On the Role of Macrophysics and Microphysics in Km-Scale Simulations of Mixed-Phase Clouds During Cold Air Outbreaks

Regional atmospheric models struggle to maintain supercooled liquid in mixed-phase clouds during polar cold-air outbreaks (CAOs). Previous studies focused on the parameterization of aerosol, microphysics and turbulence to understand the origin of this widespread model bias. This study investigates the role of macrophysics parameterizations (MacP) in the simulation of mixed-phase clouds. Km-scale simulations are performed for a large number of CAO cases over Norway, for which continuous ground observations were collected at one site over 6 months. We use a novel analysis that attributes the cloud-radiative errors to deficiencies in specific cloud regimes. We show that the MacP matters for cloud-radiative effects in CAOs, but that it is probably not the primary cause of the lack of liquid water in simulated mixed-phase clouds. Of all the MacP sensitivities explored in this study, the prognostic representation of both liquid and ice shows most promise in increasing the liquid water path. A newly proposed hybrid MacP with prognostic frozen and diagnostic liquid cloud fraction reproduces some of the benefits of the prognostic scheme at reduced cost and complexity. The two-moment microphysics scheme in this study produces too large precipitation particles. Reducing the snow deposition rate decreases the precipitation particle sizes and largely improves the liquid water path. Simulations are less sensitive to reduced riming rates. Cloud droplets and ice crystals often coexist in polar clouds, even at temperatures well below the freezing level. These clouds play an important role in the climate system, since they reflect more sunlight back to space than completely frozen clouds. Therefore it is important that climate and weather forecasting models are thoroughly evaluated in their ability to produce these so-called mixed-phase clouds. This study compares simulations of polar clouds with detailed cloud observations over Norway. We confirm that simulations produce clouds that are nearly completely frozen, while observations clearly show both liquid and frozen clouds. We test sensitivities of the simulations to specific model components to assess potential avenues for model improvement. Of all the tests performed, the ones affecting the simulated snowflake sizes seem most promising in allowing both liquid and frozen cloud to coexist.