The bigger part of aviation's warming
CO2 is not the only way flying warms the planet. The white lines behind aircraft can spread into cirrus cloud that traps heat. On the reference assessment, that effect is larger than the warming from aviation's CO2.
The numbers
At any given moment, contrails trap more heat than all the CO2 aviation has emitted since 1940.Lee et al. 2021: contrail cirrus 57.4 mW/m² against 34.3 mW/m² from all aviation CO2 accumulated since 1940 (2018).
Contrail science is uncertain, and the authors rate their confidence in the contrail term as low. On a CirculAIRity Monte Carlo of Lee et al.'s own uncertainty ranges, the contrail share lies between 27% and 73% (5–95%). We show ranges, never a single heroic number.
Ice on soot
A contrail forms when hot, humid exhaust mixes with cold air and briefly reaches water saturation. Water condenses on particles in the plume, mainly engine soot, and freezes into ice crystals. Most contrails vanish within minutes; in ice-supersaturated air they persist and spread into cirrus that can last for hours. Kärcher 2018
Trapping heat, day and night
By day a contrail reflects sunlight, which cools. Day and night it traps heat from the surface, which warms. On the global average the warming wins, and night-time and winter contrails warm the most. Fewer, larger ice crystals make a thinner, shorter-lived contrail. Kärcher 2018; Burkhardt et al. 2018
Rerouting helps some flights. It cannot reach most.
Contrail avoidance changes a flight's time, altitude or path so it misses the air most likely to form warming contrails. It works when the right flights can be moved and the forecast is right.
- A 2021 night-time live trial at Maastricht avoided contrails by moving flights ±2,000 ft. EUROCONTROL MUAC 2024
- A 2023 airline trial saw 63.6% fewer contrails on flights given avoidance guidance, against a control group. Sanekommu et al. 2024
- Short-term forecasts of ice-supersaturated air remain poor. Gierens et al. 2020; Hofer, Gierens and Rohs 2024
- In simulation, blocking flight levels for avoidance cut sector capacity by at least 20%. EUROCONTROL MUAC 2024
- Every diversion burns extra fuel: a certain CO2 cost against an uncertain contrail benefit.
- The largest planned European avoidance programme would reach about 0.04–0.05% of annual European flights. CirculAIRity calculation: about 4,500 flights a year (Blue Lines 2024, citing Aviation Week) ÷ 10.69 million flights in 2024 (EUROCONTROL 2025)
Fewer particles, fewer crystals
Published modelling and flight data show contrail forcing falls when soot particle numbers fall, by up to about 44% for fleet-wide 100% SAF. We are testing how far a specification-compliant tuned fuel can move along that curve.
A fuel lever acts on every contrail a flight forms, including the ones a forecast misses, so it complements avoidance. It works alongside SAF, not instead of it.
- Low-aromatic 50:50 blends cut soot and contrail ice by 50–70% in flight. Voigt et al. 2021
- 100% HEFA cut contrail ice by 56% in flight. Märkl et al. 2024
- Fleet-wide 100% SAF cut contrail net forcing by about 44% (model). Teoh et al. 2022
- Targeting cleaner fuel at the right flights multiplied its climate benefit 9–15 times. Teoh et al. 2022
- Forcing falls less than in proportion to particle number; at very low soot, volatile particles take over. Bier and Burkhardt 2022; Voigt et al. 2026
Tonnes of CO2-warming-equivalent
tCO2-WE expresses any emission or effect as the tonnes of CO2 that would cause the same change in global temperature. It comes from the GWP* method developed at the University of Oxford, and it is the basis on which the reference assessment reports aviation's warming. Allen et al. 2018; Cain et al. 2019; Lynch et al. 2020; Lee et al. 2021
| Metric | What it measures | How it treats contrails | Used by |
|---|---|---|---|
| GWP100 (tCO2e) | Heat trapped over 100 years after a one-off release | A year of contrails counts as a short pulse spread over a century | UNFCCC reporting; EU and UK ETS; CORSIA |
| GWP20 | Heat trapped over 20 years after a one-off release | Counts contrails heavily; the 20-year horizon is arbitrary | Some short-term climate studies |
| GTP100 | Temperature 100 years after a one-off release | Counts almost nothing: the warming has long gone | Some IPCC comparisons |
| GWP* (tCO2-WE) | The temperature effect of a change in emission rate | Counts a sustained cut in contrails as the CO2 removal that would cool the planet as much | Lee et al. 2021; CirculAIRity |
The same year of contrails is 1.9 tCO2e per tonne of fuel on GWP100, and 5.3 tCO2-WE over a 30-year period on GWP*. The period matters: about 7.6 over 20 years, 3.3 over 50 and 1.9 over 100.
CirculAIRity calculation from Lee et al. inputs (contrail cirrus 57.4 mW/m²; 327 Mt of fuel in 2018; α = 0.25, H = 100 years per Cain et al.). Lee et al.'s own Table 5 gives about 5.6, 2.0 and 7.3 per tonne on GWP*, GWP100 and GWP20, which corroborates it.
Our standard: every result in tCO2-WE by GWP*, over a declared 30-year period, with GWP100 and GWP20 alongside.
Both units are right
Regulators count in tCO2e: it is the legal unit under the Paris Agreement, and the EU's non-CO2 monitoring converts each flight's effects using GWP20, GWP50 and GWP100. GWP* has been criticised for favouring established emitters, and we say so. Rogelj and Schleussner 2019
But the question that matters for a contrail measure is how much warming it avoids, and that is what tCO2-WE measures. The IPCC recognises the difference and recommends no single metric: the choice depends on the question. IPCC 2021, AR6 WG1
tCO2e is right for accounting and compliance. tCO2-WE is more right for the question that matters: how much warming is avoided. We report both and decide on tCO2-WE.
Sources
- Lee et al. 2021
- Teoh et al. 2024
- Teoh et al. 2022
- Kärcher 2018
- Burkhardt et al. 2018
- Bier and Burkhardt 2022
- Voigt et al. 2021
- Märkl et al. 2024
- Voigt et al. 2026
- Gierens et al. 2020
- Hofer, Gierens and Rohs 2024
- EUROCONTROL MUAC 2024
- Sanekommu et al. 2024
- Allen et al. 2018
- Cain et al. 2019
- Lynch et al. 2020
- Directive (EU) 2023/958
- Rogelj and Schleussner 2019
- IPCC 2021, AR6 WG1