The Di models method: geological 3-D modeling of detrital systems consisting of varying grain fractions to predict the relative lithological variability for a multipurpose usability

Abstract

AbstractThe coexistence of a wide variety of subsurface uses in urban areas requires increasingly demanding geological prediction capacities for characterizing the geological heterogeneities at a small-scale. In particular, detrital systems are characterized by the presence of highly varying sediment mixtures which control the non-constant spatial distribution of properties, therefore presenting a crucial aspect for understanding the small-scale spatial variability of physical properties. The proposed methodology uses the lithological descriptions from drilled boreholes and implements sequential indicator simulation to simulate the cumulative frequencies of each lithological class in the whole sediment mixture. The resulting distributions are expressed by a set of voxel models, referred to as Di models. This solution is able to predict the relative amounts of each grain fraction on a cell-by-cell basis and therefore also derive a virtual grain size distribution. Its implementation allows the modeler to flexibly choose both the grain fractions to be modeled and the precision in the relative quantification. The concept of information entropy is adapted as a measure of the disorder state of the clasts mixture, resulting in the concept of “Model Lithological Uniformity,” proposed as a measure of the degree of detrital homogeneity. Moreover, the “Most Uniform Lithological Model” is presented as a distribution of the most prevailing lithologies. This method was tested in the city of Munich (Germany) using a dataset of over 20,000 boreholes, providing a significant step forward in capturing the spatial heterogeneity of detrital systems and addressing model scenarios for applications requiring variable relative amounts of grain fractions.