Retrieval of cloud thermodynamic phase partitioning from multi-angle polarimetric imaging of Arctic mixed-phase clouds

Abstract. Mixed-phase clouds are frequently observed in the Arctic and still not well represented in climate and general circulation models. The spatial distribution of cloud thermodynamic phase and its partitioning are important quantities since they affect the radiative effect of clouds as well as cloud life time. In this work, a new quantitative retrieval method of cloud thermodynamic phase partitioning based on multi-angle polarimetry is presented. The polarization signal is sensitive to cloud thermodynamic phase since liquid water and ice have different shapes and different optical properties. The basic idea of the retrieval is to fit simulations obtained from a forward operator to measurements in the cloudbow range between 135° and 165° scattering angle and the slope range between 60° and 110° to determine a quantitative ice fraction. Either plane-parallel clouds or three-dimensional (half-spherical) clouds are assumed in the simulations. The retrieval uses data measured by the polarization-resolving cameras of the specMACS instrument and provides two-dimensional fields of cloud thermodynamic phase partitioning with a high spatial resolution of about 100 m. The retrieval was validated using synthetic data. 3D radiative transfer simulations were performed for different idealized cloud cases as well as for a realistic field of low-level Arctic mixed-phase clouds. In addition, it was applied to measurements taken during the HALO-(AC)3 campaign. As the retrieval is based on polarization it is sensitive to cloud top. The retrieved ice fraction is here defined as the ratio of the ice optical thickness to the liquid plus ice optical thickness and corresponds to the mean ice fraction of the uppermost cloud layer from cloud top to an optical thickness of about 1 to 2 depending on the solar zenith angle.