Search

The general protocol for quantitative petrographic description of clastic sediments envisioned by J.C. Griffiths approximately 60 years ago is built on specification of the following five statistically independent fundamental properties of aggregates: modal composition, spatial arrangement, size distributions, shape distributions, and orientation distributions of the elements. Each element is represented by a grain falling into one of an exhaustive set of petrographic classes. We introduce the so-called Generalized Griffiths Descriptor (GGD) which extends this protocol to igneous and metamorphic rocks. Statistical independence of the fundamental properties permits rigorous testing of petrogenetic hypotheses as well as simulation of rocks based on any conceivable combination of them. The benefits of a unifying scheme for description of crystalline parent rocks and their weathering products are illustrated by exploratory analysis of a petrographic data set derived from five granodiorites from across North America, comprising the parent rocks and three size fractions (coarse, medium, and fine) of their detritus. Compositional data analysis is used to estimate sediment generation and evolution (SGE) trajectories for each of the plutons sampled. The bulk mineral composition and the mineral composition of the rock-fragment assemblage display a common trend of relative depletion of quartz and K-feldspar with decreasing grain size, compensated for by relative enrichment of biotite and accessory minerals. The evolution of crystal interface composition suggests that the balance between chemical and mechanical weathering may be reflected in relative stabilities of crystal interfaces. We propose a computationally efficient stereological inversion procedure for estimating the parameters of crystal-size probability distributions in rocks based on the assumption of lognormality. The lognormal distribution is an acceptable default model for crystal size in granitoids at the level of individual thin sections (about 87% passes the tests for lognormality), but less so at the level of entire plutons (only 62% success ratio). The current implementation of the GGD is not yet complete, because it deals explicitly with the first three fundamental properties only, as a consequence of adopting the simplifying assumption that all elements which make up a rock are spherical. Shape and orientation distributions may be included in future implementations of the GGD. Dedicated data sets comprising sediments generated from a single parent rock under different conditions will be needed to advance SGE studies by shedding more light on the balance between chemical and mechanical weathering. [ABSTRACT FROM AUTHOR]