Inferring total organic carbon and major element geochemical and mineralogical characteristics of shale core from hyperspectral imagery

mineralogical information at the millimeter scale and thus have the potential to enable investigators to characterize shale composition and heterogeneity, complementing the direct chemical and x-ray diffraction analysis of core samples and guiding detailed sampling. This method provides insight into petrophysical and geomechanical properties because these properties are significantly correlated to rock composition. We tested this approach on a continuous long core from the shale sequence of the Horn River Group in the Horn River Basin, British Columbia, sampled at a spacing of 1 m (40 in.) and analyzed for geochemical composition. These data enable the calibration of spectral imagery to rock composition and specifically predict total organic carbon (TOC) and the abundance of SiO2, Al2O3, K2O, and CaO. We then imaged nine samples from the Woodford Shale from the Permian Basin, Texas, for a blind test to assess the predictive models. The models were then used to predict TOC and geochemical data over detailed imagery of 300 m (984 ft) of Horn River Group shale core and portray their spatial variability downhole as images and profiles. In its simplest form, hyperspectral imagery can be enhanced to highlight fabric in shale core that otherwise is difficult to visualize because of low brightness. In addition, we show that spectral imagery of shale can also be processed to either convey mineralogical (quartz, clay, and carbonate) or geochemical information. The resulting views can readily be used to guide the selection of samples and may provide tools for scaling reservoir properties from individual plugs to reservoir volumes.