Article · 8 min read

Can AI fix one of aviation's dirtiest secrets? The UK's big contrail experiment, explained.

Published August 2026

The white lines behind planes are a bigger climate problem than you might think

Most people, if they think about aviation and climate change at all, think about the fuel that burns to keep a plane in the air. The carbon dioxide that comes out of a jet engine is real, and it matters. But there is another part of the problem that gets far less attention: those thin white streaks planes leave behind them in the sky.

Contrails are the thin white lines behind aircraft. They form when water vapour in jet exhaust freezes around soot particles into ice. Many vanish in minutes. In cold, humid air they persist, spread into cirrus cloud, and trap heat. That last part is the issue. When a contrail lingers, it behaves like a thin blanket over a patch of sky, stopping heat from escaping into space. Google estimates contrail warming is responsible for around one third of aviation's total climate impact. That is a striking figure. It means that if you want to cut the harm flying does to the atmosphere, contrails are not an afterthought; they are roughly as important as the CO2 itself.

Worth being cautious with that one-third figure, though: it comes from Google, a company with an obvious interest in making the problem sound large enough to justify its solution. Independent scientists broadly agree that contrails are a significant non-CO2 warming effect, but the precise share remains an area of active research and some debate.

There's more where this came from. New articles most weeks.Browse all articles →

What Operation Blue Skies is actually doing

A consortium composed of Google UK, NATS (an air navigation service provider), Contrails.org, Imperial College London, the University of Cambridge, and the Met Office has launched a 30-month, £5 million initiative called Operation Blue Skies. The project will test whether AI can predict where climate-warming contrails are likely to form and advise aircraft to alter their altitude to avoid them.

The geography matters here. Operation Blue Skies is the first state-backed effort to manage contrail avoidance across an entire oceanic airspace, according to Google. The programme centres on Shanwick Oceanic airspace, which spans the eastern half of the North Atlantic corridor and is responsible for roughly 5% of global contrail warming. Shanwick is the patch of sky above the northeast Atlantic where hundreds of long-haul flights between Europe and North America cross every day, at altitudes where the air is frequently cold and humid enough to generate persistent contrails.

The mechanics of what will actually happen in the trial are simpler than you might expect. Predictions will be provided to the UK's NATS, allowing air-traffic controllers to recommend relatively small altitude changes, typically around 2,000 feet, using existing procedures. Controllers will primarily request vertical adjustments of roughly 2,000 feet rather than rerouting planes laterally, a manoeuvre pilots already perform regularly to sidestep turbulence, according to the BBC. The idea is that contrail formation is sensitive to altitude: a zone of cold, moist air at 35,000 feet might not exist at 33,000 feet. A small shift, if timed right, could mean the plane passes through air that simply will not produce a lingering contrail.

Using satellite images and flight data, Google will pinpoint where aircraft have generated contrails in the past, then layer those findings against weather patterns to predict where new contrails are likely to form. NATS will then receive those forecasts so its controllers can steer pilots clear of the identified zones.

Around 10,000 flights are expected to cross the selected airspace during evenings and nights in the winters of 2026-27 and 2027-28. Not all of them will be rerouted. That 1% to 5% would be enough to demonstrate a statistically significant reduction in contrail formation on an airspace scale, said Paul Hodgson, Google's technical lead for Operation Blue Skies.

Why evenings and nights? The trial focuses on evening and night flights. Those produce the strongest warming, because daytime contrails also reflect some sunlight back to space. During the day, contrails have a small cooling effect that partially offsets their heat-trapping one. At night, there is no sunlight to reflect, so the warming effect is unambiguous.

This is not a brand new idea

It would be easy to read the press release and assume this came out of nowhere. It did not. Individual airlines have tested shifting altitudes to avoid contrail regions. Earlier this year, Google and American Airlines announced that the airline significantly reduced the climate impact of some flights using Google's AI-based forecasting tool to help prevent contrails. This new effort expands on individual airline trials to try mitigating contrails across an entire flight corridor.

