A carbon sink is a place that takes in more carbon than it lets go. A peatland has done that for ten thousand years with one trick: it stays wet. Cut a ditch through it and the trick stops. The carbon does not disappear. It changes direction.
If you underwrite property in a coastal county, run a state air program, carry a scope 3 line for land, or manage a refuge next to somebody else's drained field, this is the flip you are already paying for — most years as an invisible exhale, some years as a fire that will not go out.
what a carbon sink is
A carbon sink is any reservoir — ocean, forest, soil — that absorbs more carbon from the atmosphere than it releases over a given period. Peat is the slow, ancient one: it banks carbon for millennia and can give it back in a decade if you drain it — reversible in a way a growing forest is not. A peatland is the slow, patient version. Waterlogged ground holds almost no oxygen, so the microbes that would normally eat dead plant matter nearly stop. What falls stays. The stock grows on the order of a millimeter a year and, given millennia, becomes meters of peat.
The intake is modest. For the pocosin peatlands of eastern North Carolina, Poulter and colleagues estimated an intact sequestration rate of roughly 1.1 to 1.3 tonnes of carbon per hectare per year. The stock is enormous not because the sink is fast but because it is old.
That stock exists whether or not anyone counts it, credits it, or insures against losing it. The peatland is the thing. What follows is what happens to the thing when the water leaves.
the flip: a drained peatland is a carbon source
Drainage is not dramatic. A ditch lowers the water table; air enters the upper peat; the microbes wake up and finish a meal they started centuries ago. The product is carbon dioxide, released steadily, with no flame and no smoke plume for a satellite to catch.
The IPCC's default for drained organic soils farmed as cropland in boreal and temperate climates is 7.9 tonnes of carbon per hectare per year (confidence interval 6.5–9.4). The temperate drained forest / shrubland row is 2.6. A ditched pocosin with shrub regrowth sits between those rows, not automatically on the cropland number. Every year. For as long as the ditch runs.
A drained peatland does not need a fire to become a carbon source. Oxidation does the work quietly, every ordinary year, for as long as the water table stays down.
The scale is not abstract. Near Belhaven, North Carolina, a 23-square-mile tract of pocosin was ditched decades ago for a farming venture that never took. Fifty miles of ditches still drain it, and the Associated Press reported in September 2026 that the site emits about 130,000 tons of CO₂ a year while it sits dry — a project figure, cited as such. Pantheon Regeneration is blocking those ditches to bring the water table back. Duke's Curt Richardson put the state total at 11,000 miles of canals draining peat toward the estuaries. None of that was dug by villains. It was dug by people solving the problem in front of them, and the people who inherited the ditches did not choose them.
Warming makes the flip steeper. A 2026 Nature Communications study led by the University of Münster, using the largest CO₂-budget dataset yet assembled for northern peatlands, found that the deeper the water table, the more sharply CO₂ emissions rise with temperature — and that a higher table dampens that response. It is a CO₂-only analysis; methane and nitrous oxide sit outside it.
the pulse: what a peat fire actually is
Fire belongs in a pocosin. Atlantic white cedar depends on it. In a wet peatland, fire runs across the surface, takes the shrub layer, and stops at the water. The ground is too saturated to burn.
Drain the ground and the ground becomes the fuel. A dry peat fire does not flame; it smolders. Low temperature, no visible front, spreading down and sideways through the soil itself, surviving rain, surviving the first month of suppression, producing the dense ground-level smoke that eastern North Carolina calls super fog. It is a different physics from the upland wildfire cluster. An upland fire burns what stands on the land. A peat fire burns the land. If your question is why the air across the West is getting worse, that is a separate post: why is wildfire smoke getting worse. This one is about the ground.
evans road, 2008
On 1 June 2008, lightning struck private farmland about five kilometers south of Pocosin Lakes National Wildlife Refuge. The flaming phase — the part that looks like a wildfire — ran until 5 July. Then the peat took over, and the fire was declared out only after months of smoldering.
Mickler, Welch, and Bailey reconstructed the emissions from pre- and post-fire elevation surveys. The fire covered 16,814 hectares. Mean peat burn depth was 0.42 meters; the deepest measured was 1.7 meters. Total release: 9.47 teragrams of carbon, 9.16 of them from the peat below ground. That is 9.47 million tonnes of carbon — around 35 million tonnes of CO₂ — from one lightning strike on one dry summer.
Per hectare, the fire released about 563 tonnes of carbon. Against an intact intake of 1.1 to 1.3 tonnes a year, one summer undid four to five centuries of accumulation. Against the ordinary drained-year figures, that is seventy to two hundred years of quiet oxidation delivered in a single pulse — depending on whether you compare against the cropland default or the shrubland row.
