Different cloud types have distinct radiative effects on the energy budget of the earth-atmosphere system. To better understand the cloud radiative impacts, it is necessary to distinguish the effects of different cloud types, which can be achieved through the cloud radar data that can provide cloud profiles for both day-to-day and diurnal variations. In this study, we use 6-year high-temporal resolution data from the Ka-Band Zenith Radar (KAZR) at the Semi-Arid Climate and Environment Observatory of Lanzhou University (SACOL) site to analyze the physical properties and radiative effects of main cloud types. The three types of clouds that occur most frequently at the SACOL site are single-layer ice clouds, single-layer water clouds, and double-layer clouds with the annual occurrence frequencies being 29.1%, 3.4%, and 8.3%, respectively. By using the Fu-Liou radiative transfer model simulation, it is found that the distinct diurnal variations of both the occurrence frequency and their macro- and micro-physical properties significantly affect the cloud-radiation. On annual mean, the single-layer ice clouds have a positive radiative forcing of 7.4 W/m(2) to heat the system, which is a result of reflecting 12.9 W/m(2) shortwave (SW) radiation and retaining 20.3 W/m(2) longwave (LW) radiation; while the single-layer water clouds and double-layer clouds have much stronger SW cooling effect than LW warming effect, causing a net negative forcing of 8.5 W/m(2). Although all these clouds have an overall small cooling effect of 1.1 W/m(2) on the annual radiative energy budget, the significant differences of the diurnal and seasonal distributions for different type clouds can lead to distinct radiative forcing. Especially the LW warming effect induced by the exclusive ice clouds in the cold season may have an important contribution to the rapid winter warming over the semi-arid regions.