After Net-Zero, Earth Cools: An Observational Constraint Settles a Long-Running Model Fight

Picture the planet at the instant the last tonne of carbon dioxide is neutralized. The furnace has been switched off. The obvious question is whether the room now stays warm, keeps heating, or begins to cool. Until now, the answer has come back with an uncomfortably wide range, depending on which climate model you asked.

That uncertainty is not academic. It shapes how much warming the world is committed to even after it does everything right, and it feeds directly into how much carbon can still be emitted before a given temperature limit is locked in. Some models suggested temperatures would hold roughly flat for centuries. Others hinted at a gentle decline. A few left room for continued creep upward. The disagreement traced back to two opposing tugs in the Earth system that are devilishly hard to pin down, and small differences in how a model handled them swung the outcome.

Nathaniel Tarshish, Nadir Jeevanjee and Inez Fung, working out of the University of California, Berkeley, took a different route. Instead of trusting a model to get those two tugs right, they measured them. More precisely, they constrained them using multiple lines of evidence, pinning the physics to observations rather than to model tuning.

The two tugs are best thought of as imbalances that the climate system carries after emissions stop. The first lives in the carbon cycle. Even with no new emissions, the ocean and the land keep drawing down carbon dioxide that is already in the atmosphere, so atmospheric concentrations slowly fall. That pulls the warming down. The second imbalance lives in the ocean itself. The deep sea has spent more than a century absorbing heat and still lags far behind the surface; as it slowly catches up, it keeps a warming pressure on the surface for a long time. That props the temperature up.

Whether Earth cools, holds, or warms after net-zero comes down to which imbalance wins. The Berkeley team's insight was that both can be constrained from what the planet is already telling us (the observed drawdown behavior of the carbon system, and the observed thermodynamic disequilibrium of the ocean) rather than left to the internal assumptions of any one model. Constrain both, and the tug-of-war resolves.

It resolves toward cooling. With greater than 95% confidence, the authors report, global temperature anomalies decline after net-zero. If emissions were to cease today, they estimate a multi-century cooling of about 0.4 °C. In the paper's framing, that is roughly a 40% reduction of the current temperature anomaly, playing out over centuries rather than years. For other emissions pathways, where the world keeps warming for longer before reaching net-zero, they find a long-term decrease of 0.7 to 1.0 °C relative to the peak temperature at the moment net-zero is achieved. The two figures describe different starting points and should not be blurred together: 0.4 °C is the cool-down from where we stand now; the 0.7 to 1.0 °C range is how far temperatures fall below whatever peak a later net-zero would reach.

The distinction matters for how the result is read. It does not say the planet snaps back to a pre-industrial climate, and it does not say cooling is fast. Centuries is the operative timescale. What it does say is that the direction, long muddied by model disagreement, points down, and that net-zero is a hinge after which the system slowly relaxes rather than a ceiling it presses against forever.

A companion News & Views article in the same issue of Nature Climate Change, by Andrew MacDougall, carries the argument a step further, framing the takeaway as less than net-zero may be needed to stabilize the climate. If temperatures are already inclined to fall once emissions reach zero, then merely holding the climate steady could require somewhat less effort than driving all the way to net-zero. That is a subtle but consequential shift in how the goalposts are drawn.

The finding is observationally constrained, but that phrase means the two key metrics are pinned by multiple lines of evidence, not that the whole trajectory was read straight off a thermometer. The magnitudes sit comfortably within the existing zero-emissions-commitment literature, and an independent study using a different question (the hysteresis of warming under active carbon removal) reaches a directionally consistent conclusion, offering modest corroboration rather than an echo.

Still, the shape of the answer is now clearer than it has been. Switch off the furnace, and the room does not stay hot forever. It cools: slowly, over centuries, but it cools.