Abstract: Based on the precipitation data of western Hunan province from 1981 to 2019, the spatial structure and variation characteristics of annual and seasonal precipitation were analyzed by using geostatistical variogram theory, and the spatial distribution of annual and seasonal precipitation was studied based on ordinary Kriging interpolation. The results showed that there was a second-order trend in the annual and seasonal precipitation in western Hunan province. In the east-west direction, the annual, spring and summer precipitation presented an inverted U-shaped trend, increased gradually from east to west in autumn, and decreased gradually in winter; In the north-south direction, the annual precipitation presented a U-shaped trend, spring and winter gradually decreased from south to north, and summer and autumn gradually increased. Through the comparison of interpolation errors, the optimal fitting model of annual and seasonal precipitation was obtained. The annual precipitation was spherical model, spring, summer and autumn were Gaussian model, and winter was exponential model. The C₀/（C₀+C） of winter precipitation was less than 25%, which indicated that the winter precipitation had strong spatial correlation, and the spatial variation caused by random factors was very small. The C₀/（C₀+C） of annual and other seasonal precipitation was between 25% and 75%, indicating that the precipitation had moderate spatial correlation. The annual, summer and autumn precipitation in western Hunan province was decreasing gradually from north to south, while it was increasing in spring and winter.

Key words: geostatistics, precipitation, spatial variability, Kriging interpolation, western Hunan province