Output from idealized simulations of a squall line are compared with in situ aircraft-based observations from the Deep Convective Clouds and Chemistry campaign. Relative humidity distributions around convection are compared between 1-Hz aircraft observations (≈250-m horizontal scale) and simulations using a double-moment bulk microphysics scheme at three horizontal grid spacings: Δx = 0.25, 1, and 4 km. The comparisons focus on the horizontal extent of ice supersaturated regions (ISSRs), the maximum and average relative humidity with respect to ice (RHi) in ISSRs, and the ice microphysical properties during cirrus cloud evolution, with simulations at 0.25 and 1 km providing better results than the 4-km simulation. Within the ISSRs, all the simulations represent the dominant contributions of water vapor horizontal heterogeneities to ISSR formation on average, but with larger variabilities in such contributions than the observations. The best results are produced by a Δx = 0.25-km simulation with the RHi threshold for initiating ice nucleation increased to 130%, which improves almost all the ISSR characteristics and allows for larger magnitude and frequency of ice supersaturation (ISS) > 8%. This simulation also allows more occurrences of clear-sky ISSRs and a higher spatial fraction of ISS for in-cloud conditions, which are consistent with the observations. These improvements are not reproduced by modifying other ice microphysical processes, such as a factor-of-2 reduction in the ice nuclei concentration; a factor-of-10 reduction in the vapor deposition rate; turning off heterogeneous contact and immersion freezing; or turning off homogeneous freezing of liquid water.
Minghui Diao, George Bryan, Hugh Morrison, and Jorgen Jensen. "Ice Nucleation Parameterization and Relative Humidity Distribution in Idealized Squall-Line Simulations" Journal of the Atmospheric Sciences (2017): 2761-2787. doi:10.1175/JAS-D-16-0356.1