequences are three different things. Confusing them can turn a serious risk assessment into an accidental science-fiction screenplay. [17]

Back to contents

5. Inventory of major tipping mechanisms and nonlinear risks

The inventory below covers the principal documented and proposed mechanisms connecting deep or subsurface ocean change to climate risk. There is no universally accepted, complete catalog with measured temperature thresholds and dates for every mechanism. Where the evidence supplies no defensible threshold or timetable, that is stated explicitly.

A. Atlantic overturning and subpolar convection

The Atlantic Meridional Overturning Circulation, or AMOC, transports heat northward and returns colder water at depth. Surface warming and freshwater can reduce the density changes that support overturning. A weaker current can carry less salt northward, allowing further freshening and weakening: the salt-advection feedback.

Timing: weakening during this century is widely expected; collapse onset remains disputed. A 2025 indicator study gave central onset estimates around 2063 under intermediate emissions and 2055 under very high emissions. Those are conditional model-based estimates, not observed countdowns. [19]

A different 34-model analysis retained a weakened AMOC under extreme forcing because Southern Ocean upwelling helped sustain it. A 2026 study adds that the rate of forcing can determine whether the circulation keeps pace. Deep-water temperature alone cannot settle the issue. [18][20]

Labrador–Irminger Sea convection is a related regional risk. Losing that convection is not identical to losing the entire AMOC. For either system, recovery can be slow and dependent on the forcing history.

B. Antarctic Bottom Water formation and abyssal ventilation

Freshening can make Antarctic shelf water less dense, reducing its sinking into the abyss. That alters the distribution of heat and reduces oxygen renewal. Regional observations already show reduced transport, interrupted by a partial recovery associated with saltier shelf conditions. [13]

Timing: a 2023 high-emissions simulation projected a roughly 40% weakening of Antarctic abyssal overturning by 2050. This is a scenario-specific slowdown, not a prediction of a worldwide ocean-current shutdown. No measured global deep-water temperature marks a universally agreed point of no return. [12]

C. Sudden shifts in water reaching Antarctic ice-shelf cavities

A small change in density or current pathways can open a route for warmer water into a previously cold cavity beneath an ice shelf. This changes heat delivery abruptly even if offshore temperature rises smoothly.

Timing: one 2025 Earth-system-model study produced Filchner–Ronne cavity transitions in the 2080s under SSP5-8.5, 2090s under SSP3-7.0 and 2150s under SSP2-4.5. The modeled transition took about ten years and increased basal melting roughly tenfold. The Ross cavity responded more gradually in that same model. These are regional, conditional outcomes. [15]

D. Marine ice-sheet instability and loss of buttressing

Where an ice sheet rests on a bed that deepens inland, retreat can expose thicker ice at the grounding line—the boundary between grounded and floating ice. That can increase ice discharge and promote further retreat. Losing ice-shelf restraint or contact with seafloor pinning points can amplify the process.

Point of no return: some modeled grounding lines continue retreating after ocean melt rates return to earlier values. Local bed shape, lateral restraint and stabilizing ridges matter. The evidence supports basin-specific instability, rather than an identical outcome for every Antarctic glacier. [16]

Timing: initiation and increasing losses can occur this century; very large ice-sheet losses unfold over centuries to millennia. A 2024 modeling study found that reducing melt still limits future discharge even when restoring lost ice is difficult. Avoiding additional damage remains valuable after some damage becomes persistent. [34]

Published temperature ranges: surface warming, not deep-water warming

The following ranges come from a major 2022 synthesis. They refer to global mean surface warming above preindustrial conditions. They are broad research estimates, not ocean thermometer trigger settings. Transition times begin after triggering; they are not years counted from today. [17]

Greenland’s marine outlets interact with subsurface ocean heat, but whole-ice-sheet tipping also involves atmospheric warming and surface-elevation feedback. It should not be classified as a purely deep-ocean tipping point. Newer AMOC work also challenges the usefulness of a fixed surface-temperature threshold. [20]

E. Fracture, calving and proposed marine ice-cliff instability

Ocean thinning can weaken floating ice and make fracture or loss of restraint more consequential. Atmospheric meltwater can widen cracks through hydrofracture. A further hypothesis proposes that tall exposed grounded-ice cliffs could fail fast enough to sustain rapid retreat.

