I’ve always treated heat and smog as a package deal: the orange air-quality alert, the brown haze over the highway, the afternoon you keep everyone indoors. So I didn’t expect a climate model to tell me that warming the planet to 2.75°C could thin out some of that smog. I expected even less that the same model would show heatwave deaths rising in every region it looked at.
That’s the paradox in a paper published November 1 in npj Clean Air by Fuzhen Shen, Michaela Hegglin and colleagues at Forschungszentrum Jülich, with co-authors in Britain, China and Australia. In their simulations, ozone and fine-particle deaths fall across much of the world at 2.75°C, while heatwave deaths go up everywhere they checked. News-Medical ran it under the headline “A hotter world could have cleaner air, but deadlier heatwaves”, which is fair, though the first half of that headline is the half I’d read with the most suspicion.
Same winds, hotter oceans
The design is a clever bargain. The team took the punishing heat of 2018 and 2019 and replayed it in an atmospheric-chemistry model called EMAC, with the horizontal winds nudged toward the observed record so that the same high-pressure domes parked over the same places. Then they warmed the oceans to match 2°C and 2.75°C above pre-industrial levels and let temperature, humidity and chemistry respond on their own. Emissions, population, baseline death rates and people’s vulnerability all stayed at 2018–2019 levels. Heat deaths were estimated from risk relationships covering 750 locations in 43 countries or regions, and pollution deaths from published meta-analyses.
Before any warming is added, the model’s own estimates for those two years already upend an assumption I carried in. In Europe, heatwaves accounted for 148.5 attributable deaths per million people, against 29.7 for ozone and 10.4 for PM2.5, so among the three exposures the team scored, heat carried by far the largest estimated toll. Asia and Africa flip that order, with fine particles leading at 100 and 57.7 deaths per million.
| Group | Value (deaths per million) |
|---|---|
| Heatwaves | 148.5 |
| Ozone | 29.7 |
| PM2.5 | 10.4 |
Ozone’s quiet assassin
Ground-level ozone gets cooked out of exhaust in sunlight, and heat usually speeds that cooking near the tailpipe. So why would a hotter world end up with less of it?
Water. Warmer air holds more water vapor, and water is ozone’s quiet assassin. When sunlight splits an ozone molecule, it kicks loose an excited oxygen atom. In dry air that atom mostly calms down and rejoins an oxygen molecule, which rebuilds the ozone. In humid air it slams into water instead, splitting it into two hydroxyl radicals, and the ozone is gone for good. The paper pins its 2.75°C decline on “enhanced chemical destruction in the warmer and moister free troposphere,” the air above our heads that feeds background ozone down onto every continent.
Other models see the same tug-of-war. A 2022 multi-model analysis of CMIP6 Earth system models in Environmental Research Letters found a global-average surface-ozone decline of 0.96 parts per billion for every degree of warming, credited to faster destruction in moister air, while polluted regions and places rich in plant emissions moved the opposite way, gaining roughly 0.2 to 2 parts per billion per degree.
The new paper shows the same push and pull. At 2°C, ozone actually rose in several locations; the broad decline arrived only in the stronger warming scenario. I trust that detail more than the headline, because a balance between two forces, rather than a dial that only turns one way, is what real chemistry looks like.
The haze runs on its own physics. In high-NOx places like parts of the western United States, eastern Europe, India and China, much of the particle load is ammonium nitrate, a salt that hangs in the air as tiny specks. Warm it and it shifts back toward gaseous nitric acid and ammonia, which no longer counts as PM2.5. PM2.5-linked deaths fell at 2.75°C in Asia, Africa, Europe and North America. South America went the other way, which the authors connect to more secondary organic aerosol forming as heat drives up natural emissions. The team didn’t break the particles down component by component, so treat those mechanisms as the authors’ explanation rather than a measurement.
Fewer heatwaves, longer ones
Then the heat. At 2.75°C, modelled heatwave mortality rose in all six regions, according to the paper’s abstract: by more than 95 percent in South America, 81.6 percent in Africa, 66.5 percent in Australia, 58.6 percent in North America, 45.1 percent in Asia and 36.6 percent in Europe.
| Group | Value (percent increase) |
|---|---|
| South America (over) | 95 |
| Africa | 81.6 |
| Australia | 66.5 |
| North America | 58.6 |
| Asia | 45.1 |
| Europe | 36.6 |
Read those percentages next to where they start. South America’s steepest jump sits on the smallest heat baseline the paper reports, 18.4 deaths per million, while Europe’s gentler-sounding rise is stacked on 148.5. A huge percentage on a small base and a smaller one on a big base can mean very different numbers of people.
Heatwaves also stretched. In parts of North Africa, the Arabian Peninsula and the southern United States, episodes ran 5 to 10 days longer, even where the number of separate events fell. Fewer heatwaves sounds like relief, right up until you see that the ones left can last up to 10 days longer.
The trade cuts the other way too. New Scientist reported in 2025 that cleaner air has increased the number of city heatwaves, and the University of Texas at Austin researchers behind that work found that sunlight-reflecting aerosols had suppressed heatwaves in populated areas by about half since 1920. Haze shades cities, so clearing it lets more sun through. Clean air is no free lunch for temperature, and temperature is no free lunch for clean air.
What the frozen dials hide
The fine print matters more than usual here. The “cleaner air” comes from chemistry, not from anyone cleaning anything, because emissions were held fixed. Haze evaporating off a smokestack in a hotter world is not a policy achievement, and no agency or industry should get to count it as one.
You also can’t net the numbers out. The authors write that the heat, ozone and PM2.5 estimates “are not necessarily mutually additive because their underlying health pathways may overlap,” and News-Medical’s summary puts it plainly: they “cannot be added or subtracted to establish a net death toll.” So I won’t tell you heat outweighs the cleaner air, because the paper doesn’t support that sum. Nor did it test whether heat makes people more vulnerable to the pollution they still breathe, which is exactly the combination an elderly person in a stifling apartment faces.
The confidence intervals cover only the epidemiological risk curves; in the authors’ words, they “do not encompass the full structural uncertainty across the chemistry-climate model.” Vulnerability stayed frozen too, which means whatever protection people had in 2018 stays baked in and nothing new gets added. That isolates the physics cleanly, and it’s why the results are comparisons of warming effects, not forecasts of future deaths.
The work was funded through the German Research Foundation’s CLICCS cluster and a Helmholtz innovation pool, with computing time on German supercomputers. Co-author Domenico Taraborrelli is an associate editor of npj Clean Air; the paper states he was not involved in the journal’s review of the manuscript, and the authors declare no competing interests.
In this model, a warmer, wetter sky dismantles some of its own ozone and boils off some of its own haze, and the heatwave deaths rise anyway, in every region, while the longest heatwaves grow by up to 10 days. In Europe’s 2018–2019 estimates, heat already carried the biggest toll of the three. The study never compared air conditioners with air purifiers, so this is my call rather than its finding: if I had an aging parent in a top-floor apartment, my money would go to a working air conditioner and a plan to check on them every day of a long heatwave before a cent went to an air purifier.
Sources
- npj Clean Air, Shen et al. (2026)
- News-Medical: “A hotter world could have cleaner air, but deadlier heatwaves” (2026)
- Environmental Research Letters, Zanis et al.: climate change penalty and benefit on surface ozone, CMIP6 (2022)
- University of Texas at Austin: decline in aerosols could lead to more heatwaves in populated areas (2025)
- New Scientist: “Cleaner air has increased the number of city heatwaves” (2025)