Enhanced Predictive Modeling and Validation of Persistent Contrails

We present recent work in modeling the impact of the aviation industry on contrail formation and its subsequent effect on global warming by utilizing the Contrail Cirrus Prediction Model (CoCiP) with comprehensive 2019 airline data. This approach offers an in-depth, fleet-level perspective, revealing the influence on global energy forcing due to various factors including, but not limited to aircraft market class, operator region, and origin-destination pairs. Our analysis extends to a comparative study with results derived from our internally developed machine-learning (ML) emulator, the Hybrid Contrail Prediction (HyCoP) modeling framework. The HyCoP framework is designed to harness the power of physics-based CoCiP simulation data alongside satellite image-based ground truth observations to accurately predict persistent contrail formation. Central to this innovative framework is the Bayesian Deep Neural Network (BDNN) classifier. HyCoP is being developed to integrate aircraft engine-specific features, enhancing the standard inputs traditionally used in the CoCiP model. Furthermore, we detail our rigorous validation efforts for these contrail models. We employ geostationary satellite (GOES) images and data from flight campaigns such as 2023 ecoDemonstrator and 2024 Contrail Optical Depth Experiment (CODEX) to ensure the accuracy and reliability of our predictions. Our presentation includes the current status of our comprehensive start-to-end validation pipeline, which detects and tracks contrails and attributes them to their originating aircraft. This holistic approach underscores our commitment to advancing the understanding of contrail impacts on climate change through cutting-edge predictive modeling and thorough validation methodologies.