LAUSR

Ice‐phase cloud microphysical processes are very complicated, and there are still many uncertainties in current microphysics parameterization schemes. In this study, two alternative ice crystal deposition/sublimation (ICDS) parameterizations, following the Harrington et al., Journal of the Atmospheric Sciences, 1995, 52, 4344–4366 and Koenig, Monthly Weather Review, 1972, 100, 417–423 methods, were implemented into the Weather Research and Forecast (WRF) Single‐Moment 6‐class Microphysics (WSM6) scheme in the Global/Regional Assimilation and Prediction System (GRAPES) regional operational model to investigate their impact on overestimation of snowfall over North China. The results show that the snowfall amount and cloud particle composition are very sensitive to the ICDS parameterization. Sensitivity tests with WSM6 using the Harrington and Koenig ICDSs, referred to as WSM6_H and WSM6_K, respectively, notably reduced overestimation of the snowfall amount and simulated more realistic supercooled cloud water content compared with ERA5 reanalysis data. The vertical distributions, intensities, and duration of radar echoes simulated by WSM6_H are the most consistent with millimeter‐wavelength cloud radar observations. There is competition for water vapor in WSM6 due to the sequential order in which processes are updated. This tends to weaken the deposition of snow and graupel and the condensation of cloud water when the deposition of ice crystals is too strong in a cold region. In both WSM6_H and WSM6_K, the deposition processes of ice crystals are reduced and the other microphysical transformation processes become more active compared with WSM6. Overall, the WSM6_H configuration improves the forecast performance as evaluated by Taylor diagrams for eight snowfall days over North China. The ICDS parameterizations must therefore be handled carefully due to their large uncertainties in the development of the cloud microphysics schemes.