Quality Control Plot
The geochemical analysis of the GeoCore X10 combines X-ray Computed Tomography (XCT) and X-ray Fluorescence (XRF) spectroscopy, through a mineralogical model fit that matches simulated minerals to these observables. The quality of the model and its results are therefore highly dependent on the suitability of the applied mineral list (see Element Concentrations and Scan Information) and how well it can be fitted to the physical contents of the scanned drill core samples.
As a measure of the mineralogical model fit's performance, the Quality Control plot view, available via Show → Quality Control (QC) in the menu bar, displays (rescaled) residual values for the different types of measurement variables:
The residual for each observable is individually rescaled into a unity range (0 to 1), as described below.
A low (high) residual value is indicative of good (bad) performance, but cannot be taken as a conclusive demonstration thereof, since there are many possible causes for variations in each residual.
However, large residual values may often indicate that improved results could be obtained with a better adapted mineral list, or point toward areas of different / unexpected mineralogy. Alternatively, they may also indicate areas that are inherently difficult to model, such as material which is too massive for accurate reconstruction with the XCT.
The residual values are only available for sections where the XRF signals are not degraded (see Degraded XRF Signals).
The mass residual measures the difference between the estimated mass (modelled value), obtained from tabulated densities combined with mineral mixture ratios from the mineralogical fit, to the measured mass (observed value), as directly measured from the loadcell at scan time.
The rescaled residual value scales linearly from exact agreement, to a mismatch which is equal to (or greater than) 20 % of the measured mass.
Unlike the estimated density which is available in higher resolution (see Rock Density ), the mass residual is only available as one value per sample, since there is only one measured mass per sample.
The XRF residual measures the difference in element specific XRF signals between the fitted minerals (modelled value) and those obtained from the spectrometer(s) at scan time (observed value). The modelled values are obtained from a lookup table, produced from Monte Carlo simulations of electron and photon transport in ideal minerals, combined with the mineral mixture ratios of the mineralogical fit. The observed values are obtained by fitting of spectral profiles at element-specific energies, to the spectrometer readings subjected to background substraction.
The rescaled residual value scales linearly from exact agreement, to a mismatch which is equal to (or greater than) the 95th percentile of unscaled residuals, taken over all samples included in Orexplore's internal reference library for analysis performance evaluation.
Note that the mineralogical fit in fact considers one signal per detectable element (see Element Concentrations), but that the XRF residual, for comprehensibility, summarizes the them into a single value, as a weighted sum over the squared signal differences.
The XRF residual is therefore an indicator of the overall performance for all detectable elements, which does not necessarily reflect the performance for individual elements.