Timing and status: fracture and calving occur now. An additional rapid ice-cliff cascade remains a debated mechanism, not a dated certainty. A 2024 multi-model study found that West Antarctica might not be vulnerable to marine ice-cliff instability during this century under its tested conditions. [26]

F. Methane hydrates, sediments and local stability boundaries

Methane hydrates are ice-like solids containing methane within water molecules. Stability depends on pressure, temperature, salinity and sediment conditions. Warming can cross a local stability boundary and release methane.

Timing: vulnerable shallow margins can respond sooner; heat penetrating large, deeper sediment reservoirs generally takes much longer, commonly centuries or more. Much released methane can dissolve or be consumed by microbes before reaching the atmosphere. Oxidation adds dissolved carbon and consumes oxygen. There is no established date or universal deep-ocean temperature for a global “methane bomb.” [23]

Hydrate loss is also discussed as one possible contributor to sediment weakening and submarine slope failure. The causal importance is site-dependent, and the evidence does not provide a reliable global landslide or tsunami timetable.

G. Oxygen, ecosystem thresholds and carbonate chemistry

Warmer water holds less oxygen, while weaker ventilation can reduce replenishment. Organisms may also require more oxygen as temperatures rise. Crossing a species’ oxygen-and-temperature tolerance can produce nonlinear habitat loss before the average water temperature looks dramatic. A global synthesis found more than 2% ocean oxygen loss since 1960, across all depths. [24]

Acidification adds another threshold. When the aragonite saturation state falls below one, exposed aragonite becomes chemically vulnerable to dissolution. This threatens some cold-water coral frameworks. It is primarily a carbon-chemistry threshold, intensified by interacting stresses; it is not caused by deep-water temperature alone. [25]

Timing: oxygen and habitat changes are already observed, with further risks during 2026–2100 and beyond. Tolerances differ among places and species. There is no scientifically established year when the whole deep ocean becomes lifeless.

H. Thermal expansion and long-lived ocean change

Seawater’s expansion depends nonlinearly on temperature, salinity and pressure. The sea-level response to warming therefore depends on where heat is stored. This is a nonlinear physical response without a single global tipping temperature. [31]

Timing: expansion is happening now. Stored heat and deep-water chemical changes can persist on centennial to millennial timescales. This creates a long-lived commitment even without a sudden threshold crossing. [29]

I. Open-ocean convection and polynya regime changes

A polynya is an opening within sea ice. In some Southern Ocean settings, deep mixing brings warmer water upward, helping keep an opening ice-free. Exposure to cold air then promotes further cooling and mixing. Freshwater layering can suppress this exchange, allowing subsurface heat to accumulate instead. These processes can create episodic, nonlinear changes in regional heat and carbon exchange.

Timing: individual events operate over seasons to years, with longer circulation consequences. Future occurrence depends on winds, salinity, sea ice and ocean structure; there is no accepted global polynya tipping date or proof that all stored heat will suddenly be released. [37]

J. Nonlinear density, mixing and convection

Two technical processes help explain why density cannot always be treated as a simple average. Cabbeling occurs when mixing two water masses produces water denser than the average expected from their starting densities. Thermobaricity describes how pressure changes the effect of temperature on seawater density. Both can alter water-mass transformation and convection.

Timing: these physical processes already operate. They can contribute to abrupt local mixing under suitable conditions, but no measured global deep-ocean temperature or dated worldwide point of no return is established for them. [38]

K. Arctic subsurface heat and Atlantification

Warm Atlantic-origin water can erode the Arctic’s insulating density layers, increase upward heat transfer and contribute to sea-ice loss. Changed sea ice and surface exchange can then influence subsequent mixing. This is an important adjacent subsurface-ocean interaction, rather than warming of the global abyss.

