Contrail avoidance gets a budget: a £5m trial to see if AI forecasts can move aircraft

A £5m UK trial will test whether AI can predict where warming contrails are likely to form and route aircraft away from them ^[1], with a 30-month project called Operation Blue Skies bringing together Google, the UK government, the Met Office, NATS and academic researchers ^[2]. The significance is not the science, which has been argued over for years, but the plumbing: previous work of this kind has been run at airline level ^[17], while this attempt pushes the forecast into an air navigation service provider's existing procedures ^[7].

The physics is unglamorous. Hot engine exhaust meets cold high-altitude air and forms ice crystals; most trails vanish, but in cold, humid conditions they persist, spread and form cloud that traps heat which would otherwise escape ^[21]. Dr Paul Hodgson, Google's technical lead on the project, told the BBC that contrail warming is responsible for "something like a third" of all aviation climate warming, acknowledged uncertainty in the figure, and said scientists were confident it was "either quite a large problem or a very large problem" ^[4][5].

The trial targets Shanwick Oceanic airspace in the eastern North Atlantic, which the project puts at roughly 5% of global contrail warming ^[3]. Combining those two figures implies the target airspace accounts for on the order of 1.7% of aviation's total climate warming ^[1], which is a narrow slice, and the reason the design matters more than the headline. Google will use satellite imagery and flight data to find where aircraft have produced contrails before, combine that with weather patterns to forecast new ones, and pass the output to NATS controllers ^[6][7]. The intervention is mostly a 2,000ft altitude change, something already done routinely for turbulence and weather ^[8]. Ian Jopson, NATS director of sustainability, said it should be "pretty much business as usual using existing procedures" for North Atlantic operators ^[9].

Testing runs on selected evenings and nights in the winters of 2026-27 and 2027-28, when traffic is lower ^[10]. Around 10,000 flights are expected through the airspace in those windows, though only a fraction will be asked to move for contrail reasons ^[11]. That fraction is the whole experiment: it determines whether the result has enough statistical weight to justify a rule change.

On the trade-off, climbing or descending burns fuel ^[14]. Hodgson said the CO2 penalty in previous trials was "in the order of less than 1%" against a contrail cost of similar magnitude to all of aviation's CO2 ^[15], and modelling for this project puts a full deviation through Shanwick at about 100kg of extra fuel per affected aircraft ^[16]. Taken together, 100kg only stays under 1% for aircraft burning more than ten tonnes on the leg ^[2], so the penalty will look different on shorter sectors.

The money is worth reading closely. The Department for Transport is providing £2.65m and Google £1.4m in AI research, engineering and computing infrastructure ^[12], leaving roughly £950,000 of the stated £5m unattributed in the announcement ^[3]. Aviation Minister Keir Mather called the project "a world-first" and "British ingenuity leading the way" ^[13].

Watch the published results, which are due in academic literature and open to wider scrutiny ^[18], and watch the context: the same government backs UK airport expansion ^[19], and Google is among the firms spending billions on power-hungry data centres ^[20].