Cloud radar is a powerful instrument for cloud detection. However, the parameters of the radar transmitter and receiver system can drift during operation, which will cause the overall offset error of the observation data and affect the accuracy of cloud physical property retrieval. Therefore, cloud radar data need to be periodically calibrated properly. This paper uses the radar data calibration method proposed by Pavlos et al., which optimizes weak clouds and precipitation signal recognition. It also uses CloudSat satellite-borne radar observations of reflectivity factors, gas attenuation correction and other data to improve the data quality of the Ka-band Zenith Radar (KAZR) deployed at the Semi-arid Climate and Environment Observatory site (35.57°N, 104.08°E) of Lanzhou University (SACOL). It has established a 46-month historical data calibration from August 2013 to May 2017 for the KAZR radar reflectivity. That means calibration of cloud radar reflectivity data is essential for obtaining accurate cloud macro- and micro-physical properties, providing a solid foundation for long-term clouds study at the SACOL site.