Timing: related changes are already observed in the Eurasian Basin and can evolve over years to decades. Regional regime shifts are possible; no single global abyssal warming threshold or irreversible-transition date applies. [39]

Back to contents

6. Feedback loops: what reinforces change—and what can restrain it

Figure 5. An Antarctic melt feedback and its consequences. Arrows represent causal influence. A plus sign means strengthening the named change. Dashed arrows are region-dependent links; equal line widths do not imply measured strengths.

A useful systems image is a plate of spaghetti: pull one strand and several others move, often after a delay. But the strands do not all pull in the same direction. A 2025 observational study found Southern Ocean freshening restrained natural deep-carbon outgassing. That is why “more stratification” cannot automatically be translated into “more carbon released.” [22]

For the broader Job One systems-risk framework, see the Climageddon Feedback Loop. It connects physical climate interactions with human vulnerabilities; that integrated interpretation should be evaluated separately from each measured ocean trend.

Back to contents

7. Timetables and consequences for people

Figure 6. The main risk windows. A slowdown date, tipping onset, transition duration and full ice loss are different quantities. The chart preserves those distinctions.

Sea-level change is especially practical. Deep warming adds thermal expansion; ocean-driven land-ice loss adds water. Local relative sea level also depends on land motion, currents and the gravitational effects of ice loss. Use local projections rather than treating global mean sea level as a bathtub filling evenly. [28][30]

An AMOC shift could also change regional temperatures and rainfall, exposing agriculture and water systems to altered conditions. Ecological losses can affect fishing communities and food supplies. These risks interact with poverty, governance, development choices and other crises; ocean temperature alone cannot calculate future global deaths or a date for civilization-wide collapse.

Back to contents

8. Start here: reduce exposure and reduce the added heat

1. Choose one local planning question. For a coastal property or community facility, examine sea-level scenarios over the asset’s lifetime using the NASA/IPCC projection tool and current local guidance. Consider flood access, drainage, power and land movement.
2. Work on collective emissions cuts. Support effective energy, transport and industrial policies. Reducing added greenhouse forcing limits future heat uptake and threshold risks.
3. Support the ocean’s observation network. Deep Argo, repeated ship sections, oxygen measurements and Antarctic monitoring help distinguish temporary changes from sustained trends. [36]
4. Make preparation social. Try a “future-coast coffee hour”: bring a map, invite neighbors, identify one shared vulnerability and finish with one achievable improvement. A useful meeting should produce something better than a larger meeting.
5. Use a practical framework. Explore Job One for Humanity’s Plan B for preparation, adaptation, resilience and collective action. Its precautionary planning perspective is distinct from the observational measurements summarized here.

The encouraging physical fact is that persistent ocean adjustment does not make surface-warming mitigation futile. Net-zero CO₂ approximately stabilizes CO₂-driven global surface warming in central model estimates, even while the deep ocean and sea level continue adjusting. [27]

Back to contents

9. My observations and predictions for 2066

My central expectation: in 2066 the global average below 2,000 meters will be warmer than in 2026, while some regions and depth layers will depart from that average. The plausible scale illustrated by historical rates is hundredths of a degree. I would place more confidence in continued global warming at depth than in a precise numerical endpoint or tipping date.

A transparent 40-year calculation

Use the observation-based 1991–2010 below-2,000-meter trend as a reference, r = 0.00039°C per year. Then ΔT = r × 40 = 0.0156°C. Keeping its published rate uncertainty would give about ±0.0068°C from that source alone. This uncertainty is about a past fitted trend; it is not a forecast confidence interval. [3]

Figure 7. Additional below-2,000-meter average warming from 2026 to 2066 under three assumed future mean rates. These sensitivity cases are author-selected calculations, not assessed emissions-scenario projections or probability bounds.

These numbers describe changes, not absolute temperatures. They do not define a “safe” interval, and the three cases do not bracket every possible outcome. The 2026 abyssal acceleration finding is a reason to examine changing rates, not a justification for extrapolating an accelerating curve indefinitely.

