On 2 September 2026 the UN Environment Programme published a report from Nairobi that changes the question rather than the answer. Limiting Overshoot: Navigating exceedance of 1.5°C and pathways towards return does not ask whether the world will cross the Paris Agreement’s aspirational limit. It assumes the crossing, and asks how high the peak goes, how long it lasts, and whether temperatures can be brought back down afterwards.

Under the most optimistic reading of current government pledges, the report puts expected peak warming at about 1.8°C. Under current policies rather than pledges, it gives a median of roughly 2.6°C by 2100, within a range of 1.9°C to 3.6°C.

What “overshoot, peak and decline” means

The phrase describes a three-part pathway. Temperatures exceed 1.5°C. Emissions cuts, including sharp reductions in methane, hold the peak as low as possible. Then, once the world reaches net-zero and moves into net-negative emissions — removing more carbon dioxide from the atmosphere than it emits — temperatures come back down below the threshold.

UNEP identifies this as the best remaining option. UN Secretary-General António Guterres, quoted in the launch material, said: “We must make the overshoot above 1.5 degrees as small and short as possible.” The report’s evidence, consistently stated, is that higher peaks and longer periods above the threshold mean greater risks for people and ecosystems — and that some of those impacts, including ice sheet loss and sea level rise, do not reverse when the temperature does.

Why this is a change of framing, not of science

None of the underlying findings are new. UNEP’s annual Emissions Gap Report has tracked the distance between pledges and pathways since 2010, and last year’s edition already concluded that a temporary overshoot was effectively unavoidable. What is new is that a flagship UN assessment now starts from the exceedance rather than treating it as the thing to be prevented.

The arithmetic behind it: more than 400 gigatonnes of carbon dioxide have been emitted cumulatively since the Paris Agreement was signed. For pathways that hold warming to 1.5°C with limited exceedance, the report cites the requirement from the 2025 Emissions Gap Report that global greenhouse gas emissions fall 55 per cent from 2019 levels by 2035 — a figure calculated on the assumption that action began in 2020. It did not.

The part that does not currently exist

The return leg of the pathway depends on carbon dioxide removal at a scale that has no precedent. Total removal today stands at roughly 2.2 gigatonnes of CO2 a year, almost all of it conventional and land-based: afforestation, ecosystem restoration and forest management. Engineered removal — direct air capture, enhanced weathering, bioenergy with carbon capture and storage — is a rounding error by comparison.

Against that, annual emissions are in the region of 50 gigatonnes of CO2 equivalent. Getting from a world that removes 2.2 gigatonnes to one that removes more than it emits is not an incremental step, and the report is blunt about the distance. It also flags that land-based removal competes for land with food production and with the ecosystems it is meant to protect, and that stored carbon can be released again by fire, drought or a change of government.

The argument about saying it out loud

Not everyone in the field welcomes this framing, and the disagreement is worth setting out properly because it is about strategy rather than data.

One camp, including many climate scientists and small-island negotiators, has argued for years that formally conceding overshoot creates moral hazard: it converts 1.5°C from a limit into a waypoint, gives governments a reason to defer cuts on the promise of future removal, and quietly transfers the burden to countries and generations that did least to cause the problem. On this view, a pathway that relies on unbuilt technology to undo present emissions is not a plan but a deferral.

The other camp, which includes UNEP itself here, argues that pretending the threshold will hold is worse. Adaptation planning, infrastructure standards, insurance and agriculture all depend on realistic temperature assumptions, and building for 1.5°C in a 1.8°C world produces defences that fail. On this view, naming the overshoot is what makes it possible to argue about its size — which is the only variable still available.

Both positions accept the same physics. They differ on what governments do with a concession once it has been made, which is a question about politics, and neither side can prove its case in advance.

What to watch

The report’s practical test is whether “peak warming” starts appearing in national climate plans as a target in its own right. Countries currently pledge emissions cuts by a date; nothing in the Paris architecture asks them to state what peak temperature their plan implies or when they expect to reach net-negative. Watch for whether the next round of nationally determined contributions adopts that language, whether any government commits to a removal target separate from its emissions target, and whether the finance discussion at the next COP treats carbon removal as infrastructure to be built or as an offset to be bought.

In the meantime, the near-term weather is being shaped by something else entirely: the very strong El Niño that WMO now expects to last into 2027.

Sources

• UN Environment Programme, Limiting Overshoot: Navigating exceedance of 1.5°C and pathways towards return, 2 September 2026
• UNEP press release, Nairobi, 2 September 2026
• Nature news, on the report’s conclusion that warming will breach 1.5°C this decade, September 2026
• UNEP, Emissions Gap Report 2025