[Paper Review] Efficacy of Scalable Airline-led Contrail Avoidance

Tharun Sankar, Thomas Dean+15 more|arXiv (Cornell University)|Mar 6, 2026

Advanced Aircraft Design and Technologies0 citations

TL;DR

This paper reports a large-scale randomized controlled trial showing dispatcher-led contrail avoidance integrated into standard airline operations reduces observable contrails significantly without materially increasing fuel use.

ABSTRACT

Contrails account for a large portion of aviation's contribution to anthropogenic climate change. Navigational contrail avoidance is a promising solution to mitigate the warming caused by contrails. Prior trials testing navigational contrail avoidance have relied on bespoke integrations of contrail forecasts into airline operations. Here, we use a randomized control trial to test the feasibility of dispatcher-led contrail avoidance integrated into standard flight planning operations using a workflow that scales to an airline's entire network. We validated the efficacy of this intervention using satellite imagery and an automated flight-contrail attribution algorithm. Using this system, we observed an 11.6% reduction in contrail formation rate for the 1232 flights marked as eligible for contrail avoidance (intent-to-treat) relative to the flights in the control group (p = 0.011). In the 112 flights that flew contrail avoidance as planned (per-protocol flights), we observed a 62.0% lower contrail formation rate relative to the flights in the control group (p < 0.001). No statistically significant difference in fuel usage was observed between the two groups.

Motivation & Objective

• Motivate contrail avoidance as a rapid climate mitigation Option in aviation.
• Test feasibility of dispatcher-led contrail avoidance integrated into standard flight planning across an airline network.
• Quantify observable contrail reduction and related climatological warming effects.
• Assess potential fuel usage impact and operational viability in real-world airline operations.

Proposed method

• Use an ML-based contrail forecast trained on GOES-East satellite data and ECMWF inputs to estimate contrail formation probability.
• Integrate contrail forecast into Flightkeys optimizer to generate contrail-optimized vs non-avoidance flight plans.
• Conduct a randomized assignment of city-pair routes to treatment vs control with a 10 t CO2e contrail-warming threshold.
• Validate observable contrails with an automated satellite attribution system (CoAtSaC) using ADS-B and wind advection data.
• Perform stratified permutation tests and bootstrap resampling to assess statistical significance and confidence intervals.

Figure 1 : Eastbound flight routes considered for this trial. The displayed routes are great circle paths between each city pair and do not necessarily reflect the actual paths that were flown. A full list of city pairs included in the trial and their applicable time periods is provided in Table 6 .

Experimental results

Research questions

• RQ1 Does dispatcher-led contrail avoidance integrated into standard flight planning reduce observable contrail formation compared with control flights?
• RQ2 What is the magnitude of contrail warming reduction associated with contrail-optimized flight plans in a real-world airline network?
• RQ3 Is there a measurable difference in fuel usage between contrail-avoidance and non-avoidance operations after adjusting for aircraft type?
• RQ4 How does dispatcher engagement (intent-to-treat vs per-protocol) affect observed contrail reductions?
• RQ5 Are satellite-derived contrail observations well-calibrated with forecast probabilities across treatment and control groups?

Key findings

• Observed contrail formation rate was reduced by 11.6% in the intent-to-treat L1 group relative to control (p = 0.011).
• Per-protocol groups showed larger reductions: L2 reduced by 36.4% (p = 0.002) and L3 by 64.5% (p < 0.001) relative to control.
• Climatological contrail warming declined correspondingly: L1 reduced by 13.8% (p = 0.006), L2 by 37.1% (p < 0.001), and L3 by 69.6% (p < 0.001).
• Adjusted fuel usage differences between treatment and control were small; L1 showed a -0.55% change (p = 0.044) after aircraft-type adjustment, with L2 and L3 not statistically significant.
• Counterfactual analysis using minimum-cost non-avoidance plans yielded expected contrail distance reductions of 25.0% (L2) and 49.2% (L3) against counterfactual plans, with overlapping confidence intervals indicating statistical significance.

((a)) Normalized fuel usage before applying the aircraft type adjustment.