The local contrast can be far larger. Naughten and colleagues’ modeled Amundsen Sea warming trends at 200–700 meters ranged from about 0.8 to 1.4°C per century across their future ensemble means. Applying those rates to 40 years gives approximately 0.32–0.56°C. That is a regional rate illustration, not their published 2026–2066 forecast and not a prediction for the abyss. [14]

What I expect this to mean for humanity

• Continued sea-level pressure and rising adaptation demands. The warming of deep water adds to a longer climate commitment; coastal planning must extend beyond short political cycles.
• Uneven risks rather than a uniform global switch. Ocean heat will matter most where currents deliver it to sensitive ice, where ventilation weakens, and where communities lack adaptation capacity.
• A serious circulation risk that deserves monitoring and preparation. I would not attach certainty to an AMOC collapse by 2066. The conflicting evidence justifies reducing the drivers and planning for disruption.
• More ecological stress, with outcomes influenced by other pressures. Fishing, pollution and habitat protection can worsen or reduce damage in a warming, changing ocean.
• A large continuing difference between better and worse choices. Protecting ecosystems, cutting emissions and adapting early can prevent losses even while some physical changes persist.

One useful long-range comparison makes that last point concrete. A 2025 ice-shelf viability study found sharply different outcomes through 2300 under low versus extreme warming. Its headline “60%” concerns a high-end, centuries-long scenario; it is not a forecast that 60% of shelves disappear by 2066. [35]

The most important observation is that the ocean’s response is delayed, but humanity’s choices do not have to be. Acting earlier reduces both the added heat and the number of future decisions made under emergency conditions.

Back to contents

 

The climate change IPCC wildcard factor to never forget

Remember that the IPCC temperature forecasts and other significant, temperature-dependent climate-change consequence forecasts have been significantly underestimated by any combination of the seven powerful reasons detailed on this page. This means that as bad as most of the climate change information or forecasts on this page already are, they could be 20% to 30% worse in specific cases and under specific conditions. 

This is an important climate change, government, media, or IPCC report wild card to carefully consider, especially if you are doing any climate change risk assessment, emergency planning, adaptation planning, resilience building, or potential migration. If you plan for consistently underestimated IPCC timeframes or consequences, and they come 20% to 30% sooner or are 20%-30% more severe than you are prepared for, your risk mitigation plans will not suffice, and you will be in for significant to severe additional financial losses and suffering. (Our organization and staff apply this 20% to 30% underestimation factor when planning for our own and our families' climate-change future.)

Evidence Standards

• Established measurements and mainstream assessments: recent global deep and abyssal warming; large whole-ocean heat uptake; thermal expansion; observed regional ventilation changes; continuing sea-level rise and persistent ocean adjustment.
• Emerging findings: the 2026 abyssal acceleration result; model-specific Antarctic cavity thresholds; rate-induced AMOC tipping; regional feedbacks with opposing signs. Their importance does not make their timing universally settled.
• DMAP analysis and inference: the synthesis of interacting systems, delayed consequences, geographic differences and human exposure. The 2066 judgment and selected rate multipliers are analytical interpretations, not measurements or a validated forecasting model.
• Job One precautionary planning: preparation can reasonably consider severe, uncertain outcomes. Such planning assumptions are kept distinct from mainstream likelihood assessments and observed events.

