Ocean Heat Stored Deep Below Could Lock In El Niño-Like Conditions for Centuries After Emissions Stop

Reaching net zero carbon emissions is widely imagined as the moment the climate crisis begins to unwind. But new research suggests that for the tropical Pacific—the engine room of global climate variability—stopping emissions may do far less than hoped. A study published in Communications Earth & Environment shows that once an El Niño-like shift in the Pacific’s mean state is established during warming, the sheer thermal inertia of the deep ocean and Southern Ocean can sustain it for centuries, even as atmospheric carbon dioxide slowly declines.

The research team, led by Aditya Sengupta of the University of Melbourne, combined two complementary modelling frameworks. The first was the Zero Emissions Commitment Model Intercomparison Project, or ZECMIP, in which nine Earth system models branch off a transient warming simulation at the point where cumulative emissions reach 1000 petagrams of carbon and then run with emissions set to zero. The second was a set of bespoke, millennial-length simulations with the Australian Earth system model ACCESS-ESM1-5, in which emissions cease at seven different points between 2030 and 2060 and the climate is then allowed to evolve freely for 1000 years. Together, these experiments let the researchers ask a question that standard climate scenarios rarely address: what happens to the tropical Pacific after humanity stops emitting?

The answer is sobering. Under transient warming, nearly all models agree that the zonal temperature gradient of the equatorial Pacific—the difference in sea surface temperature between the warm western Pacific and the cooler eastern Pacific—weakens, producing a mean state that resembles a permanent El Niño. In the ZECMIP simulations, this weakened gradient persists for decades to centuries after emissions cease, with no meaningful recovery. The global mean surface temperature stabilises at roughly 2 degrees Celsius above pre-industrial levels, but the Pacific does not bounce back to its pre-industrial configuration.

The consequences for extreme weather are equally persistent. The researchers used a metric linking the meridional temperature gradient across the eastern equatorial Pacific to local rainfall, a well-established proxy for extreme El Niño events of the kind seen in 1997/98 and 2015/16. In the unforced pre-industrial control runs, only about 9.8 percent of El Niño events in the ZECMIP ensemble were extreme. During transient warming, that share more than doubled to about 21.3 percent—and after emissions stopped, it remained at 20.25 percent. In the longer ACCESS-ESM1-5 runs, the baseline was even lower, at 1.73 percent of El Niño events, yet in the late-cessation B2060 simulation, extreme events still accounted for 26.3 percent of all El Niño events centuries after net zero was achieved. Crucially, the changes scale with warming level: the later emissions cease, the larger and longer-lived the shift.

Not every aspect of El Niño–Southern Oscillation behaviour behaves so coherently. Changes in ENSO amplitude—the typical strength of the oscillation—turn out to be non-monotonic and strongly model-dependent. Seven of the nine ZECMIP models showed amplification during transient warming, and the ensemble median stayed elevated after cessation, but the 30-year rolling time series wobbled considerably, dominated by multi-decadal internal variability. Only when the researchers applied a 100-year averaging window in the ACCESS simulations did a sustained elevation of ENSO amplitude emerge clearly from the noise, particularly in the late-cessation pathway. The forced signal, in other words, is real but only reveals itself on multi-centennial timescales.

The mechanism behind this persistence lies in the uneven way the ocean absorbs and releases heat. After emissions stop, the surface of the Northern Hemisphere stabilises or even cools, but the Southern Ocean and the deep ocean continue to warm for a very long time. This delayed warming shrinks the temperature difference between the hemispheres, which shifts the tropical convergence zones—the ITCZ and the South Pacific Convergence Zone—net equatorward and drives cross-equatorial atmospheric flow. The result is a sustained weakening of the easterly trade winds over the equatorial Pacific, which reinforces the El Niño-like background state.

Beneath the surface, the story continues into the upper ocean’s circulation machinery. During warming, the slope of the equatorial thermocline—the tilted boundary between warm surface water and cold deep water, tracked by the 20 degree Celsius isotherm—flattens, shoaling in the west and deepening in the east. After cessation, sea surface temperatures cool slightly while the subsurface keeps warming, so the flattened thermocline never recovers its tilt. The researchers also found hemispherically asymmetric changes in the Subtropical Cells, the shallow overturning circulations that carry cold water toward the equator. Anomalous low pressure and cyclonic winds over the South Pacific drive Ekman suction that slows the southern cell, reducing the upwelling of cold water, while the water that does upwell in the east is anomalously warm. Both effects inhibit any recovery of the zonal temperature gradient.

Feedback diagnostics add a final piece to the causal chain. Using linear regression, the team showed that the atmospheric wind–SST feedback strengthens during transient warming but returns to within its pre-industrial range once emissions cease—atmospheric coupling adjusts quickly. The oceanic feedbacks, by contrast, intensify and stay intensified: the thermocline–SST feedback remains elevated for centuries, meaning eastern Pacific surface temperatures become more sensitive to subsurface anomalies, and wind anomalies tilt the thermocline more efficiently. This reorganisation of oceanic feedback, combined with the persistently weakened zonal gradient and equatorward-shifted convergence zones, keeps ENSO variability amplified long after radiative forcing has stabilised.

The findings fit into a broader, non-linear picture of how ENSO responds to climate change. Palaeoclimate proxies and previous modelling work indicate that ENSO amplitude does not scale simply with warming: moderate shifts of the convergence zones under moderate warming can enhance variability, while extreme warming under very high carbon dioxide concentrations can suppress it, producing a permanent El Niño-like state with weakened variability. Net zero scenarios, which branch off at moderate warming levels, appear to sit in the amplifying regime—locking in a weakened gradient and elevated variability for centuries without ever reaching the extreme concentrations of idealised abrupt-forcing experiments. The study also differs from earlier work suggesting robust ENSO weakening under fixed high carbon dioxide, underscoring that the response depends critically on both the level and the trajectory of forcing.

The authors are careful about limitations. The millennial-scale feedback analysis rests on a single model, and ACCESS-ESM1-5 shares common biases such as a cold tongue and double ITCZ that could exaggerate eastern Pacific convection changes. Yet checks across the ZECMIP ensemble suggest the direction of the response is robust: in eight of nine models, the zonal gradient continues to weaken after cessation regardless of each model’s background state. The clear message for policy is that early mitigation matters twice over—delaying net zero not only increases the magnitude of the Pacific shift but also its persistence, committing generations to more frequent extreme El Niño events and altered rainfall patterns over regions such as East Asia and Australia. Net zero slows further change, the study concludes, but it does not restore the Pacific we knew.

Subject of Research: Persistence of an El Niño-like tropical Pacific mean state under net zero emissions due to ocean thermal inertia

Article Title: Ocean thermal inertia sustains a tropical Pacific mean state shift under net zero

Article References: Ocean thermal inertia sustains a tropical Pacific mean state shift under net zero. (n.d.). https://doi.org/10.1038/s43247-026-04082-x

Image Credits: AI Generated

DOI: 10.1038/s43247-026-04082-x

Keywords: El Niño, ENSO, net zero emissions, tropical Pacific, ocean thermal inertia, Southern Ocean, thermocline, climate modelling, ZECMIP, ACCESS-ESM1-5, extreme El Niño, ITCZ