This paper takes advantage of fractal method to research some soil characteristics through case analysis. The multifractal spectrum and random Sierpinski carpet are used to describe the spatial variability and thermal conductivity of soil quantitatively. On the basis of predecessors, the scatter plots of various types of data have been used to supplement the multifractal results in a more detailed way. It turns out that the content of clay, silt, and coarse sand could reflect the degree of spatial variability of soil. Then based on this case, the effect of porosity on soil thermal conductivity is discussed by using random Sierpinski carpet. The result shows that the effective thermal conductivity of the clay, silt and coarse sand decreases linearly with the increase of porosity, but the degree of reduction is different. Moreover, when the porosity is definite, the effective thermal conductivity of the coarse sand is the largest, that of the clay with the highest thermal conductivity is second, and that of the silt is the smallest.