The Pollution–Climate Risk Nexus

The map has five nodes. Climate and pollution sit at the centre-line; health, ecosystems and development sit around them, because that is where the damage lands. Each arrow is a causal link with published evidence behind it. Not every pathway closes: some run from climate through pollution into health and stop there, which makes them amplifiers rather than loops. The distinction matters for strategy, because an amplifier can only be damped at its links, while a loop can be broken.

The fire loop and the methane–ozone loop both close through the climate: warming makes more fire and more methane and ozone, and those warm the planet. The cooking loop closes through development: poverty sustains polluting fuels, and the illness they cause sustains poverty. Heat and air is an amplifier: heat makes pollution more harmful and more abundant. Ocean deoxygenation is an amplifier where warming and nutrient pollution combine.

What all five share is that each crosses at least two of the three domains in the title. That is why treating pollution and climate separately misses them: a pollution agency sees the smoke, a climate agency the carbon, a health ministry the deaths and an agriculture ministry the lost harvest, and none of them sees the loop. A loop is visible only when the links are drawn together, as here.

The two charts tell opposite stories, and both are true. Burned area is dominated by fires in African and South American savannas, which have declined as grasslands are converted to cropland and pasture. The fires that close the loop are different: forest fires in boreal and temperate conifer regions, which release far more carbon per hectare, burn through peat and soils, and produce the smoke plumes that cross continents. Those are the fires that have multiplied, and the boreal series shows how far outside the earlier record 2023 fell.

The loop's health link is the most immediate. Landscape fire smoke contributed to about 1.53 million deaths a year on average over 2000–2019, according to Xu and colleagues in The Lancet, mostly through fine particles and partly through ozone. The Lancet Countdown estimated about 154,000 deaths from fire smoke in 2024 alone using a narrower method; the two figures measure different things and should not be compared directly, but both place fire smoke among the larger air-pollution burdens.

The loop's largest link — global warming — can only be weakened slowly, by cutting emissions everywhere. Its other links are local and faster to act on: fuel management and prescribed burning in fire-prone forests, rapid detection and suppression of new ignitions, protection of peatlands, and, at the health end, smoke forecasts, clean-air shelters and filtration. Section 7 returns to why that local work matters even though it does not touch the global driver.

The methane–ozone loop has an unusually cheap link. Wetland emissions cannot be switched off, but human-caused methane — from oil and gas, coal, landfills and agriculture — can be cut quickly, and because methane lasts only about a decade in the atmosphere, cuts show up in temperature and ozone within years rather than decades. The UNEP and Climate and Clean Air Coalition Global Methane Assessment found that cutting human-caused methane by 45 per cent this decade would avoid nearly 0.3 °C of warming by 2045, about 260,000 premature deaths, 775,000 asthma-related hospital visits, 73 billion hours of lost labour and 25 million tonnes of crop losses every year.

The cheapest part of that cut is in fossil fuels. The IEA's Global Methane Tracker estimates that fossil-fuel operations emitted about 124 million tonnes of methane in 2025, that about 70 per cent could be abated with existing technology, and that about 30 per cent — more than 35 million tonnes a year — could be abated at no net cost, because the captured gas is worth more than the equipment to capture it. Few climate measures pay for themselves; fixing leaks, ending routine flaring and venting, and capturing coal-mine methane do.

The growth of methane slowed in 2023–2025, to 4.8 parts per billion in 2025, the lowest in nearly two decades. Whether that reflects cooler, drier conditions in tropical wetlands, changes in the atmosphere's capacity to destroy methane, or early effects of abatement is not yet settled. The loop remains: as the climate warms, wetland emissions are expected to rise, which makes the human-caused share the part worth cutting now.

Exposure has turned a corner, but not everywhere. Population-weighted PM2.5 exposure worldwide rose from about 24 µg/m³ in 1998 to a peak of about 35 in 2016, then fell to about 30 by 2024, in large part reflecting China's clean-air programme. In high-income countries it fell from about 16 to 13. In lower-middle-income countries — including India, Pakistan, Bangladesh and Nigeria — it fell from its 2016 peak of about 51 to about 43. In low-income countries it rose over the period, from about 23 to 30. Every group remains far above the WHO guideline of 5.

Because the heat amplifier works on existing pollution, its harm concentrates where both are high: the Indo-Gangetic plain, the Sahel and the megacities of South and West Asia. These places are also where heat is rising fastest relative to people's ability to cool themselves. The amplifier cannot be broken, because heat will keep rising for decades; it can be damped from both ends — by cutting the precursors that heat turns into pollution, and by protecting people on the days when heat and pollution coincide.

The protective end is proven. Ahmedabad's Heat Action Plan, India's first, combined early warnings, public cooling and health-system preparation; an evaluation associated it with about 1,190 avoided deaths a year (95 per cent confidence interval 162 to 2,218). Pairing heat warnings with air-quality alerts — and treating hot, polluted days as a single emergency — is a low-cost way to damp the amplifier while the precursors are cut.

