Water sinking from the surface keeps the North Pacific thermocline relatively young and oxygen-rich, while the tropical Pacific contains much older, oxygen-poor water. As the ocean warms, that relationship is projected to reverse in a striking way: the younger subtropical waters become older and lose oxygen, while the older tropical waters become younger and experience relatively little oxygen change.
The contrast has appeared consistently in climate-model projections, but the physical reason for it has been less clear. A study in AGU Advances traces the pattern to changes in the ocean’s vertical circulation as warming strengthens stratification, combining the effects of warming within the Pacific with a remote response to warming in the Southern Ocean.
The researchers describe the result as a “young get older, but old get younger” response.
The distinction is based on water-mass age, a measure of how long water has been out of contact with the atmosphere. Younger water has more recently been exposed to the surface, where it can acquire oxygen. Older water has spent more time isolated from the atmosphere, allowing oxygen to be consumed through respiration.
The North Pacific subtropical thermocline, roughly the 200-to-1,000-meter layer beneath the surface, is normally supplied with relatively young water that sinks in the subtropics and circulates through the thermocline. In the tropical Pacific, by contrast, ventilation is weaker and upwelling brings older, oxygen-poor water toward the surface.
The study finds that warming alters both pathways, but in opposite directions.
Climate models show a persistent geographic contrast
The researchers first examined projections from 14 CMIP6 climate models that provide ocean oxygen output, 10 of which also provide ideal-age tracers. They focused on the high-forcing SSP5-8.5 scenario and examined changes through the 21st century.
Across the models, the North Pacific subtropical gyre showed stronger deoxygenation and increasing water age than the eastern tropical Pacific. The models differed in some details of the oxygen response, particularly in the tropics, but every model showed stronger deoxygenation in the North Pacific subtropical gyre than in the eastern tropics.
The pattern is also visible in observations-based oxygen products covering recent decades, although the authors caution that historical observations cannot be directly equated with the forced changes projected for the future. Observed North Pacific oxygen changes are strongly affected by internal variability, including the Pacific Decadal Oscillation, while oxygen observations remain comparatively sparse.
The researchers therefore used the models to ask a more specific question: what physical circulation changes could produce the projected geographic pattern?
A warmer Pacific compresses the subtropical circulation
To isolate the mechanism, the researchers used an intermediate-complexity version of the Massachusetts Institute of Technology General Circulation Model, or MITgcm, coupled to a simplified biogeochemistry model.
The ocean model has a horizontal resolution of about 2.8 degrees and 23 nonuniform vertical levels. It includes an ideal-age tracer alongside dissolved oxygen and biological processes. The model was spun up for 4,000 years before the warming experiments.
One experiment gradually increased sea-surface temperature by 3.5°C across the ocean over 100 years. The resulting pattern resembled the broad CMIP6 response: oxygen declined and water age increased in the North Pacific subtropical thermocline, while the tropical Pacific became younger with comparatively weak oxygen changes.
The researchers then used a set of idealized experiments in which sea-surface temperature was suddenly increased by 4°C either globally or within particular regions. These experiments allowed them to separate local Pacific effects from the influence of warming elsewhere.
The clearest local response came from warming the Pacific itself.
The upper ocean becomes more strongly stratified as its surface warms. Stronger stratification makes it harder for vertical motions to carry water downward. In the subtropical gyres, the circulation therefore becomes compressed into a shallower layer.
That does not mean the entire subtropical circulation stops.
Surface transport associated with the wind-driven circulation continues. But because the wind stress is held fixed in these experiments, the depth structure of the circulation changes rather than its total horizontal transport changing dramatically. The ventilation pathways become shallower.
The consequence is particularly pronounced near the bottom of the previously ventilated North Pacific thermocline, between about 500 and 1,000 meters and around 30 degrees north. Water that once received a supply of young, oxygen-rich surface water is no longer reached as effectively by the shoaling ventilation pathways.
There, the model produces a pronounced increase in water age and loss of oxygen.
The oxygen response contains another component. Warming reduces the amount of oxygen that seawater can hold, producing a broad oxygen decline near the surface in regions that are warmed. But when the researchers remove this solubility effect using apparent oxygen utilization, the resulting pattern closely follows the changes in water age. That indicates that circulation changes are central to the non-thermal oxygen response in the model.
