It is found that without the Hallett-Mossop ice multiplication process the GSFC ice microphysical scheme fails to reproduce the observed dense cloud ice particle region around -5°C in two cases. The inclusion of this second cloud ice production processes in the model microphysics scheme helps to reproduce this detected region with the order of magnitude of simulated number concentration of cloud ice being close to the observed by the 1D-C and 2D-C probes. This indicates that this ice-producing process is vital for accurate simulation of microphysical properties although its existence increases by only 5% the mean areal precipitation in the Folsom Lake basin for the two case studies. In addition, it is found that the simulation of number concentration of precipitating particles depends on the accurate selection of the intercept N0 of snow particle size distribution. This parameter can modify mean area precipitation in the Folsom Lake basin by about 7-10%, that is, more significantly than the Hallett-Mossop process.
Division of precipitation types and comparison between the simulations with/without the riming process reveal that (a) graupel (or rimed snow) is the dominant precipitation type especially in the windward slope and (b) the riming process is critical for distributing precipitation in the direction normal to barrier. This suggests parameterizing riming degree in microphysics schemes and investigating its effects on the improvement of QPF.