Analysis methods#
Different methods have been used to analyse the Rocklea samples, giving an opportunity to compare the measurement and methods associated. The hyperspectral data set of the Rocklea dome aims to provide new informations about this method and improve datamining for exploration and mining, integration of independent geoscience data sets, resource modelling and different approaches for processing of hyperspectral data. Two types of samples have been drilled in this area : diamond core and drill chips.
Methods used in the past to describe the area#
The main method used in the studies is infrared spectroscopy on pieces of drill core. The relevance and interest of the study varies according to the number of cores and the number of elements considered. In our case, to study iron, most of the figures compare the contents of goethite, hematite, kaolinite, aluminum, according to the depth and the different drilling locations. The drilling data come from a study that carried out a drilling campaign in a part of the paleochannel overlying the Archean, and corresponding to an area historically exploited for iron, being one of the areas whose tonnage is important enough to justify a mining interest (Area belonging to Murchison Metals Ltd., having >50 wt% in iron).
X-Ray Fluorescence (XRF)#
This method has been used to determine the geochemistry of the samples, as well as the loss-on-ignition method. This method uses the characteristics X-rays of an element, generated when X-rays irradiate this element. Samples passing under the X-ray analysis need to be prepared following different steps. First, samples are dried in order to remove all the water they can contain. Samples were then crushed to a particular diameter (75 µm for the Rocklea study) and fused in the furnace before being casted into 40mm diameter beads. This beads are then used to determine the weigth percentage of FeO, P, S, SiO2, Al2O3, Mn, CaO, K2O, MgO, and TiO2 in each sample thanks to the X-ray analysis. This method give a chemical information about the drill core.
Loss-On-Ignition (LOI)#
Loss on ignition is a method used to study the composition of a mineral and give a chemical information. It consists in strongly heating the sample at a certain temperature to make volatil substances escape. The sample is heated until its mass ceases to change. we can calculate the weight loss with the following formula
This method is mostly used to study the organic matter of a sample. In the Rocklea data set, samples were heated to 1000 \(^\circ\) C. Goethite releases its water between 260 and 425 \(^\circ\) C, organic matter disappear by 550 \(^\circ\) C, clay material cecomposed between 530 and 605 \(^\circ\) C and inorganic carbon transforms itself in CO2 between 700 and 850 \(^\circ\) C.
Spectral Unmixing#
For remote sensing methods, the signal detected by a sensor into a single pixel is often a combination of numerous disparate signals (for instance vegetation or soil component can perturbate the analysis of a rock sample). The process to unmix the signal begins with a reduction of the dimensions of the picture. Then we have to select the endmembers
VNIR–SWIR Drill Core Spectroscopy#
These methods were used to draw the reflectance spectras of the samples studied. Drill core are put on an automated tray that move in a snake-like pattern under the optical fibre of the spectrometer. Spectral datas are collected for each sample from a 1 x 1 cm area. These measurements constitute the hyperspectral datas. Spectras for this data set were collected in the VNIR (380-1000nm) and SWIR (1000-2500nm). High-spatial-resolution images are collected too (0.1 mm pixel).
MFEM (Multiple Feature Extraction Method)#
Feature extraction methods are used to determine the mineralogy of a sample thanks to the intensity and wavelength position of absorption features in the reflectance spectra. Hyperspectral technologies use various range of wavelength in the SWIR (short wave infrared), VNIR (visible and near wave infrared) and thermal-infrared wavelength ranges. MFEM has the advantage of being only based on the visible and/or infrared active functional groups of minerals (molecules from the same functional group have the same characteristical chemical reactions).