ABSTRACT
Loess is the product of natural history. The spatial variability of soil parameters is complex and has strong regional characteristics, which is difficult to be evaluated accurately by the traditional random variable model. In view of this, this paper based on the Huayin test site of Zhengzhou Xi’an passenger dedicated railway, the correlation distance and spatial variability of loess soil parameters are statistically analyzed by using random field theory and semi-variogram model. The semi-variogram method is used to solve the correlation distance and variance reduction function. With the help of variance reduction function, the point variation coefficient is transformed into spatial variation coefficient, and the spatial variation coefficient is used as the basis for spatial variability evaluation. According to the semi-variogram, Kriging interpolation is used to obtain the predicted value of coefficient of collapsibility and the calculated collapse value by Kriging method. The results show that spherical model and exponential model are more suitable for the random field model of loess soil parameters. The correlation distance of soil parameters varies due to the influence of error disturbance. From small to large, they are collapsibility index, consistency index, mechanical index and physical index, and the spatial variability of mechanical index is generally greater than physical index. The correlation distance of soil parameters varies due to the influence of error disturbance, from small to large, they are collapsibility index, physical state index, compressibility index and physical property index. The spatial variability of mechanical index is generally greater than physical index. The calculated value of collapse value considering the spatial variability of soil parameters is increased compared with the standard method.
