Testing the existing mineral interpretation workflow on hyperspectral imagery of the new Specim camera of Deep Atlas

F.J.A. van Ruitenbeek, Wim Harry Bakker, Wijnand van Eindthoven

Research output: Book/ReportReportAcademic

Abstract

In this study we have tested the calibration, processing, and mineral interpretation routines in the hyppy-based workflow, developed in previous studies for DA (Deep Atlas) on the new imagery generated by the Specim camera of DA. Differences in the instrument specifications and calibration and the wavelength, summary and mineral maps and other derived products obtained with the DA and UT (University of Twente) cameras were described and quantified where possible. Solutions to the observed differences were proposed to make the mineralogical interpretation system compatible with the new camera of DA.
The following differences between the images and derived products acquired by the UT camera (on which the workflow was developed) and the DA camera were observed: i) A different number of bands, ii) different center wavelengths of the bands, iii) a different accuracy of the wavelength calibration of the bands and iv) changes to the general shape of the spectra. Another difference was the spatial resolution of the images acquired by the two cameras; that of the images acquired by the UT camera was higher (200 µm) than that of DA camera (~350 µm), the value was estimated and has to be verified).
The different number and wavelength positions have a minor effect on the results produced by the mineral interpretation workflow. The difference in the accuracy of the wavelength calibration is more prominent since it produces changes to the wavelength positions of absorption features and in the classification of some minerals (e.g., kaolinite). The change of spectral shapes causes changes to the continuum removed depth of absorption features used for calculating the clay content. The changes in spectra shapes also impact on the Shannon entropy values. The lower spatial resolution of the images of the DA camera result in less detailed mineral maps.
Five of the seven spectral parameters that were used to model rock permeability (see the previous study for DA, i.e., van Ruitenbeek et al., 2021) showed different values in the images acquired by the two different cameras: The wavelength position of deepest absorption features between 1300-1600 nm, 1850-2100 nm and 2100-2400 nm, the Shannon entropy and the spatial entropy. The ferrous drop and the illkaol2 ratio only showed very minor deviations.
We conclude that, even though the two Specim cameras of the UT and DA are of the same type, there are slight differences in the instrument specifications and the calibration of bands and spectra. These result in differences between the hyperspectral images acquired by the two cameras. Although the differences are minor, they impact some of the results of the (semi-) automated workflow and the spectral parameters used to predict rock permeability. Correcting these differences using the recommendations given in this report is advised to get a higher degree of consistency between the results obtained with the two cameras. Applying these corrections will make the results obtained with the DA camera similar to those obtained with the UT camera.
Original languageEnglish
Number of pages29
Publication statusPublished - 25 Aug 2022

Fingerprint

Dive into the research topics of 'Testing the existing mineral interpretation workflow on hyperspectral imagery of the new Specim camera of Deep Atlas'. Together they form a unique fingerprint.

Cite this