Three-dimensional narrow cold-frontal rainband (NCFR) simulations are run with an anelastic time-dependent cloud model to investigate the impact of changes in the bulk microphysics parameterization, which adapts Kessler’s warm rain physics and the three-class ice phase of Rutledge and Hobbs while using a temperature-weighted diagnostic partitioning of cloud phase. Precipitation generally includes rain, graupel and snow, each assumed spherical with inverse exponential size spectra and also constant collection efficiencies for all accretion processes.

A control NCFR simulation and three sensitivity variations are run out to 168 min. The domain, Cartesian with periodic boundaries along-line and open aperiodic boundaries cross-line, is initialized with a sounding adapted from a vigorous wintertime California NCFR case, imposing a deep wavy-edged cold pool as forcing. One variation doubles the graupel fallspeeds for all diameters. Another replaces the spherical snow with columns, hexagonal plates and broad-branch ice crystals, assigning columns an inverse exponential length distribution and using variable collection efficiencies for snow riming. The remaining experiment modifies the capping algorithm that reduces hydrometeor sinks as needed to prevent overdrawing available supplies.

Each NCFR simulation spins up a strong quasi-steady rainband that propagates rapidly akin to a density current and has a vigorous jump-type updraft with an upshear tilt, despite marginal buoyancy and modest cloud top heights. Cloud is mostly liquid and contributes far less mass than precipitation, which is largely frozen. The control case produces a very narrow primary rainband closely trailed by a somewhat wider and weaker secondary band, while the hydrometeor fields consist of graupel, rain, snow, cloud water and cloud ice in decreasing order of abundance. Mature updraft speed is quasi-steady aside from a brief drop of ~20% in early mid-run. Among other salient findings:

1) Doubling the graupel fallspeeds reduces hydrometeor mass ~35% overall, and ~50% for graupel and snow. The secondary rainband is suppressed, and the width of the primary rainband more than doubles, although the mass and source/sinks of rain change little. Riming becomes a more important secondary graupel source, while melting dominates the graupel sinks more lopsidedly. The temporary updraft retrenchment starts earlier and is more gradual, although after early mid-run the updraft strength is nearly the same as in the control experiment.

2) Replacing exclusively spherical snow by the three aspherical habits affects airflow little but the microphysics considerably. Hydrometeor mass more than doubles for snow, while decreasing slightly for graupel and in toto. The primary rainband is enhanced, the trailing secondary one somewhat weakened. Accretion of cloud water is bolstered as a secondary rain source, while accretion by snow dominates evaporation as a rain sink, rather than nearly equaling it as in the control case. Among secondary processes, deposition is a much enlarged snow source, sublimation is a much enlarged snow sink, and accretion of rain by snow is enhanced as a graupel source.

3) Modifying the capping algorithm for hydrometeor sinks also produces several significant microphysical effects, although NCFR structure and draft evolution are little changed except that the maximum downdraft is strengthened somewhat. Hydrometeor mass increases by ~20% overall, ~50% for snow and cloud water, and ~15% for graupel. Among snow sources, riming is more than doubled, as are the Bergeron and accretional growth rates from cloud ice, while conversion from cloud ice is nearly halved. Accretional graupel growth is nearly doubled for cloud water, though slowed somewhat for snow.