The ocean is losing oxygen. Warmer water holds less dissolved oxygen and mixes less with the deep ocean, and the open ocean has lost about 2 per cent of its oxygen since the mid-twentieth century, according to the studies summarised by Breitburg and colleagues in Science. Some regions have lost far more. This is the climate half of the amplifier: warming on its own lowers oxygen everywhere.

The pollution half is concentrated on coasts. Fertiliser and sewage running off the land feed algal blooms that, as they decay, strip oxygen from coastal waters. Reported coastal dead zones have risen from fewer than 50 in 1950 to more than 500, and monitoring is patchy enough that the true number is likely higher. Where warming and nutrient pollution combine, low-oxygen zones expand, last longer and push fish and shellfish out of their habitats.

The development consequence falls on the coastal communities that depend on those fisheries. The amplifier is damped most directly at the pollution end: more efficient fertiliser use, wetland buffers on farmland, and wastewater treatment in coastal cities. These measures are local, well understood and pay off within years — which, as in the other loops, makes them the cheapest place to start while the global driver is addressed.

The cooking loop's cheapest link is the fuel. The IEA estimates that universal access to clean cooking by 2030 would require about $8 billion of investment a year — a very small sum against the 2.9 to 3.7 million deaths a year that household smoke causes, and against the global energy system's annual investment. The same transition would reduce greenhouse gas emissions by about 1.5 billion tonnes a year, mostly by reducing unsustainable fuelwood harvesting and the methane and black carbon from incomplete combustion.

Breaking the loop at the fuel does more than remove smoke. It frees time for school and paid work, reduces pressure on forests and cuts emissions of short-lived warmers, so it weakens the development loop and touches the fire and climate loops at the same time. This is the pattern this dashboard is built to show: the best interventions cut several loops at once.

The obstacle is affordability, not technology. Households switch when clean fuels are cheaper and more reliable than the alternatives, which is why the successful national programmes subsidised stoves and refills and expanded distribution. Sub-Saharan Africa, where the number without access is still growing, is where the next $8 billion a year would do most.

Two of the breakers stand out because they touch several loops at once. Cutting methane weakens the methane–ozone loop, reduces the ozone that heat turns into a health emergency, and protects crops and forests. Clean cooking weakens the development loop, reduces pressure on forests and fire, and cuts short-lived warmers. These are the interventions to fund first, because each dollar works in more than one loop — and in both cases the cost per unit of harm avoided is among the lowest known.

One loop works against the others. Sulphur dioxide from coal and shipping is a deadly air pollutant, but the aerosols it forms also cool the planet by reflecting sunlight. Cleaning it up — rightly — removes some of that cooling, as this series examined in One Source, Two Signs. The answer is not to keep the sulphur but to pair clean-air policy with fast cuts in methane and other short-lived warmers, so that the health gains of clean air are not paid for in extra warming.

Breaking loops at their cheapest links does not replace cutting carbon dioxide; it buys time and health while that slower work proceeds. Every loop in this dashboard is driven in the end by warming, and every one gets harder to break as warming continues. The practical message is about sequence: act first where the cost is lowest and the benefit spans the most loops, then keep going on the links that cost more.

Charted series are from public datasets: NOAA's global annual mean methane and growth rate (1984–2025); the Global Fire Emissions Database (GFED4s) annual fire carbon for boreal North America and boreal Asia (1997–2023), whose 2017+ values GFED labels as beta and unsuitable for trend analysis; GWIS annual burned area via Our World in Data (2012–2024; 2025–2026 incomplete); SatPM population-weighted PM2.5 exposure by World Bank income group via Our World in Data (1998–2024), which includes natural dust and smoke; and the WHO/World Bank share of the population without clean cooking fuels (2000–2024).

Other figures are from the literature cited in the text and listed below: State of Global Air 2025, Cunningham et al. (2024), Byrne et al. (2024), Xu et al. (2024), Mills et al. (2018), the Global Methane Assessment (2021), the IEA Global Methane Tracker 2026 and its clean cooking analysis, Breitburg et al. (2018), the WMO summary of IPCC ozone projections, the Indian heat–PM2.5 study (2025) and the Ahmedabad Heat Action Plan evaluation (2018). Where sources disagree — household air pollution deaths of 2.9 million (WHO) or 3.7 million (IEA) — both are reported.

The loop calculator treats a reinforcing loop as three links with strengths s₁, s₂, s₃; the loop gain is their product G, and the amplification of an external push is 1 / (1 − G). Cutting a link by share c multiplies its strength by (1 − c). The default strengths (0.8, 0.8, 0.78) are illustrative, chosen so that G ≈ 0.5; they are not estimates of any real loop. Related in this series: megacity air quality, the pollution–poverty–climate nexus and one source, two signs.