A Model Intercomparison Study of Mixed‐Phase Clouds in a Laboratory Chamber

Abstract Supercooled liquid water often persists in mixed‐phase clouds for hours to days, even though a simple, well‐mixed Wegener‐Bergeron‐Findeisen model predicts rapid depletion of liquid water in the presence of ice. The persistence is well documented, but the relative importance of the maintaining mechanisms remains uncertain. This model intercomparison study addresses this issue by comparing results from 10 model configurations against observations from the Pi Cloud Chamber at Michigan Technological University, which produces cloudy Rayleigh‐Bénard convection. To simplify the problem, we exclude ice‐nucleating particle processes and assume that the ice crystals are spherical. Each model was first tuned to reproduce the observed steady liquid‐cloud state and then run with ice injected at a range of rates. The evolution of spherical ice crystals is consistent across models, but differences emerge in the liquid phase. Most models that simulate the full chamber tend to retain liquid droplets for the entire simulation, because persistent supersaturation near the bottom activates new droplets that replace those that mix into the core and evaporate. In contrast, models that assume a well‐mixed domain, exclude the near‐wall region, or use coarse resolution do not sustain droplets and undergo complete glaciation. Models that use Lagrangian microphysics expose particles to greater supersaturation variability than bin schemes. Overall, these results demonstrate that liquid persistence in mixed‐phase clouds is highly sensitive to the representation of spatial heterogeneity and mixing, and provide guidance for improving microphysical parameterizations in atmospheric models.