Google AI-based prediction study with American Airlines shows 54% reduction in aircraft contrails

A study led by Google Research and Breakthrough Energy, together with support from American Airlines, has found that by using artificial intelligence (AI) based predictions to make small modifications to routes that were projected to create contrails, their formation could be reduced by 54%, although avoidance burned 2% additional fuel. Other recent studies have shown that only a small percentage of flights need to be adjusted to avoid the majority of contrail warming. Therefore, say the researchers, the total fuel impact could be as low as 0.3% across an airline’s flights, suggesting contrails could be avoided at scale for around $5-25/ton CO2e using the predictions, making it a cost-effective warming reduction measure that could be improved still further with enhancements. A small group of American pilots flew 70 test flights over six months, after which Google Research analysed satellite imagery.

Cirrus clouds created by aircraft contrails are said to account for around 35% of aviation’s global warming impact During the day, the clouds reflect back incoming sunlight and warmth but nighttime contrails trap heat and a net warming effect takes place. However, contrails only form in ice-supersaturated air below a critical temperature threshold, within areas tending to be tens to hundreds of kilometres in the horizontal and only a few hundred metres in the vertical extent.

“The challenge is knowing where these regions will occur,” said researchers from Google. “By combining massive quantities of weather data, satellite data and flight data, AI can create state-of-the-art predictions of when and where contrails are likely to be formed. Pilots and dispatchers can then use this information to adjust the altitudes of their flights.”

The research team developed an approach to leverage and analyse satellite imagery, and spent hundreds of hours labelling tens of thousands of contrail images collected by the geostationary satellite, GOES-16. Labelled data was used to train a computer vision model to detect contrails when they are formed. The detection model can identify contrails that are visible within GOES imagery in a matter of 30 minutes, say the researchers, who then combined the computer vision model with large-scale weather data, flight data and satellite images to develop a more accurate prediction model.

The contrail-likely zone predictions were integrated into the tablets American’s pilots used in flight, so they could make real-time adjustments in altitude to avoid creating contrails, just as they would do to avoid turbulence. The prediction model’s performance was then evaluated using satellite imagery to compare the number of contrails produced in flights where pilots used the predictions against where they were not.

“We now have the first proof point that commercial flights can use these predictions to avoid contrails, as verified in satellite imagery,” said Juliet Rothenberg, Head of Product for Climate AI at Google Research. “We’re grateful for our partnership with American Airlines and Breakthrough Energy – together we’ve taken a significant step towards understanding a high-potential climate solution.”

Google Research said it planned to extend its models to geostationary satellites over Europe, Africa, the Indian Ocean, East Asia and Western Australia.

“The results from this small-scale test are encouraging and, while there are more questions to answer about how to operationalise contrails avoidance across our industry, we’re excited to have played a role in establishing their first proof point,” said Jill Blickstein, VP Sustainability at American. “And we’re looking forward to sharing what we learned with stakeholders in the aviation industry and beyond.”          

Added Marc Shapiro, Director of Breakthrough Energy Contrails: “Avoiding contrails might be one of the best ways to limit aviation’s climate impact, and now we have a clear demonstration that it’s possible to do so. This study is a great example of what happens when creative, ambitious organisations work together to better understand and solve a tough problem.”

The study follows the setting up of a cross-sector task force last year of aviation industry, tech sector and academic leaders to explore opportunities to address the warming impact of certain contrails. Assembled by RMI and Breakthrough Energy, the Contrail Impact Task Force includes airlines Alaska, American, Southwest, United and Virgin Atlantic, as well as Airbus, Boeing, Flightkeys, Google Research and Imperial College London.

It aims to reduce the climate impact from contrails by:

• Sharing and expanding on the latest science on the climate impact of contrails;
• Developing actionable strategies to avoid warming contrails;
• Analysing the operational and financial challenges of implementing potential solutions; and
• Establishing a roadmap for implementation and validation of contrail mitigation tools.

“The creation of the task force is a significant step in fostering the industry’s climate action. As a cross-industry stakeholder group, it demonstrates leadership and continued commitment to reduce aviation’s climate impact,” said an RMI blog. “The exact contribution of contrails to climate warming may still be uncertain, but a collaborative approach can serve as a catalyst for developing a greater understanding of contrail science, prediction and verification tools, and mitigation opportunities.”

Top image: Google AI detecting contrails over the United States, based on satellite imagery

Christopher Surgenor

Editor

• Linkedin