This work reports on the sensitivity of accumulated precipitation to the microphysical parameterization in simulations of deep convective storms using a three-dimensional, nonhydrostatic cloud model with a simple liquid–ice microphysics scheme. Various intercept parameters from an assumed Marshall–Palmer exponential size distribution are tested along with two particle densities for the hail/graupel (qh) category. These variations allow testing of unique qh distributions that have been observed and documented in previous literature. Tests are conducted for a single thermodynamic profile and three idealized wind shear profiles. The amount of accumulated precipitation at the ground is very sensitive to the way the qh category is parameterized. Distributions characterized by larger intercepts and/or smaller particle density have a smaller mass-weighted mean terminal fall velocity and produce smaller qh mixing ratios spread over a larger area. For example, for a qh category weighted toward graupel, only a fourth as much precipitation accumulates on the ground over 2 h (and none is hail) compared to a qh category weighted toward large hail (with baseball-sized stones common). The inherent uncertainty within the qh distribution for this simple cloud-scale three-class ice microphysics scheme suggests limited usefulness in the forecasting of ground-accumulated precipitation and damaging hail.

Weisman and Klemp suggested that their liquid-only, deep convective storm experiments should be repeated with a liquid-ice microphysics scheme to determine if the solutions are qualitatively the same. Using a threedimensional, nonhydrostatic cloud model, such results are compared between three microphysics schemes: the ''Kessler'' liquid-only scheme (used by Weisman and Klemp), a Lin-Farley-Orville-like scheme with liquid and ice parameterization (Li), and the same Lin-Farley-Orville-like microphysics scheme but with only liquid processes turned on (Lr). Convection is simulated using a single thermodynamic profile and a variety of shear profiles. The shear profiles are represented by five idealized half-circle wind hodographs with arc lengths (U s) of 20, 25, 30, 40, and 50 m s Ϫ1. The precipitation, cold pool characteristics, and storm evolution produced by the different schemes are compared. The Kessler scheme produces similar accumulated precipitation over 2 h compared to Lr for all shear regimes. Although Kessler's rain evaporation rate is 1.5-1.8 times faster in the lower troposphere, rain production is also faster via accretion and autoconversion of cloud water. In addition, nearly ϳ40% more accumulated precipitation occurs in Li compared to Lr. This can be attributed primarily to increased precipitation production rates and enhanced low-level precipitation fluxes in Li for all shear regimes. Differences in the amount of precipitation reaching ground and the low-level cooling rates also cause differences in storm cold pools. For the U s ϭ 25 shear regime, microphysics cases with colder low-level outflow are shown to be associated with temporarily weaker (Li) or shorter-lived (Kessler) supercells as compared to cases with warmer outflow (Lr). This is consistent with a previous study showing that the cold pool has a greater relative impact on the storm updraft compared to dynamic forcing for weaker shear.

No abstract