Abstract

Short-wave infrared (SWIR, 1.90–2.36 μm) and long-wave infrared (LWIR, 8.1–11.1 μm) hyperspectral images collected using the SisuROCK system were used to develop an automated methodology for generating kimberlite dilution maps. Smoothed and denoised images from two Snap Lake (Northwest Territories, Canada) kimberlite drill cores were processed, and SWIR and LWIR spectral endmembers were extracted from the images with each mineralogical endmember assigned to one of four compositional groups: undiluted kimberlite, microdiluted kimberlite, micro- and macrodiluted kimberlite, and crustal rocks. These endmembers were used to classify the SWIR and LWIR images, and the results were validated using linescan data, drill core logs, petrology reports, and the results of X-ray diffraction, Raman spectroscopy, and Micro-Fourier Transform Infrared (FTIR) spectroscopy. This study demonstrates that hyperspectral imagery can be used to generate dilution maps for hypabyssal kimberlites that far supersede other current techniques in terms of spatial resolution.

You do not currently have access to this article.