Predicting ice supersaturation for contrail avoidance: ensemble forecasting using ICON with two-moment ice microphysics

Abstract

Persistent contrails and contrail-induced cirrus clouds are considered the most significant non-CO₂ contributors to aviation's climate impact. These clouds primarily form in ice-supersaturated regions (ISSRs), defined by relative humidity over ice (RH_ice) exceeding 100 %. Reliable prediction of RH_ice in the upper troposphere and lower stratosphere allows mitigating their formation by re-routing flights. We implemented a two-moment cloud ice microphysics parameterization within a ten-member Ensemble Prediction System (EPS) using the global ICON (ICOsahedral Nonhydrostatic) model. RH_ice predictions were evaluated against radiosonde and aircraft observations from the Northern Hemisphere during 2024–2025. Treating ISSR prediction (RH_ice > 100 %) as a binary classification problem, we find that the probability of detection (POD) of ISSRs increases to 0.6 for the two-moment scheme (ICON 2-Mom), compared to 0.4 for the operational ICON with a one-moment ice microphysics scheme, while maintaining a low false positive rate (FPR < 0.1). Further evaluation of the ICON 2-Mom EPS using Receiver Operating Characteristic (ROC) analysis shows a POD of 0.8 for a decision model that requires at least three ensemble members to predict ISSR, with an FPR of 0.13. Additionally, we incorporate ensemble spread information to develop a meta-model that further reduces the FPR. Since June 2024, more than 100 flights have been rerouted based on ICON 2-Mom EPS predictions in a contrail avoidance trial, demonstrating the practical value of improved ISSR forecasts for climate-conscious aviation. This study highlights the significant potential of ensemble-based modeling for predicting ISSRs and RH_ice, supporting environmentally optimized flight planning and advancing applications in weather and climate science.