Cathay Pacific and Google Are Testing AI-Guided Contrail Avoidance

Persistent contrails are a difficult aviation-climate problem because the conditions that create them can occur in narrow layers of the atmosphere. Cathay Pacific and Google are testing whether better forecasts can help aircraft avoid those layers through small changes in cruise altitude. The aim is to reduce a near-term warming effect using today’s aircraft and fuel systems—not to replace the need to cut aviation’s carbon dioxide emissions. 48

What the trial does

Google’s contrail forecasting tools combine large-scale weather data, flight data and satellite imagery with computer vision. They produce forecasts of contrail-likely areas and their potential warming impact for the next 48 hours. 17

The system uses two linked models:

• A machine-learning model identifies zones where contrails are likely to form.
• A physics-based Contrail Cirrus Prediction model, or CoCiP, estimates their warming effect. 19

That information can be incorporated into flight planning. If a flight is expected to pass through a thin, humid layer conducive to persistent contrails, dispatchers and pilots can consider flying slightly above or below it, subject to normal safety, air-traffic-control and fuel requirements. 1721

This is not a claim that an aircraft can simply eliminate every contrail. The practical question is whether it can selectively avoid the contrails expected to have the greatest warming impact without causing disproportionate extra fuel burn or operational disruption.

Why small altitude changes can matter

Some contrails dissipate quickly. The concern is the subset that persists and spreads into cirrus-like cloud, which can trap heat. The weather conditions involved are often localized vertically, so a modest cruise-altitude adjustment may be enough to avoid them. 421

That makes contrail avoidance distinct from a full route diversion. Rather than redesigning an entire journey, the trial tests targeted vertical changes when the forecast identifies a high-value avoidance opportunity.

Independent commercial-flight research supports the basic operational premise. In one per-flight avoidance test, the treatment group produced four detectable contrails versus 11 in the control group—a 63.6% reduction in detectable contrails. 1

What Cathay Pacific has reported so far

Cathay Pacific began its operational trial with Google in late 2025. Google says the early phase covered more than 80 flights and achieved a roughly 40% estimated reduction in contrail warming impact when flights followed contrail-avoidance routes. Cathay is Google’s first commercial airline partner in Asia for this work. 8

The estimate should be read carefully: it is an early result from the partners’ trial, not a final, independently replicated measure of fleet-wide climate benefit. The value of the next phase is to test the forecast and the operational process across a wider set of conditions and flights. 48

Reporting on the initial work identifies the Hong Kong–Singapore corridor as a major contributor to the observed reductions. 6 Google and Cathay also position the work as a test of ultra-long-haul flying, where decisions must account for long sectors, changing weather and tighter fuel-planning constraints. 38

The expansion with Contrails.org

The partners are expanding the program across Cathay Pacific’s network to evaluate operational feasibility and contribute data to contrail research. 8

The next phase is expected to involve Contrails.org and target ultra-long-haul Asian and transpacific flights. It is intended to include a randomized controlled trial, which can more rigorously compare flights using avoidance guidance with comparable flights that do not. 520

That matters because a successful system must work beyond a small set of favorable flights. It needs forecasts accurate enough for dispatch, flight changes that can be accommodated safely, and credible verification of whether contrails were actually avoided.

Why Asia-Pacific is an important test case

Much prior operational work on AI-assisted contrail avoidance has been conducted in the United States and over the North Atlantic. Google’s earlier work with American Airlines included 70 test flights using AI-based predictions to avoid altitudes likely to produce contrails. 27

Cathay’s program brings the approach into Asia-Pacific and onto ultra-long-haul operations. This expands the evidence base to a different operating environment, with its own weather patterns, airspace constraints and route structures. The expansion is therefore as much a validation exercise as a deployment effort: it tests whether forecasting and flight-planning methods can perform reliably across a broader network. 820

A complement to SAF and newer aircraft—not a replacement

Contrail avoidance has an immediate practical attraction: it can be tested with the existing fleet. It primarily requires better prediction, flight-planning integration and operational coordination, rather than a new aircraft type or a new fuel-distribution system.

Research reviewed by ICAO indicates that selectively targeting the most harmful contrails can produce substantial reductions with relatively small fuel penalties in modeled scenarios. One cited airline study found a 73% reduction in contrail climate forcing at a 0.11% extra-fuel cost, while another modeling study found that diverting a small fraction of flights could materially reduce contrail energy forcing with a very small aggregate fuel increase. 215

Sustainable aviation fuel and more efficient aircraft still matter because they address the long-lived CO2 emissions from burning fuel. SAF can also reduce soot emissions: Contrails.org cites commercial-flight studies of up-to-50% SAF blends that found 50% to 70% lower soot emissions, alongside reductions in contrail ice-particle size and concentration. 12

But lower soot does not remove the need for accurate contrail forecasts, and route or altitude management does not remove CO2 emissions. The strongest case for the Cathay-Google approach is as a complementary measure: use targeted operational changes to reduce a potentially avoidable non-CO2 warming effect while aviation works on fuel, aircraft and infrastructure transitions.

The real test: operational scale

The reported 40% result is an encouraging signal, not a settled answer. Scaling it will depend on forecast quality, satellite-based verification, approval of altitude changes, fuel trade-offs and the ability to integrate recommendations into normal airline and air-traffic workflows. 817

If those conditions can be met, contrail avoidance could offer airlines a comparatively low-disruption way to reduce part of aviation’s climate impact now—while the longer work of deploying cleaner fuels and fleets continues.