Back to contents

Uncertainties, Assumptions, Qualifications, and Limitations

• Industrial history: no continuous global observational series below 2,000 meters spans 1750–2026. Figure 1 is a model-based historical reconstruction, converted from energy using an approximate fixed volume, density and heat capacity. Its narrow supplied band excludes several structural uncertainties. The model and independent observed trends disagree in recent decades.
• Ancient history: the approximately 108-million-year curve is digitized from a published central reconstruction; it is not a precise sample-by-sample global average. Its uncertainty envelope was not digitized. The separate North Atlantic record includes reported measurement-based intervals, not all possible systematic uncertainty. Geological records cannot resolve a 1750 or 2026 endpoint.
• Tipping inventory: coverage is broad, not a claim that science has identified every mechanism. Some items are true tipping candidates; others are persistent trends, chemical boundaries or proposed feedbacks. Global surface threshold estimates must never be relabeled deep-water temperature thresholds.
• AMOC: the 2025 indicator dates, 34-model resilience result and 2026 rate-dependence experiments address different diagnostics and forcing designs. No single study establishes the actual collapse year or its probability. A near-zero flow definition and a severely weakened circulation definition also need not give the same “collapse” date.
• Antarctic regimes: the 2080s/2090s/2150s dates belong to one model’s scenario experiments. Other studies use different forcing and cavity geometry. The Naughten simulations held ice geometry fixed; increased melting is not itself a coupled forecast of complete West Antarctic collapse.
• Feedback signs: freshening can enhance melt in one region while inhibiting it elsewhere; stratification can suppress natural CO₂ release. Positive feedback does not automatically mean runaway change. The total biological carbon-pump feedback remains uncertain.
• 2066: the numerical cases extend a historical fitted rate using chosen multipliers. They are not a CMIP ensemble, a formal probability distribution or a complete treatment of acceleration. They intentionally avoid claiming an absolute global deep-ocean temperature in 2066.
• Human consequences: temperature alone cannot yield global mortality, migration totals or a civilization-collapse deadline. Sea level involves upper-ocean warming, ice loss, currents and land motion as well as deep warming.
• Action and persistence: fixing atmospheric CO₂ concentration is different from reaching net-zero emissions. Deep-ocean inertia does not establish large inevitable additional surface warming after net-zero CO₂. Continued emissions create avoidable additional harm.

Back to contents

Frequently asked questions

How can hundredths of a degree matter?

A very large volume stores substantial energy even when its average warms slightly. Global averages also hide the warmer currents that directly contact vulnerable ice. See the energy calculation.

Why does the first chart have blank periods?

Because the selected historical archive begins in 1870 and ends in 2018. Filling the gaps would give a false impression of evidence. Newer studies are shown separately.

Do cooling regions disprove global warming?

No. Circulation redistributes heat and carries old climate signals. Parts of the ocean can cool while its total stored heat increases. See the Little Ice Age example.

Does the deep ocean heat up mainly because of volcanoes?

Geothermal heat enters through the seafloor, but it does not explain the observed modern global ocean heat buildup. The dominant contemporary driver is the planetary energy imbalance associated with human-caused warming, with circulation controlling where the excess energy goes. [1][2]

Have all the important tipping points already passed?

That is not established. Some regions may have committed persistent losses; others remain strongly sensitive to future warming. A local point of no return is not proof that every ocean or ice-sheet system has crossed one.

Will net-zero emissions immediately cool the deep ocean?

No. Deep water and sea level respond slowly. That makes preventing additional heat uptake important; it does not erase the benefits of stabilizing surface warming.

Is a near-term methane explosion from the deep ocean established?

No. Local hydrate instability is a legitimate research concern, but global claims must account for sediment heating, methane dissolution, microbial oxidation and atmospheric transfer. See the hydrate section.

Is +0.016°C by 2066 your exact prediction?

It is the result of a stated reference-rate calculation. My broader judgment is continued deep warming on roughly that small global-average scale, with consequential and potentially much larger regional changes. The calculation is not a probability forecast.

Back to contents

Glossary

Back to contents

AI Prompt Support

• “Explain the difference between an AMOC slowdown, tipping onset and complete circulation collapse. Use primary sources, identify disagreements, and do not convert a model year into a forecast date.”
• “For [location], summarize current local sea-level projections for 2050, 2075 and 2100. State the baseline, emissions assumptions, land-motion effects and the source of each number.”
• “Audit a claim that deep-ocean warming causes an imminent methane burst. Separate seabed release from atmospheric release, and check sediment-heating times and oxidation.”

Back to contents

Linked bibliography and data sources

Research reviewed through October 20, 2026. References link to original studies, author or government repositories, data archives and assessment chapters. The charts’ downloaded data, extracted coordinates and calculation scripts accompany this article.

