Abstract:

【Objective】Taking Pinggu district as the research region, the paper studied spatial variability and explored mechanism of spatial variability of soil texture on a county scale by traditional statistics methods and geo-statistics.【Method】 The paper analyzed the relationship between different soil particle composition and some factors by the Least Significant Difference (LSD) according to the test results by Levine’s method. As soil textures are the composition data, they were transformed by SLR and then the spatial distribution of the particle composition was predicted by Ordinary Kriging.【Result】Mean values of soil particle composition of different elevation groups, parent material, and land use and water buffer groups had some differences, and the differences were more significant between elevation, parent material types and soil particle composition. Generally speaking, soil particles became finer gradually with the decrease of elevation. Soils developed from acid rock with high quartz content are coarser. Soil particles of vegetable plot were relatively finer. Sand content increased with distance increased from water. Results analyzed by geo-statistics showed that strong spatial autocorrelation of different soil particle compositions and spatial variability was caused by structural properties. Interpolation results showed that the overall trend characteristics of spatial distribution of different soil particle compositions are obvious. 【Conclusion】By using methods of combining statistical analysis and geostatistics, deterministic and stochastic of soil texture can be comprehensively and systematically reflected. Distributions of soil particles mainly were affected by natural factors, such as topography and parent material. After translated by SLR,the prediction results of soil texture met the requirements for composition data interpolation.

Key words: soil texture, spatial variability, composition data, factor analyses