Hill, June1 and Poulet, Thomas1
1MinEx CRC, CSIRO Mineral Resources, Kensington, WA, Australia
The collection and analysis of structural data is an important part of exploration for structurally controlled mineral deposits. For mineral deposits under cover, drill core is the main source of structural information. Methods for collecting high density structural data from core images are being investigated to complement traditional geology logging with data that is non-subjective and rapid to collect. Automated measurement of anisotropy can be used to quantify scale, intensity and orientation of vein networks, compositional layering and grain fabrics; these are key components of structural analysis. We have explored two published measures of anisotropy for rock images (1) variogram maps and (2) inertia tensor analysis to assess their practicality and usefulness for collecting continuous structural data from core images.
The variogram is a popular method of texture analysis. It measures the spatial continuity of the input variables. If the spatial continuity is stronger in one direction than another, then the fabric can be considered to be anisotropic. Variograms result from comparing the values of two points in the image at increasing distances from each other (lag). When the direction of the measured lag is included, this produces a 2D variogram map. Diaz et al. (2019, Mathematical Geosciences) tested the variogram map method for a set of rock images and demonstrated its ability to detect anisotropy. In this study, the Diaz method has been modified for continuous testing along the core image. A threshold is applied to the variogram map and an ellipse is fitted to the resulting data points. Varying the threshold provides results for different scales. The relative length of the ellipse axes provides a measure of anisotropy and the orientation of the principal axis provides the direction of maximum spatial continuity; i.e. the direction of the anisotropy. Prior to analysis the core photo is segmented into overlapping square images and the algorithm is applied to each image. The result is a continuous downhole measurement of anisotropy and angle.
The inertia tensor analysis provides a simple and general approach to quantify anisotropy in terms of position, direction and scale. We use the method described by Lehoucq et al. (2015, Frontiers in Physics), where an image is partitioned into analysis boxes at a chosen scale, with an ellipse being fitted in each box to determine the prevalent direction of the signal. This allows the user to determine an anisotropy measure with orientation at all scales up to the image size and the resulting signal provides useful information about the original image. This approach reproduces some results of the variogram method and also captures the variation of anisotropy and orientation with scale. In this study we investigated the signature of conceptual scenarios as well as real vein networks at different scales to understand how this quantitative measure can be used to characterise vein arrays on drill core images, including characteristic lengths and orientation. The variability of anisotropy as a function of scale allows us to identify different characteristics of vein arrays.
Using mathematical, statistical and computational techniques, including machine learning, June Hill specialises in automating the analysis and interpretation of numerical drill hole data.
Thomas Poulet’s research focuses on multiphysics instabilities in porous media. His expertise includes theoretical and numerical modelling, geomechanics, optimisation, software engineering and high-performance computing.