Study On Spatial Variability Of Soil Nutrients And Reasonable Sampling Number At County Scale

The soil fertility is one of the most important ecological functions of soil, and the soilvariation can cause the variation of crop growth. Accurately mastering the sptial distributionof soil nutrients is the foundation of managing soil nutrients and rational fertilization. Thus,in order to ensure the accuracy of the spatial interpolation of soil nutrients, we shouldconsider sampling number when studying spatial variation.Therefore. It was greatsignificance of quantitative research of soil nutrient variation、implementation of precisionagriculture and sustainable development of agriculture to understand the spatial variabilityof the soil nutrient and determineing the sampling number scientifically. Taking Fei countyof Shandong Province as a case, this paper discusses the spatial variability of soil organicmatter, total nitrogen, alkali solution nitrogen, available P and available K. And on this basis,both the ordinary kriging interpolation and cokriging interpolation were used to determinethe sampling number, and the influence factors of soil nutrient variation were also analyzed.（1）The spatial variability of soil nutrientsGS+9.0softwore was applied to analyze the semivariograms of soil nutrient, the resultsshowed SOM, TN, available P and available K were well simulated. The semivariogramsmodel can greatly reflect the spatial structure features of the soil nutrients.AN shows nospatial dependence, and can not be simulated with the theoretical model. SOM、TN and AKpresent moderate spatial dependence and larger range, the spatial variation of which wascaused by the interaction of random factors and structural factors.It shows that the threenutrients have very good space continuity, mainly influenced by the effect factors in largerscale.Whereas available P had shorter range and present strong spatial dependence only inthe small range, which was more vulnerable to the influences of extrinsic factors such ashuman activities. We can obtain the regional distribution of soil nutrient from spatialdistribution map. The content of SOM、TN and AK are in the medium level and thedistribution is distributed uniformly. Available P has higher content, and had much smallspots. (2)Determining reasonable sampling number by Or-krigingBased on the spatial correlation of soil nutrients and validation, we concluded that thereasonable sampling number of SOM、TN and AK was about1035、842and1033andsampling interval was about1400m and1500m. AN had no spatial dependence andsampling densities should be increased later for the further study of the spatial structurefeatures of the AN itself. Whereas available P had shorter range and present strong spatialdependence only in the small range, which was more vulnerable to the influences ofextrinsic factors such as human activities. Subsequent sampling cannot below the currentsampling densities. In order to ensure the spatial dependence of SOM、TN and AK, weshould accord to the minimum sampling spacing of the nutrients when sampling.Thereforthe rational sampling spacing was dermined about1400m.(3) Determining reasonable sampling number by Co-krigingTo compare the prediction accuracy of the soil nutrients by corkriging and ordinarykriging under different sampling numbers, the accuracy of ordinary kriging interpolationwas taken as the standard. The results show that both cokriging method and ordinary krigingmethod can reflect the spatial distribution features of the soil nutrient at county scale, whenthe sampling number were sufficient. Compared with the ordinary kriging under the samesampling number, the prediction accuracy of co-kriging is higher.When the original data ofTN reduced to538and SOM reduced to662, co-kriging still can better express the soilspatial distribution information. When AN was interpolated by corkriging, cross variablesshowed strong dependence and842points can better express the space distributionfeatures；The co-kriging interpolation of SK does not show obvious advantages, andreasonable numbers were still1033. Interaction variable of available P had no spatialdependence, which can’t be accurately simulated by cross-semivariograms. In order toensure the spatial dependence of SOM、TN、AN and AK, we should accord to theminimum sampling spacing of the nutrients when sampling.Therefor the maximumsampling spacing was dermined about1484m.(4) Factor analysis of soil nutrient spatial variabilityResearch shows that the nutrient content present a bigger difference under the differenttypes of land use.Except TN, the other four kinds of nutrients presented significantdifference under the five different types. The influence of the soil types to SOM、TN、AKand available P all reached significant level, this suggests that regional soil type is one ofthe important factors affecting soil spatial variability. At the same time, SOM、TN、AK and available P have larger coefficient of variations and standard deviation,and the wholedistribution is uneven.