Impact of Physical and Numerical Representation of Ice Cloud Microphysics on the Simulated Ice Clouds and Their Radiative Effects in the E3SM Atmosphere Model

Ice clouds play an important role in regulating the Earth’s radiative budget and influencing the hydrological cycle. However, the simulated radiative effects of ice clouds in global climate models are not well quantified and constrained. Apart from the dynamical influence at various scales, the simulated radiative effect of ice clouds largely depends on the physical and numerical representation of ice microphysical processes in the model. Our previous work has shown that: a) the artificial treatment of the ice-to-snow conversion process has a large impact on the net radiative balance at the top-of-the-atmosphere; b) less careful treatment of the numerical coupling of competing and compensating processes, such as the ice nucleation, ice depositional growth/sublimation, and source/sink of water vapor, can lead to inaccurate solutions. To address these problems and improve the atmosphere component of E3SM (Energy Exascale Earth System Model), we have implemented the single-ice-category Predicted Particle Properties (P3) scheme in the model. The scheme was revised for use at coarser resolutions and to better represent the interaction between processes. We performed single column simulations and global nudged simulations using both P3 and MG2 (the original microphysics scheme) in E3SM and investigated the ice cloud radiative effects. Results show that E3SM-P3 performs well in simulating macrophysical and microphysical properties of ice clouds. The differences in simulated ice cloud properties between E3SM-P3 and E3SM-MG2 are mainly caused by changes in the simulated ice deposition (including Bergeron process), sublimation, and sedimentation processes arising from the use of a single-ice-category treatment (all processes) and consideration of the capacitance for vapor deposition over non-spherical ice crystals (deposition/sublimation). We also find significant differences between E3SM-P3 and E3SM-MG2 in the simulated frequency distribution of precipitation and the ratio between stratiform and convective precipitation. Preliminary analysis indicates that these changes are linked to the differences in the simulated diabatic heating in the two simulations. The model sensitivity to using different numerical methods for coupling the microphysical/sedimentation processes and to using a three-moment configuration of P3 will also be discussed