Grain-size dependence of sediment composition and environmental bias in provenance studies

This article investigates both experimentally and theoretically the compositional changes associated with textural effects and hydraulic sorting during sediment transport and deposition, which cause systematic distortion in quantitative provenance analysis (environmental bias). Traditional procedures aimed at eliminating textural noise find limited success. The Gazzi-Dickinson method cannot remove hydrodynamic-related modal variability. Multiple size-window strategies are time-consuming. Narrow or moving size-window strategies represent misleading or impractical short cuts, being less convenient options than simply analysing each sample in bulk. New concepts introduced here unravel the superposed causes of compositional variability in modern sediments. Intrasample modal variability, fundamentally explained by settling-equivalence relationships, can be accurately modelled mathematically. Intersample modal variability, principally resulting from selective entrainment, can be assessed and removed by a simple principle. In absence of provenance changes and environmental bias, the weighted average density of terrigenous grains (SRD index) should be equal, for each sample and each grain-size class of each sample, to the weighted average density of source rocks. By correcting relative abundances of detrital minerals in proportion to their densities, we can restore the appropriate SRD index for any provenance and subprovenance type in each sample or grain-size class. Modal variability is effectively reduced by this procedure, which can be applied confidently to modern sediments deposited by tractive currents in any environment. Good results are obtained even for placer sands and finest classes where heavy-mineral concentration is strongest. Such SRD correction also successfully compensates for biased narrow-window modes, thus providing a numerical solution of general validity to the problem of environmental bias in sedimentary petrology. After compensating for settling-equivalence and selective-entrainment effects, residual size-dependent compositional variability may be provenance-related. Minor in Ganga-Brahmaputra sediments, provenance-related effects are spectacularly displayed in the Nile basin, where volcaniclastic silt mixes with basement-derived quartzofeldspathic sand and wind-blown Saharan quartz. (C) 2008 Elsevier B.V. All rights reserved.