A Turbulence‐Informed Parameterization of Phase Partitioning in Stratiform Mixed‐Phase Clouds for the LMDZ Model

High‐latitude clouds, present over the Arctic Ocean, the Southern Ocean and the Antarctic continent, are very often mixed‐phase clouds (MPCs), that is, composed of both supercooled liquid droplets and ice crystals. Despite being essential for the climate of the poles, they remain a major modeling challenge for climate models. In this study, we present a new cloud phase partitioning parameterization developed for the LMDZ atmospheric model. This parameterization is based on the theory of the evolution of supersaturation in a turbulent environment and is inspired by previous theoretical and modeling works. This scheme completely abandons the standard temperature dependent phase partitioning used in the model to predict the amount of supercooled liquid water in clouds as a function of turbulent kinetic energy, resolved vertical velocity and pre‐existing ice crystal properties. This new scheme is evaluated over the Southern Ocean with observation from the MARCUS campaign and results show an improvement in the simulation of the cloud phase spatial variability. The sensitivity to the crystal number concentration, determined by a prescribed concentration of ice nucleating particles, is also assessed. A second evaluation is performed in the Arctic region with observations collected in mid‐level frontal clouds during the RALI‐Thinice campaign and a perturbed parameter ensemble experiment is conducted to assess the parametric sensitivity. The new scheme suppresses the systematic overestimation of liquid far from cloud top and shows the ability to simulate patches and thin layers of supercooled liquid water as commonly observed in polar frontal clouds.