Abstract: The spatial variability of soil properties is widely accepted, and the response of a geo-structure can be significantly affected by the spatial variability of the surrounding soil mass. The random field theory is a popularly used method to simulate the spatial variability of soil properties. The stochastic analysis of horizontal convergence of tunnel is carried out using the random field difference method considering the spatial variability of Young's modulus. The random field difference method is combined with the Monte Carlo method and finite difference simulation based on random field theory. A large number of Monte Carlo simulations are adopted in the proposed random field difference method. Meanwhile, three simple and easy-to-use methods for the spatial variability of soil are proposed: reduction factor method, amplification factor method and reliability partial factor calibration method. Based on the statistical analysis of the calculated results, the suggested values of three simplified analysis methods under different combinations of the horizontal scale of fluctuation, the vertical scale of fluctuation and the coefficient of variation are given. This study may provide references for the application of spatial variability research to practical geotechnical engineering.