How attribution science works for marine heatwaves — fraction of attributable risk, model counterfactuals, and what it means for baselines and reversal risk.
When a marine heatwave kills a kelp forest or a seagrass meadow, an uncomfortable question follows every report: was that climate change, or just weather? Attribution science exists to answer that question properly — not with a shrug or a slogan, but with a probability that can be stated, defended, and priced. This explainer covers how marine heatwave attribution actually works, why the ocean is in some ways easier to read than the atmosphere, and — because this is where it starts to matter commercially — what attribution means for the baselines and reversal-risk language in blue carbon and restoration deals. For the phenomenon itself, start with our marine heatwaves explainer.
Attribution does not claim that climate change "caused" a specific heatwave, in the way a hit-and-run has a sole driver. The ocean has always had warm spells. The precise question is counterfactual: how much more likely — and how much more intense — was this event because human activity warmed the planet?
To answer it, scientists compare two worlds:
Large ensembles of climate model runs generate each world many times over, because you need hundreds of simulated decades for rare events to show up at all. The heatwave in question is then counted in both worlds: how often does an event of this magnitude, duration and location occur naturally, versus in the warmed world?
The headline output is often the fraction of attributable risk (FAR): the share of the event's likelihood that can be assigned to human influence. If an event class occurs naturally once in a while but becomes, say, several times more likely in the warmed world, then most of the risk of that event is attributable — the FAR climbs toward one. Closely related is the probability ratio, the simple multiplier of how much more likely the event became.
Attribution can also be framed as intensity rather than frequency: how much warmer did this heatwave run because of background warming? For marine heatwaves this framing is intuitive, because so much of the attribution signal is a shifting baseline — the whole distribution of ocean temperatures has moved toward warmth, so events that used to define a region's extreme tail now sit closer to its middle. We unpack why a spike and a slope are different beasts in marine heatwave vs ocean warming.
Extreme-event attribution on land wrestles with complicated machinery: soil moisture feedbacks, land use change, chaotic circulation patterns that can raise or flatten extremes regionally. Ocean surface temperature attribution has two structural advantages:
That does not make every marine heatwave trivially attributable. Regional drivers — El Niño phases, blocking weather patterns, anomalous currents, reduced mixing from calm winds — stack on top of the background trend, and good studies separate the layers. But the direction of the literature is unambiguous: attribution studies of individual recent marine heatwaves have repeatedly found that human-caused warming made events like them far more likely, and some analyses find that most of the large marine heatwaves of the modern record would have been vanishingly improbable without it. Several global analyses report that the frequency, duration and intensity of marine heatwaves have all increased over the satellite era, with a dominant human contribution.
Alongside the probability-based method, scientists increasingly use storyline attribution: hold the actual weather pattern fixed (this high-pressure ridge, this slack wind, this weak upwelling) and ask how warm the event would have been in a cooler ocean. Storylines trade the probability question for a conditional one — they can't tell you how likely the pattern was, but they answer the question a manager actually asks: given this weather, how much of the heat is ours?
Both methods are legitimate; they answer different questions, and the strongest attribution papers say which question they are asking.
Attribution explains the past and present; marine heatwave forecasting is developing in parallel, with seasonal prediction systems now showing real skill in some regions months ahead. The combination is potent for coastal managers: attribution tells you the background risk has structurally risen; forecasts tell you this summer's concrete odds. We review the state of that art in can we predict marine heatwaves.
Here is where attribution science stops being an academic sport. Any credible blue carbon project deals with heatwaves in at least three places, and attribution thinking sharpens all three.
Baselines that assume a stationary climate are wrong. A baseline built from the 1990s return period of heat stress silently imports a climate that no longer exists. If heatwave frequency is trending — and attribution says the trend is human-forced and ongoing — then the "without project" world is itself a moving target. Conservative baselines model a worsening disturbance regime, which (counterintuitively for sellers) can make some restoration claims larger and some avoided-loss claims smaller. Either way, the honest baseline is a non-stationary one. See setting a kelp carbon baseline.
Reversal-risk language should name attribution. When a heatwave reverses stored carbon, is that an insurable "natural disaster" or the predictable unfolding of a priced-in trend? The older deals treat all extremes as acts of God. The newer reading — the one we argue for — is that a portion of heatwave risk is attributable, quantifiable and therefore underwritable: not an excuse for lost credits, but a number that should size buffer contributions and insurance premiums. The mechanics sit in our explainers on reversal risk and buffer pools.
Buyers should ask attribution-shaped questions. Does the project's reversal-risk analysis reference the region's attribution literature? Does the monitoring plan distinguish a climate-attributable trend from a containable local pressure? A project that cannot say whether its hazard model is stationary is pricing risk with yesterday's ocean. Our reporting methodology describes how we treat disturbance trends in our own (fictional) programme accounting.
Attribution is strong at the regional-to-basin scale for temperature extremes and weakest exactly where restoration often lives: the shallow, local, topographically complicated coastal strip where a single headland can split one water mass into two fates. Fine-scale coastal attribution is data-hungry — long local temperature records are rare — and model resolution strains at bays and headlands. Practically, that means attribution statements get stronger as you aggregate (this coastline, this decade) and softer as you zoom to your project site. Rigorous practitioners use regional attribution to bound the trend and local records to tune the tail.
It also answers only the physical question. It does not tell you a heatwave was "unnatural" in every sense — the ocean has always produced hot years — and it does not by itself assign legal liability, though attribution evidence is increasingly cited in climate litigation. What it gives a contract is something more useful than blame: a defensible probability.
Not with sole-cause certainty — and good scientists don't try. The rigorous statements are probabilistic: human-caused warming made an event of this magnitude this many times more likely, or this much hotter. Because ocean warming is so strongly forced by greenhouse gases, marine heatwave attribution statements are often among the most confident in all of extreme-event science.
It is the share of an event's odds that comes from human influence. A FAR of 0.9 for a class of heatwave means: in the warmed world, nine-tenths of the risk of that event exists only because of human forcing. It converts "climate change played a role" into an auditable number — which is exactly what risk contracts need.
No single event is wholly the work of any one driver — weather modes like El Niño still organise the timing and placement of major events. What attribution finds is that the background warming stacks the deck: the same weather pattern now produces hotter, longer events than it used to, and some recent extremes would have been essentially impossible in an unwarmed ocean.
Ask three things: whether the project's hazard model assumes a stationary climate (it shouldn't); whether its reversal-risk assessment cites the regional attribution literature; and whether buffer contributions or insurance reflect the trend, not the historical average. Our buyer's due diligence guide folds these questions into a full checklist.
Attribution per se looks backward and present-ward, but its outputs feed forward: the same model ensembles used for attribution are used to project how heatwave likelihood will keep evolving under different emissions paths. That projection, not the historical record, is the right input for any long-lived carbon storage claim in the coastal ocean.
Blue carbon is stored in living ecosystems that can burn, sicken or wash away. How reversal risk works, what triggers it, and how honest projects insure it.
Priya Raman ·
A carbon credit is only real if the baseline against which it's measured is honest. How baselines are set for kelp and blue carbon projects — and why they decide integrity.
Priya Raman ·
Stored carbon is a promise about the future. How permanence risk, buffer pools and reversal mechanisms keep that promise honest across carbon markets.
Priya Raman ·