How Operation Blue Skies will use AI to help planes avoid warming contrails

Operation Blue Skies is a UK-backed research and operations trial testing whether aircraft can reduce part of aviation’s climate impact by avoiding persistent contrails. The £5 million programme will run for 30 months in Shanwick Oceanic airspace, the eastern section of the North Atlantic corridor. Its organisers describe it as the first state-backed contrail-avoidance trial conducted at the scale of an entire oceanic airspace.

The project is not assuming that every flight should change course. Instead, it will test whether weather and AI forecasts can identify a relatively small number of flights and atmospheric locations where a modest altitude adjustment could prevent a contrail with a disproportionately large warming effect.

Why persistent contrails can warm the climate

Contrails form when water vapour in warm aircraft exhaust mixes with very cold air, producing clouds of ice crystals. Many disappear quickly. In sufficiently cold and humid conditions, however, a contrail can persist, spread and develop into wider ice cloud. These clouds can affect the amount of heat escaping from Earth’s atmosphere, making contrails an important part of aviation’s non-CO2 climate impact.

The effect varies with atmospheric conditions and time of day. Persistent contrails are especially relevant to evening and night operations because the resulting ice cloud can reduce the escape of infrared heat. That makes targeted avoidance potentially more useful than treating all contrails as equally harmful.

How the AI-guided flight changes are expected to work

Google’s forecasting work will combine satellite observations, historical flight and contrail data, and weather forecasts from the Met Office. The aim is to predict “ice-supersaturated” regions—parts of the upper atmosphere where persistent contrails are more likely to form.

The process is expected to work in three stages:

1. Forecast the risk: AI models identify areas and altitudes where a flight is likely to create a persistent contrail.
2. Assess an operational option: Controllers consider whether a small altitude change is safe and practical for a selected aircraft.
3. Check what happened: Satellite imagery and flight data are compared with the forecast to determine whether the aircraft avoided a persistent contrail and what the change cost in fuel.

NATS controllers are expected to direct selected aircraft to make relatively small climbs or descents—up to about 2,000 feet—rather than rerouting all transatlantic traffic. The operational trials are planned around winter periods, including evening and night flights when the expected climate value of avoidance is greatest.

The trial’s central question: does avoidance create a net climate benefit?

Changing altitude is not automatically a climate solution. A climb or descent can increase fuel consumption, which produces additional CO2 emissions. Operation Blue Skies is therefore designed to measure the trade-off between that fuel penalty and the warming avoided by preventing a persistent contrail.

A convincing result would need to show more than accurate-looking forecasts. Researchers will need to establish:

• how often the model correctly predicts persistent contrails;
• how many warming contrails are missed or falsely predicted;
• how much additional fuel the altitude changes consume;
• whether satellite observations confirm the predicted and avoided contrails; and
• which climate metric and time horizon are used to compare contrail warming with CO2 emissions.

The trial is consequently an experiment, not proof that the manoeuvres already deliver a net climate benefit. Contrail impacts are difficult to estimate, and the UK government’s review of aviation’s non-CO2 effects highlights the broader challenge of measuring them consistently.

Who is involved and how large is it?

The consortium brings together Google UK, the UK Department for Transport, NATS, the Met Office, Contrails.org, Imperial College London and the University of Cambridge. UK government support is provided through the Aerospace Technology Institute programme.

The programme’s significance comes from its operational scale. Previous work has included airline-led demonstrations and tools that detect contrails in satellite imagery, attribute them to individual flights and produce contrail forecasts. Google’s earlier research used computer vision trained on labelled satellite images to detect contrails and connect them with aircraft.

Blue Skies is intended to move beyond isolated flight experiments by testing forecast-led decisions within a busy oceanic airspace managed by air-traffic controllers. Shanwick airspace accounts for approximately 5% of estimated global contrail warming, according to the project’s organisers.

How the results should be independently checked

The programme says its methods and results are intended for publication. That is important because the strongest evidence will need to be reproducible and open to scrutiny rather than relying only on model projections or promotional estimates.

Academic partners can provide scientific review, but independent verification should also cover the operational accounting. The most useful final reporting would disclose model performance, observed contrail changes, flight-level fuel impacts, the climate assumptions used and the energy required to run the forecasting system.

That level of reporting would help separate three different claims: that AI can predict contrail-forming conditions, that controllers can use those predictions without disrupting safe operations, and that the resulting intervention reduces total climate impact.

What could change if the approach works?

If the trial demonstrates a reliable net benefit, contrail avoidance could become a relatively near-term way for aviation to address part of its non-CO2 climate impact. It would use existing aircraft and air-traffic systems rather than waiting for new fuels or wholesale fleet replacement.

It would not replace efforts to reduce CO2. Aircraft would still burn fuel, not every contrail would be avoidable, and altitude changes could sometimes increase emissions. The value of the approach would depend on targeting the right flights and atmospheric regions with enough accuracy to make the overall balance favourable.

The unresolved AI-energy question for Google

The project also raises a legitimate accounting question about the energy used by AI. Google reported that its data-centre electricity demand increased by 27% in 2024 while data-centre energy emissions fell by 12%, which it attributed to clean-energy procurement and efficiency measures.

Those figures do not by themselves show whether the Blue Skies system will produce a net climate benefit. A fair assessment would compare the forecasting system’s marginal computing energy and emissions with measured changes in aircraft fuel use and observed contrail warming.

Until those figures are published, the most accurate verdict is cautious: Operation Blue Skies is a significant test of whether AI can make contrail avoidance practical at airspace scale, but its climate payoff remains unproven.