中国土壤土层厚度的空间变异性特征

以全国第二次土壤普查的1627个土壤剖面资料为基础,在地质统计学和地理信息系统的支持下,以变异函数为基本工具初步分析中国土壤土层厚度的空间变异特征,并应用普通克里格法进行最优无偏线性插值,制作出分辨率为30km×30km的中国土壤土层厚度的空间分布图。结果表明:中国土壤土层厚度具有较好的可迁性和空间结构性特点,实验变异函数值的变化趋势基本上随着距离的增加逐渐上升,拟合变程在680km以上,土壤厚度的相关性可大于680km,土层厚度具有明显的块状或连续分布的特点

Characteristics of spatial variability of soil thickness in China

The spatial variability of soil properties is very important to agriculture, soil quality assessment, sustainable development and global change research. Geostatistics is a means of spatial variability, spatial structures and spatial prediction that is used widely for soil properties overseas.However, this kind of research was started relatively late in China. On the other hand, our research on soils paid more attention to chemical and biological properties than to physical properties. This paper aimed to study spatial distribution characteristics of soil depth in China, especially that of physical ones using Geostatistics and Geographic Information System, and provide raster data for global change research.This paper built the China Soil Geographic Database which includes soil physical and chemical properties database and 1627 soil samplers database by data from the second national soil surveys and primarily studied the spatial distribution characteristics of the thickness of soil layer using GIS and Geostatistics. The spatial distribution features were quantitatively described by semi-variogram. Based on the semi-variogram, the maps illustrating soil thickness are made by ordinary Kriging with cell of 30km×30km. 　　Soil thickness being studied has relatively good spatial structures, and their experimental semi-variogram could be fitted by exponential model with nuggets. And optimal interpolation could be done depended on the semi-variogram features. Basically, the experimental semi-variogram of soil properties being studied increases with increasing lag distance to approach or attain a maximum. The range of soil thickness semi-variogram is up to 680km, which is consistent with average distances among patches with the same feature in the distribution maps. This indicates the significant trend known as “drift” in distribution maps.