This study demonstrates how mineralogy generated from spectroscopic visible, near, and shortwave infrared reflectance data, collected using rapid drill core logging systems, can be used to build one-, two- and three-dimensional models of the architecture of ore systems, with applications toward exploration and mining. In particular, this study examines both spectroscopic and traditional mining data (laboratory geochemistry and field geologist logging) from 180 drill holes through the Rocklea channel iron deposit in the Hamersley region of Western Australia. Valuable infrared reflectance spectroscopy-based mineralogy includes: the types and abundances of iron (oxyhydr-)oxides (hematite, vitreous goethite, and ochreous goethite); clays (well- and poorly ordered kaolin, Al and Fe smectite); and carbonates (calcite and dolomite). Algorithms to determine the abundance and composition of these minerals from infrared reflectance spectra were validated in an associated study by Haest et al. (2012) and were found to be accurate (e.g., root mean square error (RMSE) for Fe (oxyhydr-)oxide abundance prediction of 9.1 wt % Fe). Petrographic analysis and validated infrared reflectance spectroscopy-based mineralogy constrained the stratigraphy in the paleochannel and the paragenetic history of each horizon. The Rocklea channel iron deposit developed over weathered basalt and metasedimentary rocks, with an internal channel stratigraphy dominated from bottom to top by the following: (1) well-ordered (in situ) kaolinite and a partially denaturated channel iron deposit, (2) poorly ordered (transported) kaolinite, (3) ochreous goethite with scarce ooidal textures, no clays, and late-stage vitreous goethite/silica replacement, (4) mostly poorly ordered kaolinite, and (5) calcrete with associated Fe smectite.
The mineral paragenesis of the Rocklea deposit was compared against current models for channel iron deposit formation. A key finding is the relationship between parent rock composition, superimposed regolith cover, and channel iron deposit ore quality, which are linked through ground-water interactions and the physicochemistries of both the parent rock and the deposit. This has implications for both exploration and ore deposit characterization: (1) ore-grade channel iron deposits can develop several kilometers downstream (50 km, in this case) from the proposed source rock areas above mafic volcanics and metasediments, (2) vitreous goethite, calcrete, Al and Fe smectite cover the Rocklea deposit and are proposed as potential vectors to channel iron deposit mineralization at depth, (3) improved mineralogy from hyperspectral data in terms of iron (oxyhydr-) oxide and clay content/composition can improve iron ore resource delineation, iron ore processing (separation of vitreous and ochreous goethite), and pit design (differentiating “swelling” smectite (requiring pit walls with lower slopes) from kaolin or white mica).