The tropical Pacific responds in the opposite direction
The tropical Pacific starts from almost the opposite physical condition.
Its thermocline is relatively old and oxygen-poor because ventilation is weaker and upwelling brings deep, oxygen-depleted water upward. When the Pacific is warmed, the stronger stratification also slows vertical motion there.
But in this setting, reducing vertical motion means reducing the upwelling of old, low-oxygen water.
That makes the tropical thermocline younger.
In the Pacific-4C experiment, the tropical circulation responds relatively quickly. During the first decade, reduced upwelling produces a substantial decrease in age and an increase in oxygen in the tropical region.
The response does not remain isolated from the subtropics, however. Warming has already made the subtropical thermocline older and lower in oxygen. Those anomalies are gradually transported toward the tropics by lateral circulation, including the Equatorial Undercurrent, and by diffusion along density surfaces between the tropical and subtropical regions.
Those later processes counteract the initial effect of reduced upwelling.
For water age, the influence of the subtropical anomaly eventually overtakes the reduced-upwelling effect after about 17 years. For oxygen, the opposing influence becomes dominant in less than five years. The result is a relatively small net change in the tropical thermocline compared with the much larger age increase and oxygen loss in the subtropical North Pacific.
The oxygen response also reflects biology. Reduced upwelling supplies fewer nutrients to the sunlit surface ocean, suppressing productivity and the export of organic matter that is later remineralized at depth. That reduces oxygen consumption and therefore opposes oxygen loss.
Within the tropical oxygen minimum zone examined by the researchers, however, the effect of reduced respiration was less than half the effect of the change in vertical advection. The authors therefore describe biological changes as significant but not dominant in their experiment.
Southern Ocean warming sends a different signal northward
The North Pacific and tropical responses cannot be explained by Pacific warming alone.
When the researchers warmed only the Southern Ocean, the model produced younger, more oxygenated water throughout much of the North Pacific thermocline. At the same time, the South Pacific south of about 10 degrees south became older.
The mechanism begins in the Southern Ocean, where warming reduces local downwelling associated with water-mass formation. That change alters the ocean circulation beyond the region that was directly warmed.
In the model, a compensating downwelling anomaly develops near New Zealand. It propagates as a Kelvin wave, moving northward along the Australian coast and then eastward along the equator. Near the eastern Pacific, the signal travels poleward along the American coasts before gradually propagating westward as a Rossby wave.
This circulation adjustment reduces upwelling in the tropical and eastern Pacific. Instead of bringing as much old, oxygen-poor water upward, the altered circulation produces younger and more oxygenated water there.
The timing is important. The wave-driven circulation change reaches the eastern tropical Pacific during the first decade, before physical transport of Southern Ocean water can carry the Southern Ocean anomaly that far. The researchers therefore attribute the early tropical response primarily to the circulation wave rather than to the direct arrival of water from the Southern Ocean.
The Southern Ocean experiment also reveals why the North and South Pacific do not respond identically. In the South Pacific subtropical gyre, waters affected by reduced Southern Ocean ventilation are gradually transported northward and eastward. Advective transport is followed by diffusion along density surfaces, progressively eroding the initial age decrease in the South Pacific.
The North Pacific response is instead dominated, to first order, by the weakened upwelling of old, low-oxygen water.
Two warming pathways combine in the global experiment
The globally warmed experiment brings the two mechanisms together.
Local Pacific warming shoals the subtropical gyre circulation and reduces the downward transport of young, oxygenated water into the deeper part of the North Pacific thermocline. That produces increasing age and oxygen loss there.
At the same time, Southern Ocean warming reduces subduction in the high-latitude Southern Ocean and triggers the wave-mediated circulation adjustment that weakens tropical and eastern Pacific upwelling. That makes those waters younger and tends to increase their oxygen concentration.
The authors use this combination to explain why the strongest projected deoxygenation occurs in the relatively well-ventilated subtropical North Pacific, while the poorly ventilated tropical oxygen minimum zones show much weaker oxygen changes.
The mechanism also helps explain why the response is not simply a matter of the entire subtropical circulation becoming weaker. The study finds that the depth-integrated lateral volume transport changes relatively little. Instead, the circulation becomes concentrated closer to the surface.
That vertical redistribution matters because ventilation depends on where the water actually travels. A shallower circulation can continue to transport water horizontally while failing to deliver newly oxygenated surface water to the deeper part of the thermocline.
