By use of a three-dimensional compressible non-hydrostatic convective cloud model with detailed microphysics featuring spectral bins of cloud condensation nuclei （CCN）, liquid droplets, ice crystals, snow and graupel particles, the spatial and temporal distributions of hydrometeors in a supercell observed by the （Severe Thunderstorm Electrification and Precipitation Study） STEPS triple-radar network are simulated and analyzed. The bin model is also employed to study the effect of CCN concentration on the evolution characteristics of the supercell. It is found that the CCN concentration not only affects the concentration and spectral distribution of water droplets, but also influences the characteristics of ice crystals and graupel particles. With a larger number of CCN, more water droplets and ice crystals are produced and the growth of graupel is restrained. With a small quantity of CCN the production of large size water droplets are promoted by initially small concentrations of water droplets and ice crystals, leading to earlier formation of small size graupel and restraining the recycling growth of graupel, and thus inhibiting the formation of large size graupel （or small size hail）. It can be concluded that both the macroscopic airflow and microphysical processes influence the formation and growth of large size graupel （or small size hail）. In regions with heavy pollution, a high concentration of CCN may restrain the formation of graupel and hail, and in extremely clean regions, excessively low concentrations of CCN may also limit the formation of large size graupel （hail）.
A 3-D convective cloud model with compressible non-hydrostatic dynamics and the spectral bin microphysics of a 2-D slab-symmetric model has been used to simulate an observed supercell storm occurring on 29 June, 2000 near Bird City, Kansas, USA. The main