The variable was water. The refuge lands inside the perimeter were managed with gated culverts by the U.S. Fish and Wildlife Service to hold the table high; adjacent private lands were managed to lower it more than two meters for agriculture and hunting. Burn depths on the public side were markedly shallower. Three years later the Pains Bay Fire burned a similar footprint — 18,329 hectares, sixty kilometers away — on wetter ground: mean burn depth 0.13 meters, 5.01 teragrams of carbon. Same coastal plain, different water table — and, the paper is honest, different vegetation associations — one-third the depth, about half the carbon. Nobody chose the fire. What had been chosen, decades earlier and for ordinary reasons, was how deep the water would be when the lightning came.
The world-scale version is Indonesia in 2015, when drained peat across Sumatra and Kalimantan burned through an El Niño dry season and released an estimated 227 teragrams of carbon in eight weeks — the same mechanism as Evans Road, at the scale of a nation, in communities who breathed it first.
sink, source, stock
| condition | what the carbon does | ordinary year, per hectare | when it burns | who carries it |
|---|---|---|---|---|
| intact — wet | sink | banks about 1.1–1.3 t C | surface fire; peat too wet to burn deep | the owner still carries the opportunity cost of not farming the ditch |
| drained | source | loses about 7.9 t C (IPCC temperate cropland default) | smolders 0.4–1.7 m into the ground for months; Evans Road averaged 563 t C/ha | whoever is downwind, downstream, or holding the inventory line |
| rewetted | stock held | oxidation clock stops; the sink returns slowly; methane can rise for a period | fire returns to the surface | whoever funds the water table |
Figures are per hectare and drawn from Mickler et al. (2017), the IPCC 2013 Wetlands Supplement, and Poulter et al. (2006). Rewetting is honest work with an honest caveat: raising the table cuts CO₂ but can lift methane, and the two do not net to zero on day one. Rewetting is keeping the peat wet goes through the mechanism.
who already pays for the failure
The failure has a payor list. It just is not labeled.
State and federal agencies pay suppression on a fire that cannot be put out, only outlasted, and then pay the health and road-closure costs of months of ground smoke. Insurers with coastal-county exposure pay the smoke days, the closed highways, and the claims that arrive under other names; no line of business prices peat smolder; the exposure sits in the book anyway, under smoke days, closed highways, and claims that arrive under other names. Evans Road was declared out in January 2009 at roughly $20 million in suppression cost (a conference figure, not a peer-reviewed loss study). Superfog — visibility under ten feet, a term from USDA meteorologist Gary Achtemeier, not local slang — has a documented auto-liability record: I-4 in Florida in 2008, I-75 in 2012. Corporations that have started counting land in their inventories find a line that grows every year the water stays down. And the refuge that spent decades installing culverts still breathes the neighbor's fire.
Every one of those bills is a receipt for a decision about water made long before the lightning. None of them is the worth of the peatland. The peat is a living stock with a ten-thousand-year memory, a sponge for an estuary, a nursery for the fisheries downstream. The fire bill only tells you what it costs when the water is gone.
fund the water table before the smolder
The order of operations is the whole point. Suppression, health costs, and inventory write-downs arrive after the water leaves. Restoration credits arrive after the ditches are blocked and the reductions are verified. Both are real. Both are late.
Ensurance funds the wet condition now, on a named place. A certificate of ensurance is a position in a named agent's account — not title to the land — and buying it funds the present condition of that place: the water table today, not the tonne avoided in a future audit. The live doors are small and we will say so: peatland.syndicate coordinates peatland agents, and inland-wetlands.ensurance is the purpose account for the wetland stock these pocosins belong to. Volumes are early. The mechanism is not a promise; it is a place to put the water-table decision before the dry summer makes it for you.
If you hold coastal exposure, see how insurers use ensurance — the argument is that funding the wet condition is cheaper than outlasting the smolder. If you came here for the definition, go back to the start: what peatlands actually are.
Price is the bridge. The peatland is the thing. Drain it and it burns.
the series
Six posts on peatlands — what they are, why they are only peatlands while wet, and what to hold if you want them to stay that way.
- what peatlands actually are — a wet living carbon stock, not a swamp amenity and not a credit
- rewetting is keeping the peat wet — block the ditch, raise the table, stop the peat from becoming air
- drain it and it burns — the sink flips to a source, then to a smolder (this post)
- a carbon credit is not a wet peatland — the receipt can be real; it is still not the water table
- bogs, fens, and pocosins are not the same peatland — rain-fed, groundwater-fed, and a swamp on a hill
- who pays to keep peat wet — the credit buyer, the refuge, and the drainage district already pay for failure