An investigation of ice formation in mixed-phase Arctic boundary layer clouds observed on 7 May 1998 during SHEBA

In prior work using large-eddy simulations with fully-periodic boundary conditions and size-resolved microphysics with simplified spherical ice to represent observed rimed ice, we found that ice nuclei measured via aircraft with the Continuous Flow Diffusion Chamber (CFDC) were unable to explain the ice concentrations observed during the Mixed-Phase Cloud Experiment (M-PACE) even when large uncertainties associated with ice shattering were included. The latest advances in understanding of ice shattering uncertainties have not changed that conclusion, although it remains subject to the many other sources of uncertainty in model representation of ice and mixed-phase processes noted in that work. Here we present a follow-on study of ice formation in mixed-phase boundary-layer clouds observed over Arctic sea ice during the Surface Heat Budget of the Arctic (SHEBA) experiment based on a recent model intercomparison case study (see Morrison et al. SHEBA results at this conference). While the SHEBA measurements were taken more than 10 years ago, the data set still provides a unique opportunity that is rich in well documented and archived measurements and extensively studied conditions. The ice crystal properties are also relatively uniform in habit with negligible numbers of crystals larger than 3 mm in maximum dimension, which distinguishes this SHEBA case study from other, more recent case studies in which there is evidence of active riming or aggregation (e.g., M-PACE and ISDAC). Taking the SHEBA intercomparison set-up as a starting point, we first adjust the initial thermodynamic profile and large-scale forcing conditions in order to better match conditions when in situ ice size distribution data are available, during the final two hours of the 12-hour intercomparison time period, and we replace fixed surface fluxes with interactive fluxes from surface similarity. When we use an improved scheme to represent the properties of non-spherical ice, and we choose ice properties consistent with Cloud Particle Imager probe measurements (radiating assemblages of plates), we find that we are able to match observed mean droplet number concentrations and ice crystal size distributions (the latter at sizes larger than several hundred microns), as well as observed cloud radar reflectivities and Doppler velocities, but only when background ice nucleus concentrations are increased by a factor of 20 beyond the CFDC mean measured values. However, we are unable to reproduce the irregular horizontal variability in ice concentration on scales of order 10 km that is evident in the radar measurements. We investigate the ability of surface sources of ice nuclei consistent with CFDC observations during this case study (likely to be emanating from nearby sea ice leads, associated also with sensible and latent heat fluxes) to simulate the observed variability in cloud ice. M-PACE results also indicated a possible role for a surface source of ice nuclei (active above liquid saturation) consistent with both observations and model results. We place these results in the context of other possible ice formation pathways, as well as the limitations of large-eddy simulations to reproduce mesoscale variability.