The scientific idea, that persistent contrails contribute meaningfully to warming and that avoiding certain altitudes could reduce them, has been around for decades. The challenge has always been operational: how do you predict, reliably and in advance, exactly where the dangerous air will be? Weather at cruising altitude is not perfectly predictable. A bad forecast could send a plane to an altitude where it makes no contrail, burning slightly more fuel in the process, or send it to exactly the wrong spot. The AI piece is specifically about making those predictions accurate enough to be worth acting on at scale.

The case for the trial

The strongest argument in favour of Operation Blue Skies is the cost-effectiveness angle. Research suggests that avoiding contrails could be one of the most cost-effective ways to reduce aviation's climate impact. The intervention, if it works, does not require new aircraft, new fuels, or new engines. It requires a forecast and a radio call. The fuel penalty for a 2,000-foot deviation is small. If the warming reduction is even a fraction of what the optimistic models suggest, the ratio of benefit to cost looks very attractive compared with most other climate interventions in aviation.

There is also something genuinely useful about the trial's independent verification structure. Independent verification of the project's climate impact will be conducted by Imperial College London and the University of Cambridge. That matters. Having universities with no commercial stake in the outcome measure whether the thing actually works is the right way to do this.

To ensure public funding directly advances sovereign scientific capabilities, 100% of the government grant funding is allocated directly to academic, non-profit, and aviation partners. Google UK is participating entirely on a pro-bono basis, contributing £1.4 million in-kind through AI research expertise, engineering time, and high-performance computing infrastructure.

The questions worth asking

None of this means the trial is beyond scrutiny. A few things are worth keeping in mind.

Google's AI is doing the forecasting, and Google is not a neutral party. The company has a significant business interest in demonstrating that its AI tools can solve real-world problems. The trial has independent academic partners, which is a meaningful safeguard, but the forecasting model itself is Google's. Exactly how it works, and how confident anyone should be in its predictions before the trial runs, is not something the public can currently verify. Take any single-company research claim with a pinch of salt until it's independently replicated. That applies here too.

The fuel cost is small but real. Diverting a plane up or down 2,000 feet burns slightly more fuel, particularly when climbing. If the AI forecast is wrong, that fuel was spent for nothing. At the scale of hundreds of flights over two winters, the total extra emissions from bad predictions are unlikely to outweigh the potential contrail savings, but it is a real consideration and the trial ought to be transparent about measuring it.

Contrail science is less settled than the press materials imply. The one-third figure circulates widely, but estimates of non-CO2 aviation warming vary considerably depending on the method and assumptions used. Some researchers argue the figure is in the right ballpark; others think it may be overstated for certain flight routes and seasons. The scientific confidence is higher for the qualitative claim (persistent contrails do trap heat) than for any specific quantitative share.

Scaling this would require international buy-in. Shanwick airspace is one corridor. If the trial works, applying the approach globally would mean convincing dozens of national aviation authorities, hundreds of airlines, and multiple competing air traffic control systems to use compatible forecasting tools. That is a long road from a two-winter UK trial.

What to watch for

The trial will examine whether targeted interventions can significantly reduce contrail formation without noticeable disruption for passengers, with findings expected to be published in academic literature for wider scientific scrutiny. When those papers appear, the key things to look for are: how accurate were the AI forecasts compared with where contrails actually formed; what was the measured fuel penalty; and did the independent verifiers find a statistically meaningful reduction in contrail warming in the test corridor versus comparable nights when no deviations were made.

If the answers to those questions are encouraging, Operation Blue Skies could become a template. Aviation accounts for a meaningful slice of global warming, and the sector has very few quick wins available to it. An AI-assisted, operationally simple intervention that does not require replacing any aircraft would be a genuine step forward, not a complete solution, but a step.

If the forecasts turn out to be too unreliable, or the measured climate benefit too small, the honest conclusion would be that the idea still needs more science before it becomes policy. Either result, published openly and scrutinised by independent researchers, would be a useful contribution. That is how a trial is supposed to work.

From Telltale

Keep reading

If this one was useful, there's plenty more on the site. Pieces on how AI works, plus coverage of AI news, the downsides included. All free to read, no account needed.

See all articles →

References

Published August 2026 · telltale-ai.com
All articles · Privacy · Terms