Abstract: Ice clouds are one of the most important components in the climate system, and the parameterization of their life cycle in climate models has always been a prominent issue. The terminal fall velocity of ice cloud particles is a key parameter affecting the life cycle of ice clouds. The Ka-band Zenith Radar (KAZR), which has been deployed at the Semi-Arid Climate and Environment Observatory of Lanzhou University (SACOL) since July 2013, has been continuously operated for six years. By using continuous observations of the Ka-band millimeter-wave cloud radar from August 2013 to July 2015, we retrieve the particle terminal fall velocity (V_t) and calculated the values of the coefficients a and b based on the relationship between the radar reflectivity factor Z and V_t. We then further divide the ice clouds into four categories using cluster analysis, and discuss the spatiotemporal distribution of the reflectivity, the terminal fall velocity, the coefficients a and b of the four different types of ice cloud. Furthermore, we also study the change of microphysical processes at different locations inside the clouds through the parameter vertical distribution feature recognition. The results show that the distribution of V_t corresponds well to Z, and the maximum frequency appears at about 7 km altitude above the ground level (AGL). In addition, they all show significant seasonal changes and the terminal fall velocity can be increased by 25% in the warm season compared to that in the cold season. The average values of radar reflectivity factor, particle terminal fall velocity and fitting coefficient a and b of the first and third types of ice clouds, which are thicker with longer duration, are significantly larger than those of the second and fourth types of clouds, which are thinner clouds with shorter duration. Seasonal changes of parameters of all types of ice clouds are relatively consistent, i.e., they all peak in June and September. In the vertical direction, Z, V_t and fitting coefficient b from the top of to the bottom of the cloud show a tendency to increase first and then decrease with the decrease of the height of the clouds, which reflects the microphysical processes at different locations in the cloud. In the top part of the cloud, the nucleation of particles and the deposition of water vapor in the clouds are dominant. With the decrease of cloud height, the deposition of water vapor and the aggregation of particles play a major role, and the cloud particles gradually increase. In the cloud layer close to the cloud bottom, the cloud particles evaporate and sublimate, and the particle size decreases. This indicates that the formation of ice clouds in arid and semi-arid areas in the middle latitudes of China is a top-down process from the top to the bottom of the cloud. While the distribution characteristics of fitting coefficient b in the vertical direction also indicate that this parameter can be used for the identification of different physical processes within the cloud.