Based on cloud macro and micro characteristic parameters (hereinafter referred to as cloud parameters) retrieved by the FY-2G geostationary satellite data, the temporal and spatial distribution of cloud characteristic parameters in Qinghai Province and 3 sub-regions from 2018 to 2020 were analyzed.The result show that the annual average cloud top height (CTH), cloud top temperature (CTT), overcooled layer depth (OLD), cloud optical depth (COD), effective radius (ER) and liquid water path (LWP) in Qinghai Province are 3.8 km, -9.7 ℃, 2.0 km, 7.1, 7.1 μm and 63.7 g∙m^-2, respectively. Except for CTT, the monthly variation of cloud parameters in the Qaidam Basin and Northeastern Qinghai Province with the same latitude showed roughly two peaks and two valleys and its peaks basically appeared in May and November, and the valleys basically appeared in August, September, December and January. Each cloud parameter was roughly unimodal in Three River Source Region, with a peak in November. The spatial distribution of annual average of each cloud parameter was roughly distributed along the topography and mountain range. Except for CTT, high-value areas corresponded to high mountains, low-value areas corresponded to desert basins and low-altitude areas, there was a low-value area in four seasons in the Qaidam Basin, and its range was largest in summer. There were obvious high-value areas in the Three River Source Region and the Qilian Mountains in Qinghai in spring and winter. The OLD, COD and LWP in Three River Source region were larger in spring and autumn, OLD and LWP in the northeastern Qinghai region were largest in spring. Spring and autumn were good time for artificial rainfall enhancement for the purpose of water conservation, drought resistance and disaster reduction.

Based on the cloud characteristic parameters retrieved by FY-2G satellite and hourly precipitation from July to September 2018, April to September 2019, and April to July 2020 in the eastern region of Qinghai Province, the indications of cloud top height, cloud top temperature, cloud optical thickness and cloud particle effective radius to precipitation frequency and precipitation intensity were analyzed. The results are as follows: (1) The cloud optical thickness was the strongest indicator of precipitation frequency for single cloud characteristic parameter. The frequency of moderate rain and heavy rain showed an obvious increasing trend with the decrease of cloud top temperature, increase of cloud top height and cloud optical thickness, while the frequency of light rain showed a decreasing trend. (2) The double cloud characteristic parameters (cloud optical thickness and cloud top temperature) were better indicators for precipitation frequency than single cloud characteristic parameter, the precipitation frequency increased with increase of cloud optical thickness and decrease of cloud top temperature. When cloud optical thickness was between 21 and 30 and cloud top temperature was greater than 0 ℃, the frequency of light rain was largest. When cloud optical thickness was greater than 40 and cloud top temperature was between -45 and -31 ℃, the frequency of moderate rain was largest. When cloud optical thickness was greater than 40 and cloud top temperature was less than -45 ℃, the frequency of heavy rain was largest. (3) The triple cloud characteristic parameters (cloud top temperature, cloud optical thickness and cloud particle effective radius) were more indicative for precipitation frequency than single cloud characteristic parameter and less indicative than double cloud characteristic parameters.

 Based on the precipitable water vapor calculated by radiosonde data (PWV_RS) at 00:00 UTC and 12:00 UTC in 2014 at Golmud station of Qaidam Basin, the accuracy of precipitable water vapor retrieved by GPS data (PWV_GPS) was tested. Meanwhile, the variation characteristics of PWV_GPS was analyzed at Golmud and Delingha stations in Qaidam Basin. The results show that daily variations of PWV_GPS and PWV_RS were consistent and the correlation coefficient between them was more than 0.9. The correlation was better in summer and autumn than in winter and spring, the correlation coefficient at 00:00 UTC was slightly higher than that at 12:00 UTC. At 00:00 UTC and 12:00 UTC, the RMSE of PWV_GPS was 1.8 and 2.4 mm, and the MRE was 0.2 and 0.4, and the BIAS was 4.2 and 4.3 mm, respectively. Monthly variation of PWV_GPS presented a single peak distribution, the maximum appeared in July and the minimum appeared in December. PWV_GPS was rich in summer and shortage in winter, and it was more in south and less in north. The diurnal variation of PWV_GPS ranged from 0.4 to 28 mm and the hourly PWV_GPS ranged from 6.9 to 7.3 mm.