Spatial modeling of discontinuous geologic attributes with geotechnical applications

Geologic data obtained by drilling investigations, field measurements, and laboratory tests are irregularly distributed with respect to location both horizontally and vertically. Spatial discontinuities impact soil classification, hydraulic conductivity, and chemical content. In this study, a three-dimensional modeling method, which combines the optimization principle method and a stochastic simulation method, was developed (OPTSIM). This method consists of three steps: the transformation of geologic attributes into binary (0 or 1) under certain thresholds, the three-dimensional interpolation of these transformed binary data, and the construction of conditional cumulative distribution functions (CCDF) by repeatedly changing the threshold and interpolation analysis. The Monte Carlo method was then used to inversely determine the attributes from the constructed CCDFs. Three types of attribute expressed by value, code, and class were examined. The OPTSIM was applied to spatial modeling of geotechnical structures in the Kumamoto plain, southwest Japan, using two kinds of data, lithologic units and physical soil properties. Resistivity data obtained by electric logging and the locations of screens were linked to geologic distribution. The constructed model contributed to the location of chief aquifers and estimates of amount of shallow groundwater. A combination of granular composition, N value, and water content was examined. Transgression and regression patterns following the last glacial stage were also inferred through this analysis. In both applications, discontinuities in lithologic units and soil properties resulting from fault movements and sea-level changes were successfully expressed in the distribution models.