Layered soil medium is a commonly existed soil layer structure. The accurate division of soil layers determines the accuracy of related researches on soil science. Therefore, it is of practical significance to study the techniques of fine soil layering. In order to classify the upper soil non-destructively in the field scale, the resistivity imaging technology was used. Based on the changes of the resistivity curve longitudinally, the stratification results were adopted to study the spatial variability of the soil resistivity. Results showed that:After finely classifying resistivity data using resistivity curve method, the result map could meet the need of visualization and fine expression in the research process. Using the Wenner quadrupole inversion resistivity curve method, the shallow soil in the research area were divided into four layers, including 5-13, 13-24, 24-50 and 50-70 cm, the prediction value and actual value were closest. I was found that the coefficient of variation of 5-13 cm (plough layer) was the highest (0.28) and the average resistivity value at 13-24 cm (plow bottom layer) was the lowest (17.35 Ω·m). The results conformed to the actual soil conditions in the test area, through this method, the soil condition could be obtained accurately, quickly and even non-destructively. The findings attained in this research provided technical support for the research on accurate agriculture and preferential flow of soil fracture.