1. NASA. Ocean Warming: Earth Indicator. Whole-ocean heat uptake; the approximately 90% figure applies to the entire ocean.
2. IPCC (2021). AR6 Working Group I, Chapter 9: Ocean, Cryosphere and Sea Level Change. Assessed ocean circulation, warming, ice-sheet processes and long-term sea-level commitments.
3. Desbruyères et al. (2016). Deep and abyssal ocean warming from 35 years of repeat hydrography. Geophysical Research Letters. Observed layer-average temperature trends; Figure 2 and the 2066 reference-rate calculation.
4. Zanna et al. (2019). Global reconstruction of historical ocean heat storage and transport. PNAS. Historical ocean heat reconstruction and its assumptions.
5. Zanna et al. Updated reconstruction archive, 1870–2018. Zenodo. Downloaded OHC_GF_1870_2018.nc; below-2,000-m heat and provided standard deviations underpin Figure 1.
6. Johnson (2026). Observed Multi-Decadal Acceleration of Globally Averaged Abyssal Ocean Warming. Geophysical Research Letters. DOI 10.1029/2026GL124104. Published observational estimates used in Figure 3.
7. Johnson and Purkey (2024). Refined Estimates of Global Ocean Deep and Abyssal Decadal Warming Trends. Geophysical Research Letters. DOI 10.1029/2024GL111229. Independent estimate of deep and abyssal heat uptake.
8. Cramer et al. (2011). Late Cretaceous–Neogene trends in deep ocean temperature and continental ice volume. Journal of Geophysical Research. DOI 10.1029/2011JC007255. Figure 4c was digitized for the approximately 108-million-year reconstruction.
9. Meckler et al. (2022). Cenozoic evolution of deep ocean temperature from clumped isotope thermometry. Science. An independent deep North Atlantic temperature reconstruction.
10. Meckler et al. (2022). Cenozoic clumped isotope temperature record from the deep North Atlantic. PANGAEA. Downloaded sample means and reported 95% temperature intervals used in Figure 4; CC BY 4.0.
11. Gebbie and Huybers (2019). The Little Ice Age and 20th-century deep Pacific cooling. Science. DOI 10.1126/science.aar8413. Historical cooling signals and the deep ocean’s delayed response.
12. Li et al. (2023). Abyssal ocean overturning slowdown and warming driven by Antarctic meltwater. Nature. Conditional high-emissions simulation of Antarctic overturning and warming to 2050.
13. Gunn et al. (2023). Recent reduced abyssal overturning and ventilation in the Australian Antarctic Basin. Nature Climate Change. Regional observational evidence for circulation and oxygen-supply change, including a partial recovery.
14. Naughten, Holland and De Rydt (2023). Unavoidable future increase in West Antarctic ice-shelf melting over the twenty-first century. Nature Climate Change. Amundsen Sea 200–700-m warming and melt projections; not a coupled prediction of complete ice-sheet collapse.
15. Song et al. (2025). Regional conditions determine thresholds of accelerated Antarctic basal melt in climate projection. Nature Climate Change. Filchner–Ronne modeled cavity transitions: 2080s, 2090s and 2150s under specified scenarios; contrasting Ross response.
16. Hill, Gudmundsson and Chandler (2024). Ocean warming as a trigger for irreversible retreat of the Antarctic ice sheet. Nature Climate Change. Conditional ice-sheet response to imposed warm-cavity regimes; thresholds and hysteresis.
17. Armstrong McKay et al. (2022). Exceeding 1.5°C global warming could trigger multiple climate tipping points. Science. Synthesis of global surface-warming threshold ranges and transition times; these are not deep-water temperatures or calendar forecasts.
18. Baker et al. (2025). Continued Atlantic overturning circulation even under climate extremes. Nature. 34-model analysis of mechanisms sustaining a weakened AMOC; includes an April 2025 correction.
19. van Westen et al. (2025). Physics-Based Indicators for the Onset of an AMOC Collapse Under Climate Change. Journal of Geophysical Research: Oceans. Conditional indicator-based estimates of collapse onset, including mid-century central estimates.
20. van Westen, Börner and Dijkstra (2026). Failure to track a stable AMOC state under rapid climate change. Nature Climate Change. DOI 10.1038/s41558-026-02730-w. Rate-induced tipping in model experiments; published online August 13, 2026.
