Abstract: The spatial variability of shear strength parameters of rock and soil mass on slopes is structural. Aimed at studying the effect of spatial variability structure of rock and soil parameters on failure probability of slopes, the transformation relationship between range and correlation distance is deduced based on the mathematical intension of variogram firstly. The simulation method of structural cross-constraint random field is proposed to simulate the cross-correlated parameters random fields. In order to study the effect of shear strength parameters with different spatial variation structures, a slope calculation model of structural cross-constraint random field is established. The results show that the structural cross-constraint random field can reflect real formation parameters by considering random deviation, conditional data and different spatial variation structures. Compared with conditional parameter interpolation field, the proposed method is suitable for parameters with more complex spatial variation structures, i.e. directionality and anisotropy, and the generated data have less fluctuation. The reliability analysis also reveals that the failure probability of slopes is prone to be overestimated without considering the spatial structure parameters. The failure probability increases with the increase of the correlation coefficient when the correlation between the two factors is considered. It is easier to overestimate the slope failure probability under the scenario of negative correlation between c' and φ'.

Key words: spatial variation structure, structural cross-constraint random field, slope stability, failure probability