Three in four flights leave nothing lasting
In 2019, 24% of flights formed persistent contrails. Just 2.7% of flights accounted for 80% of the annual warming.
Share of all flights (top) and where the contrail warming comes from (bottom), 2019
Cold, ice-saturated air, and ideally no sun
A contrail lasts only where the air is already supersaturated with ice. Those contrails averaged 2.4 hours. By day they also reflect sunlight; after dark they only trap heat.
When the strongest contrails tend to form, local time
Schematic of a tendency, not a measured intensity.
Contrail net forcing by season, 2019, mW/m²
In winter, 5.9% of distance flown forms lasting contrails (summer 3.8%), and they last 2.8 hours (summer 2.4), even though traffic is 14% below its summer peak.
Air travel grew nearly fourteenfold. The pandemic showed how fast contrails respond
Contrails last hours, not centuries, so when flying dropped in 2020 the forcing dropped with it: distance flown fell 43%, contrail net forcing fell 56%.
World air passengers per year, billions
Global contrail net forcing, mW/m², change from 2019
From 2006 to 2019, contrail forcing grew 2.7% a year while distance flown grew 3.6% a year.
Possibly aviation's largest warming term, and heaped over northern skies
Contrail cirrus outranks aviation's CO₂ in the leading 2018 tally, with a much wider error bar. Forcing is many times the world average over Europe and the USA.
Aviation's warming by source, 2018, ERF, mW/m² (whisker: 5–95% range)
Teoh's 2019 estimate of 62.1 mW/m² net RF is 44% below the older 111 mW/m² RF estimate, and sensitivity tests span 34.8–74.8. The number is still moving.
Contrail net forcing over each region, 2019, mW/m²
The reroute works on flights that actually fly it
Recent trials show big cuts on flights that followed the plan, and much smaller ones across the whole eligible group. Getting plans flown looks like the next hurdle.
Reported reduction in contrails or their climate impact (metrics differ by trial)
Cheap to try, harder to prove
The estimated price is low, but the extra fuel adds CO₂ that lasts far longer than any contrail, so the forecasts have to be right.
per ticket on Paris–New York, fuel and technology included.T&E 2024, advocacy estimate
extra fuel across the fleet in a year; 5% or less on rerouted flights.T&E 2024
later: warming due in 2048 arrives in 2061, if contrail warming halves by 2040.Klima Consulting 2026, modelled scenario
mW/m²: the 2019 forcing range under different model assumptions.Teoh et al. 2024, sensitivity test
Operation Blue Skies
A UK-backed trial will have controllers move some North Atlantic flights by small altitude steps, to test whether avoidance can be measured at airspace scale. No results yet.
- Flights through Shanwick in test periods~10,000
- Rerouted1–5%
- Maximum altitude change2,000 ft
- Test days per winter20–40
- Shanwick's share of contrail warming~5%
- Programme cost£5M
A rare climate fix that could work within hours
CO₂ from a flight lingers for centuries. A contrail avoided today is warming gone today, which makes this worth testing properly, alongside cutting CO₂ and not in place of it.
- Look upA trail that fades fast is not the kind in question. One that lingers and spreads into a thin sheet is.
- Ask your airlineDoes it plan around contrails, how many flagged flights actually reroute, and does it publish the results?
- Watch the follow-through rateTrial headlines quote the best subset. The figure across all eligible flights is the one that scales.
- Keep CO₂ in the frameSupport rules that count both effects, so a reroute is only rewarded when the climate comes out ahead.
Sources & notes
- Teoh R. et al.. "Global aviation contrail climate effects from 2019 to 2021," Atmos. Chem. Phys. 24, 6071 (2024). Flight shares, net RF by year and region, lifetimes, seasons, time of day, sensitivity range. All contrail forcing values are modelled (CoCiP on ADS-B flight tracks), not directly measured.
- Lee D.S. et al.. "The contribution of global aviation to anthropogenic climate forcing for 2000 to 2018," Atmos. Environ. 244 (2021). Effective radiative forcing, 2018, with 5–95% range for contrail cirrus.
- World Bank / ICAO. Air transport, passengers carried (IS.AIR.PSGR), world, 1970–2023. 1972 is missing and is bridged by the line.
- Sankar T. et al.. "Efficacy of Scalable Airline-led Contrail Avoidance," arXiv 2603.06909 (2026), preprint. American Airlines transatlantic trial of 2,400+ flights, 2025. Metric: contrail formation rate. No significant fuel difference. City pairs were randomised; the 11.6% compares the 1,232 treatment flights with 1,172 controls. Flying the plan was voluntary, so the 62.0% subset could reflect selection, though the authors' counterfactual check suggests it was not driven by it.
- Air France. North Atlantic trial, Feb–Apr 2026, as reported by PlaneSight (Sep 2026). The airline's own analysis, not peer reviewed. Metric: contrail climate impact.
- Google Research, American Airlines, Breakthrough Energy. 2023 test of 70 flights. Metric: contrail distance. Company-reported and a small sample.
- Transport & Environment. Contrail avoidance cost study, Nov 2024. Estimate from an advocacy group; the under-€4 figure is its Paris–New York example.
- Boucher O. et al., Klima Consulting. "Defining a way forward for the science and technology of contrail mitigation," May 2026, commissioned by T&E. The 2048→2061 delay is one modelled scenario.
- Operation Blue Skies. AP via WTOP (Aug 18, 2026); Imperial College London (2026). Trial plan, no results yet.
Not comparable side by side: Lee reports effective radiative forcing (ERF); Teoh reports net radiative forcing (RF). Teoh's abstract rounds the flight share to 2%; its body gives 2.7% for 2019 and 1.7% for 2020–21. The strip uses 2019, where the 21.3% is derived as 24% minus 2.7%. "Warming" means annual contrail energy forcing, net of cooling contrails. Trial percentages use different metrics and are not directly comparable. Illustrative only: the 100-line sky graphic (schematic of 2 in 100) and the time-of-day strip, which shows when strong contrails tend to form, not a measured intensity.