The researchers connect this behavior to the ocean’s potential-vorticity and Sverdrup dynamics. In their analysis, the vertical velocity changes are tied to the vertical structure of the wind-driven meridional circulation. As stratification increases, the depth at which the circulation operates shifts upward, concentrating the transport into a shallower layer.
The resulting circulation change provides a physical explanation for the strong oxygen loss along the base of the North Pacific ventilated thermocline.
North Atlantic warming has a much smaller Pacific effect
The researchers also warmed the North Atlantic separately to test whether remote warming from another major high-latitude region could produce the same tropical Pacific response.
It did not.
The North Atlantic experiment generated a much weaker opposing influence in the tropical Pacific than Southern Ocean warming. In the model, anomalously old North Atlantic water does not reach the tropical Pacific on the relevant timescales.
The comparison helps distinguish the Southern Ocean mechanism from a generic response to warming anywhere outside the Pacific. The Southern Ocean is dynamically connected to the Pacific circulation in a way that allows its surface buoyancy change to alter tropical Pacific vertical motion rapidly.
The model is designed to isolate dynamics, not reproduce every oxygen detail
The experiments deliberately simplify the climate system.
The MITgcm configuration is relatively coarse and lacks the resolution needed to fully represent some processes important to ocean ventilation, particularly in tropical oxygen minimum zones. It also has no Arctic Ocean and only minimal representations of marginal seas.
The model has biases in its average oxygen distribution. It contains too much oxygen in the shallow North Pacific oxygen minimum zone and too little oxygen in Southern Hemisphere intermediate waters, leaving its oxygen minimum too deep compared with observations. Those biases can affect the relative influence of vertical and lateral transport, especially below the observed oxygen minimum.
The authors nevertheless argue that the central mechanism is robust within the model because the North Pacific response occurs across a strong oxygen gradient between well-ventilated surface waters and the underlying oxygen minimum zone.
The warming experiments also impose a uniform 4°C temperature increase in selected regions. The authors note that this may exaggerate the effect of Southern Ocean warming because projected warming there is weaker than elsewhere as a result of background ocean processes.
The model’s winter mixed layers around New Zealand are also unrealistically deep, making that region particularly sensitive to warming. Its simplified coastal bathymetry favors propagation of the Southern Ocean-generated Kelvin wave deep into the subpolar North Pacific, potentially exaggerating the modeled age decrease and oxygenation there.
The experiments hold wind stress fixed and do not include changes in freshwater flux. The authors note that changes in westerly winds and freshwater input can influence the Southern Pacific and subpolar North Pacific and could help explain aspects of projected oxygen change that are not captured by these warming-only experiments.
For those reasons, the researchers present the MITgcm experiments primarily as a way to identify first-order dynamical drivers rather than as a precise prediction of future oxygen concentrations.
The observed ocean does not yet provide a simple test
The circulation mechanisms identified in the experiments may already operate in the real ocean, but the researchers say it is difficult to isolate their contribution from observations.
Oxygen measurements are relatively sparse, and the record is affected by changes in wind stress, uneven warming and natural variability. In the North Pacific, the Pacific Decadal Oscillation produces substantial variability that can obscure externally forced trends.
Observations over recent decades do show substantial North Pacific thermocline deoxygenation, particularly along the densest seasonally outcropping density surfaces. Tropical waters have also lost oxygen. But the study cautions against treating those observed changes as a direct measurement of the long-term forced response represented by the climate-model experiments.
The authors also note that aerosol deposition could modify oxygen changes in the North Pacific by supplying micronutrients, increasing production and ultimately affecting oxygen consumption at depth. These processes add another source of uncertainty when comparing observed oxygen trends with warming-only experiments.
The study therefore points to continued measurements of upper-ocean oxygen, including observations from Biogeochemical-Argo floats, as important for distinguishing long-term circulation-driven changes from natural variability and other influences.
The central result remains a circulation response to warming: in the model, stronger stratification slows vertical motion. Where downwelling normally keeps water relatively young, reduced downwelling allows that water to become older. Where upwelling normally brings old water toward the surface, reduced upwelling makes that water younger. The combination produces the contrasting North Pacific and tropical Pacific oxygen and age responses seen in the experiments and broadly reproduced in CMIP6 projections.
The study was published in AGU Advances.
