The white lines planes leave in the sky do not always disappear quickly. Some turn into clouds of ice crystals called condensation trails, or contrails, which can trap heat and increase aviation’s climate impact. Google estimates that these trails account for nearly one-third of the sector’s total climate impact.
Now, Google Research and a British consortium are launching Operation Blue Skies, the first government-backed trial to reduce contrail formation across an entire oceanic air corridor.
How AI helps avoid condensation trails
Contrails form when water vapor from engines condenses around soot particles in very cold, humid areas of the atmosphere. However, not every flight produces persistent trails, and not all of them cause the same amount of warming.
The strategy is to identify in advance the atmospheric regions where an aircraft could create an especially long-lasting trail. To do this, AI models combine weather forecasts, flight data, satellite images, and atmospheric conditions.
With this information, flight crews and air traffic controllers can make small adjustments to a route or altitude. The goal is not to radically change flights, but to divert them slightly when they pass through an area especially prone to contrail formation.
AI does not eliminate contrails on its own. It turns complex atmospheric data into operational decisions that can be applied during a real flight.
Operation Blue Skies tests the solution at scale
Until now, trials have focused on individual flights or airlines. Operation Blue Skies aims to demonstrate that mitigation can be coordinated across an entire air corridor, where aircraft from multiple companies and countries share the same airspace.
The trial will take place in Shanwick, the eastern half of the North Atlantic corridor under the United Kingdom’s control. This area accounts for approximately 5% of global warming associated with contrails.
The program will run for 30 months and include two operational trials lasting around four months each. During the trial hours, approximately 10,000 flights from hundreds of cities will pass through Shanwick airspace each year.
Only a small proportion of those aircraft will need to divert. They will be the flights predicted to pass through areas with a high probability of producing strongly warming trails.
A technical system with independent verification
The project does not rely solely on an AI model. For an intervention like this to be reliable, it must integrate prediction, aviation safety, operational coordination, and follow-up impact measurement.
The system includes several components:
- Google models: predict where persistent contrails could form and help select alternative routes.
- Machine learning and satellite images: make it possible to track the trails and verify whether the changes actually reduced their formation.
- Weather forecasts: the Met Office will develop a new British capability to predict conditions favorable to contrails.
- Air traffic management: NATS will coordinate operations, safety assessments, and controller training.
- Scientific evaluation: Contrails.org, Imperial College London, and the University of Cambridge will analyze the results and provide independent verification.
This approach matters because predicting a trail is not enough. It is also necessary to confirm that the diversion was safe, that it did not create disproportionate operational impacts, and that the climate reduction actually occurred.
What changes for airlines and passengers?
For an airline, avoiding contrails could mean slightly changing the altitude or path of some flights. Those decisions must take fuel consumption, travel time, airspace congestion, safety, and possible effects on other aircraft into account.
That is where one of the central technical challenges appears: a diversion may increase fuel consumption, while avoiding a persistent trail could deliver a greater climate benefit. The models must compare both effects, rather than simply look for the shortest route.
The trial will also make it possible to assess inference latency—that is, how long the system takes to turn new weather and traffic data into a useful recommendation. In an operational environment, an accurate forecast that arrives too late is no longer practical.
A consortium turning research into operations
Operation Blue Skies brings together technology companies, air navigation authorities, universities, and scientific organizations from the United Kingdom. The collaboration aims to address a problem that no single institution could tackle on its own.
Google UK will participate at no cost and contribute £1.4 million toward AI research, engineering, and high-performance computing infrastructure. The program also has funding from its industry partners and the Department for Transport through the ATI program.
According to the announcement, 100% of the government grant will go directly to academic partners, nonprofit organizations, and aviation entities. The intention is for public funding to strengthen the United Kingdom’s own scientific and operational capabilities.
A possible model for the future of aviation
Condensation trails are difficult to observe from the ground and complex to attribute to a specific flight. That is precisely why combining atmospheric models, machine learning, satellite images, and traffic control could be so relevant.
The value of Operation Blue Skies will not depend only on how many trails it manages to avoid. It must also demonstrate that the predictions are accurate enough, that the decisions can be integrated into daily operations, and that the climate benefit outweighs the additional costs of modifying some flights.
If the trial produces strong results, it could become a reference for designing mitigation systems in other air corridors. In this case, AI does not replace pilots or controllers: it gives them a more detailed view of an invisible phenomenon that is already occurring above our heads.
Original source
https://blog.google/innovation-and-ai/models-and-research/google-research/blue-skies