Robust satellite-based tracking of contrail forming regions

Contrails are estimated to be one of the largest contributors to the aviation sector’s climate impacts. A potential mitigation approach is avoiding flying in regions where contrails form and persist by rerouting aircraft above or below these regions.

Implementing such contrail avoidance strategies requires accurately forecasting the location of contrail forming regions. Although models exist, their prediction ability is limited by uncertainties in local weather conditions and contrail modeling. Understanding how these limitations affect our ability to predict contrail formation at a regional scale is necessary to improve forecasting of contrail avoidance regions.

To address this, we develop an observational inventory of the evolution of contrail forming regions over the United States. By developing a deep-learning algorithm and ensemble Kalman filter, we generate robust contrail detections on geostationary satellite imagery at a 5-minute frequency. Observed contrail forming regions are tracked over their lifespan allowing for the derivation of properties such as lifetime of the region, or rates of formation of contrails. These observed properties are compared to contrail model outputs using numerical weather prediction data, as well as correlated to patterns such as flight traffic density or spatial extent of the region. We also investigate the variability in conditions across the United States which support contrail formation.

Direct comparison of model outputs to large-scale high temporal resolution imagery of contrail forming regions will inform our understanding of contrail formation regions by providing observational evidence as to when and why current predictions can succeed.