Space‐local spectral texture segmentation applied to characterizing the heterogeneity of hydraulic conductivity

Abstract

Spatial variability of hydraulic conductivity exerts a predominant control on groundwater flow by influencing advective pathways, hydrodynamic dispersion, and density‐dependent instabilities. Space‐local spectral texture segmentation aids in the macroscale characterization of the spatial heterogeneity of natural porous media via an outcrop analogue approach. Detailed photographic data sets were obtained for a 45 m × 3 m vertical section of glacial‐fluvial sand and gravel deposit in the Fanshawe Delta area (Ontario, Canada). High‐resolution texture maps of the sedimentary exposure are generated using a texture segmentation routine based on the space‐local S transform with the photographic data sets used as input. Geostatistical analyses of the texture maps reveal similarity between the spatial correlation structures of spectral texture and hydraulic conductivity as determined from constant‐head permeameter testing of sediment cores. Conditioned on the permeameter measurements, texture maps can be used to provide local continuous estimates of the hydraulic conductivity field at a spatial resolution equal to the sediment core dimensions.