Data properties under hyperspectral analysis#
We found a lot of informations on the hyperspectral file, concentration of differents chemical elements, identification of every hole and every sample, their position on a map but also some less explicit data that are describe below.
LOI (Loss On Ignition)#
Loss on ignition is a method used to study the composition of a mineral. 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. Its expressed as a weight percentage of the dry mass and calculated with the following formula
In the data set, samples are heated in an electric furnace to 1000 \(^\circ\) C.
(The Rocklea Dome 3D Mineral Mapping Test Data Set, paragraph 3.4)
Fe Ox ai (Ferric oxide abundance)#
This column give the abundance of the ferric oxide elements (Hematite, goethite, jarosite, “limonite”). Values around 0.04 indicate a low content in ferric oxide. There is no particular value for a high content in these elements, it depends of the other drill core samples. The accuracy of the multiple feature extraction method is high for this measure but complicated because of water vapour residuals and mix with other elements (dry and green vegetation, carbon black and ferrous iron in silicates/carbonates). For the Rocklea data set the RMSE (Root-Mean-Square deviation) is of 9.7%.
Hem/goe (Hemathite-Goethite distribution)#
Hematite is an iron oxide which formula is \(Fe_2O_3\). Goethite is an iron oxyhydroxide which formula is \(FeO(OH)\). The Hematite-Goethite ratio has been determined thanks to the Multiple feature extraction method (MFEM). Tracking the wavelength position for the same absorption feature has allowed to distinguish goethite from hematite. Values are taken for an absorption feature of 900nm and corresponds to a wavelength position, around 890nm being for a more hematitic sample and around 910nm being for a more goetitic sample. However the accuracy of these measures is moderate because of the mix with green and dry vegetation, ferrous-bearing carbonate and silicate minerals.
Kaolin abundance#
Kaolin abundance has permits to detect the kaolin group minerals, thanks to the MFEM, composed of kaolinite \((Al_2O_3 2SiO_2·2H_2O)\), halloysite (\(Al_2Si_2O_5(OH)_4)\), dickite (\(Al_2Si_2O_5(OH)_4\)) and nacrite (\(Al_2(Si_2O_5)(OH)_4\)). These are clay minerals that we can differentiate by the cristallisation level. Values around 0,02 and under indicate a low content of the precedent minerals.
Kaolin composition#
This column give the composition and cristallinity of the kaolin group minerals ranging from well-ordered kaolinite to halloysite to dickite and finally to nacrite. Low values correspond to a low cristallinity level (kaolinite) and high values correspond to a high cristallinity level (dickite and nacrite).
wmAlsmai (white micas and Al-smectite abundance)#
This column give the abundance of white micas (illite, muscovite, paragonite, brammalite, phengite, lepidolite, margarite) and Al-smectites (montmorillonite, beidellite). Values around 0.02 and under indicates a low content of the precedent minerals.
wmAlsmci (white micas and Al-smectite composition)#
the MFEM permits to determine the composition of the sample in white micas and Al-smectite. White micas minerals depend of the Tschermak substitution (it changes the total number of Mg and Fe atoms in mineral formulas) and we have detected paragonite, brammalite, illite, muscovite and phengite. Concerning Al-smectites minerals, beidellite and montmorillonite has aroundve been detected. Values around 2180nm indicates a sample Al-rich in mica (muscovite, illite, paragonite, brammalite, lepidolite) while values around 2220nm indicates a sample Al-poor mica (phengite). Values are given in nanometre and the accuracy of this method is very high.
carbai3pfit (Carbonates abundance)#
This column measure the ratio of carbonates/MgOH-bearing silicates and is based on the left-asymmetry of the CO3 feature. Values around 0.05 and under indicate a low content in carbonates. Values indicating a high content in carbonate are not specified and depend on the other drill cores results.
carbci3pfit (Carbonates composition)#
The MFEM works well to distinguish some carbonates, especially magnesite, dolomite and calcite but can’t be used to recognize all of them. Values in this column are given in nanometre. Around 2303nm it is magnesite, around 2326nm it is dolomite and around 2343nm it is calcite.
(The Rocklea Dome 3D Mineral Mapping Test Data Set, Table 1. Base scripts and multiple-feature extraction method scripts used for the Rocklea Dome 3D Mineral Mapping project)