Stochastic modelling of hydraulic conductivity derived from geotechnical data: an example applied to central Glasgow

摘要: Characterising the three-dimensional (3D) distribution of hydraulic conductivity and its variability in shallow subsurface is fundamental to understanding groundwater behaviour developing conceptual numerical models manage subsurface. However, directly measuring situ can be difficult expensive rarely carried out with sufficient density urban environments. In this study we model for 603 sites unconsolidated Quaternary deposits underlying Glasgow using particle size description widely available from geotechnical investigations. Six different were applied MacDonald formula was found most applicable heterogeneous environment, comparing well few data. The range calculated values between 5th 95th percentile 1.56×10 –2 –4.38mday –1 a median 2.26×10 mday . These modelled data used develop suite stochastic 3D simulations conditioned existing representations lithology. Ten per cent input excluded modelling process use split-sample validation test, which demonstrated effectiveness approach compared non-spatial or lithologically unconstrained models. Our spatial reduces mean squared error estimated observed at locations over those predicted simple homogeneous by 73 %. resulting much higher resolution than would have been possible only direct measurements, will improve flow reduce uncertainty parameters methodology employed could replicated other regions where significant volumes suitable site investigation are predict ground conditions areas complex superficial deposits.