Abstract: The spatial variability of soil properties in reliability analysis is of great importance, of which the inherent variability of soil can be analyzed through random field theory by determination of coefficient of variance and scale of fluctuation. In this paper, efforts have been devoted to the fluctuation function method to determine scale of fluctuation. Meanwhile, factors such as trend removal method, sampling spacing and sampling range are analyzed based on tip resistance from cone penetration test(CPT), modeling as a random field. It turned out that, as to analyzed data in this paper, if more than one peak point exists when employing the fluctuation function method, the first peak point should be taken as the scale of fluctuation of the soil, which has never been analyzed before. As to affecting factors, some very useful conclusions are obtained. Firstly, selection of the trend removal method should be based on the condition of soil properties; and the quadratic trend removal method is suggested if nonlinear method adopted. Secondly, the scale of fluctuation increases with increasing of sampling spacing when the number of CPTs is large. Last but not the least, as to a same sampling spacing, the scale of fluctuation decreases with increasing of the number of penetration tests; but such difference gradually becomes not obvious for large sampling spacing.

Key words: scale of fluctuation, fluctuation function method, trend removal, sampling spacing, sampling range