Compositional linear modeling, non-central principal component analysis, Ω and relative variation curves: using higher dimensional space to unravel elemental mass redistribution patterns within the regolith


Tina Zeidan
MS Candidate
Advisor: Dr. Gary H. Girty

Friday, December 11th, 2015
CSL 422 – 11:00am
watch Tina’s defense

Within the arid climate of southern California, USA, we uncovered through trenching a granodioritic-tonalitic corestone, a partially encircling fractured rind, and an A2 layer composed of sandy soil.  Chemical data derived from an adjacent erosionally exhumed corestone was used as a proxy for the unweathered bedrock from which the above section of regolith was derived.  Using non-central principal component analysis, the compositional linear trend derived from the weathering of the granodioritic-tonalitic bedrock was calculated.  Average weathering intensities (t-values) and 95% confidence intervals derived from orthogonally projecting weathered samples on to the resulting compositional linear trend are as follows: uncovered corestone = 0.56 (± 0.15), inner fractured rind = 0.93 (± 0.08), outer fractured rind = 1.5 (± 0.06), and A2 layer = 1.8 (± 0.10).

Ti was immobile during weathering, and was used as a reference frame in our calculations of Ω, the modeled fractional change in a given elements mass relative to the geometric mean of its mass in the erosionally exhumed corestone.  Resulting relative variation curves were plotted on 10 bivariate graphs, where the x-axis represents t-values and the y-axis is Ω.  With the exception of P2O5, samples from each of the textural groups, clustered and plotted closely about the 10 relative variation curves.

Mass balance calculations are consistent with the modeled relative variation curves and the results of our XRD and petrologic studies.  Together they suggest that the progressive decrease in the loss of Ca and Na mass through the A2 layer, and outer and inner fractured rind, is the result of the incongruent dissolution of plagioclase.  This result implies an increase in pH through these three textural zones, while the lack of loss of Ca and Na mass within the corestone uncovered by trenching signifies slightly acidic to near neutral pH conditions. The products of acidic fluids reacting with plagioclase within the A2 layer and outer fractured rind included silicic acid and kaolinite.  Notably, though Si mass was lost in the above two textural zones, it did not accumulate in the adjacent inner fractured rind and corestone.

Acidic solutions percolating through the A2 layer and outer fractured rind also reacted with biotite leaching and removing K, Mn, and Mg mass, and transforming it into kaolinite.  The removal of K mass continued into the inner rind while the absence of a loss of Mn mass in the inner rind and corestone are consistent with increasing pH and near neutral conditions.  The continued loss of Mg mass within the inner rind and adjacent corestone reflects the highly soluble character of Mg, and the alteration of chlorite to vermiculite. Finally, the relative stability of Fe mass throughout the sampled regolith implies oxidizing conditions. In short, the clustering and ordering of samples analyzed from the uncovered corestone, inner rind, outer rind, and A2 layer about the relative variation curves suggests that the weathering of the section of regolith uncovered by trenching occurred in an orderly and sequential manner as the result of changing pH and fluid/mineral reactions.