21. Youngs et al. (2026). Antarctic ice-shelf basal melt shaped by competing feedbacks. Nature Geoscience. DOI 10.1038/s41561-026-01975-6. Both amplifying and restraining regional meltwater feedbacks.
22. Olivier and Haumann (2025). Southern Ocean freshening stalls deep ocean CO₂ release in a changing climate. Nature Climate Change. Observational evidence that stratification can reduce natural carbon outgassing.
23. Ruppel and Kessler (2017). The interaction of climate change and methane hydrates. Reviews of Geophysics. DOI 10.1002/2016RG000534. Hydrate stability, sediment response, oxidation and atmospheric-release constraints.
24. Schmidtko, Stramma and Visbeck (2017). Decline in global oceanic oxygen content during the past five decades. Nature. Ocean oxygen inventory change; applies across ocean depths, not exclusively below 2,000 m.
25. IPCC (2022). AR6 Working Group II, Chapter 3: Oceans and Coastal Ecosystems and their Services. Marine ecological risks, acidification, oxygen loss and habitat effects.
26. Morlighem et al. (2024). The West Antarctic Ice Sheet may not be vulnerable to marine ice cliff instability during the 21st century. Science Advances. DOI 10.1126/sciadv.ado7794. Important limitation on an additional proposed rapid ice-loss mechanism.
27. MacDougall et al. (2020). Is there warming in the pipeline? A multi-model analysis of the Zero Emissions Commitment from CO₂. Biogeosciences. Why continued deep-ocean adjustment does not imply large unavoidable additional global surface warming after net-zero CO₂.
28. NASA. IPCC AR6 Sea Level Projection Tool. Global and location-specific projections through 2150; use local planning scenarios.
29. IPCC (2021). AR6 Working Group I, Summary for Policymakers. Centennial-to-millennial persistence of ocean changes and assessed global sea-level projections.
30. NASA (2020). Sea Level 101, Part Two: All Sea Level Is Local. Regional sea-level differences, including land motion and ice-loss fingerprints.
31. Widlansky et al. (2020). Increase in sea level variability with ocean warming associated with the nonlinear thermal expansion of seawater. Communications Earth & Environment. Temperature-dependent thermal expansion; a nonlinear response without a single global tipping temperature.
32. IPCC (2021). AR6 Working Group I, Chapter 5: Global Carbon and Other Biogeochemical Cycles and Feedbacks. Ocean carbon feedbacks and limits on knowledge of biological-pump changes.
33. Johnson et al. (2007). Recent Bottom Water Warming in the Pacific Ocean. Journal of Climate. Published hypsometry: total ocean volume about 1.33 × 10¹⁸ m³; about 52% below 2,000 m.
34. Alevropoulos-Borrill et al. (2024). Sustained ocean cooling insufficient to reverse sea level rise from Antarctica. Communications Earth & Environment. Reduced melting can limit future losses even when previous losses are difficult to reverse.
35. Burgard et al. (2025). Ocean warming threatens the viability of 60% of Antarctic ice shelves. Nature. Strong scenario dependence through 2300; the headline percentage is not a prediction for 2066.
36. Zilberman et al. (2023). Observing the full ocean volume using Deep Argo floats. Frontiers in Marine Science. Deep Argo capabilities and the large remaining observation gap.
37. Review article (2026). Weddell Sea Polynya formation, cessation and climatic impacts. The Cryosphere. Open-ocean convection, sea-ice openings, heat release and competing regional processes.
38. McDougall (1987). Thermobaricity, cabbeling, and water-mass conversion. Journal of Geophysical Research. Nonlinear seawater density effects that influence mixing and circulation.
39. Polyakov et al. (2017). Greater role for Atlantic inflows on sea-ice loss in the Eurasian Basin of the Arctic Ocean. Science. DOI 10.1126/science.aai8204. Subsurface heat, weakened stratification and Arctic sea-ice change.

Choose one action that protects your community this month, and one that reduces the warming added over its lifetime. Continue with Job One for Humanity’s Plan B.

This document was created by Lawrence Wollersheim, Senior DMAP analyst at the Universe Institute. AI was used in the process of researching, assembling, and verifying the accuracy of this article.