Random field characterization of CPTU soil behavior type index of Jiangsu quaternary soil deposits

The soil behavior type index (I (c)) obtained from piezocone penetration tests (CPTU) has been widely applied in a series of geotechnical issues including the identification of soil stratum, estimation of soil properties and liquefaction analysis. To provide spatial statistics of I (c) for the probabilistic analysis of these geotechnical applications, this paper investigated the inherent vertical variability of the I (c) data using Gaussian random field theory. To achieve the analysis, 134 CPTU soundings were performed at the sites of four typical geologic formations. These formations included marine, Yangtze River Delta, long river floodplain and abandoned Yellow River floodplain deposits in the Jiangsu Province, China. Statistically homogeneous soil units (HSUs) were firstly identified based on a screening procedure. Then the probability density distribution of the I (c) data in each HSU was studied using a formal normality test and the quantile plot method. Afterwards, the method of moment was used to estimate the three random field model parameters of I (c), including the mean value (mu (Y) ), coefficient of variation (COV), and vertical scale of fluctuation (delta (v)). It was shown that the normality hypothesis may be suitable for both raw I (c) and the fluctuation components of I (c) data in a qualified sense, whereas, in a strict sense, the feasibility of this assumption is complicated with about 50 % data passing the formal normality test. The COV of I (c) data of the Jiangsu soils varies from 1.9 to 14 %. The delta (v) of I (c) profile ranges from 0.1 to 1.17 m. The impacts of the geologic formation and soil behavior type on the probability density distribution and random field characteristics of I (c) data were observed as the normality, COV and scale of fluctuation values of I (c) tend to be similar, provided that the soils are of the same geologic formation. Finally, suggestions for future selections of the normality hypothesis, coefficient of variation and vertical scale of fluctuation of I (c) data are proposed.