Abstract:Zoning based on spatial variability and spatial autocorrelation of soil nutritions on a field scale,is a key to variable-rate fertilization in precision agriculture.First of all,spatial variability and spatial autocorrelation of three main soil nutrients were analyzed.The variation coefficient of available P,alkalytic N and available K was 44.3% ,13.2% and 34.3% ,respectively.Available P and alkalytic N demonstrate medium spatial autocorrelation,but available K did not show any on the sampling scale.Precision of the Kriging interpolation of alkalytic N was improved by 42.8% considering anisotropy of spatial variability. Secondly,fuzzy-K-means cluster analysis of interpolated soil distribution maps was conducted to delineate management zones based on available P and alkalytic N.Fuzziness Performance Index(FPI) and Modified Partition Entropy(MPE) were employed to determine optimal zone number.The whole field was divided into four management zones.Finally,the zoning statistics showed that variation coefficient of alkalytic N and available P within a zone decreased from 13.2% on the field scale to 3.0%~5.3% , from 44.3% to 10.1%~17.3% ,respectively,while the means of the soil nutrients differed sharply between management zones.