Case studies on the origin of basalt

Abstract

This paper presents a method for the systematic trace element modelling of a cogenetic suite of lavas. It is based on the geochemical inversion technique of Allègre and coworkers and utilizes the variations in the trace element concentrations of the lavas to calculate initial concentrations and source mineralogy. We reduce this inversion to a simple, step-by-step procedure: (1) correcting for fractional crystallization; (2) testing the inferred primary melt compositions for consistency with a model of equilibrium partial melts (with constant partition coefficients) formed from identical sources; (3) estimating the proportions of mineral phases entering the melt; (4) computing concentrations and bulk partition coefficients in the initial source relative to the concentration of a common reference element; (5) estimating relative mineral abundances in the source. Except for the fractionation correction, the calculations are done element by element using a direct analytic solution. For the purpose of comparison we apply this method to the same set of data used by Minster and Allègre (1978), a suite of lavas from Grenada (lesser Antilles) originally analyzed by Shimizu and Arculus (1975). The results of both methods agree well for the source abundances of the light REE, whereas the heavy REE abundances are shown to be poorly constrained by the data. Both methods require residual clinopyroxene and garnet in the source, but the ratio of these minerals is not well constrained. We are unable to reproduce the shape of D₀ pattern (=bulk partition coefficients of the initial source) given by Minster and Allègre. The reason for this cannot be evaluated without repeating their calculations in detail.

The set of data from Grenada is useful for comparison of the methods only, because it is now known from isotopic data that the samples are not truly cogenetic. Possibly better suited sets of samples for petrogenetic modelling are presented in parts II and III of this series.