Spatial variation of cone tip resistance for the clay site at Texas A & M University

The cone tip resistance data available for the clay site at Texas A & M University (one of the National Geotechnical Experimentation Sites) are used to show how the spatial variation of a soil property can be quantified. It is suggested that the first step in quantifying the spatial variation of a soil property should be the identification or selection of the statistically homogeneous soil layers. A new simple procedure is suggested to identify statistically homogeneous layers in a soil profile. Through examples it is shown that the procedure works well in identifying the statistically homogeneous layers. For the chosen statistically homogeneous layers, the spatial variation of cone tip resistance in the depth-wise is quantified in terms of a constant mean or a mean trend, variance/standard deviation/coefficient of variation or a standard deviation around the mean trend, and a correlation or variogram function. Correlation distances in the depth direction were ranged between 0.1 and 0.5m for the two soil layers investigated. It was shown that the correlation distance can decrease in the presence of a global mean trend for the soil property. In such cases, it is important to note that a part of the correlation is automatically included in the mean trend function. Cone penetrometer data available at spacings greater than or equal to 1m failed to indicate presence of correlation structures in the horizontal direction. Self-affine fractals seem to have potential in capturing the different auto-correlation structures that exist with spatial variation of soil properties.