ABSTRACT
In the past 10 years, much attention has been given to probabilistic analyses in modelling of rock dicontinuities. However, less attention has been paid to modelling the spatial correlation of geometric and mechanical parameters (autocorrelation). Geometric parameters of rock joints have been shown by geostatistical methods to be spatially dependent and are spatially correlated. The paper reviews the concepts and models used for modelling auto-correlation of joint geometric parameters. It includes discussion of basic definitions and methodology of geostatistical methods for spatial correlation — variogram, spherical model and hole-effect models, and that of probabilistic methods — autocovariance functions. Actual correlation distances for the geometric parameters are discussed. Also, random field modelling is explained due to its success in modelling the mechanical behaviour of soil slope stability, which also seems to provide a method for modelling autocorrelation in jointed rock parameters.
