Chemical composition of Earth's primitive mantle and its variance: 1. Method and results

We present a new statistical method to construct a model for the chemical composition of Earth's primitive mantle along with its variance. Earth's primitive mantle is located on the melting trend exhibited by the global compilation of mantle peridotites, using cosmochemical constraints on the relative abundances of refractory lithophile elements (RLE). This so-called pyrolite approach involves the least amount of assumptions, thereby being probably most satisfactory compared to other approaches. Its previous implementations, however, suffer from questionable statistical treatment of noisy geochemical data, leaving the uncertainty of model composition poorly quantified. In order to properly take into account how scatters in peridotite data affect this geochemical inference, we combine the following statistical techniques: (1) modeling a nonlinear melting trend in the multidimensional compositional space through the principal component analysis, (2) determining the primitive mantle composition on the melting trend by simultaneously imposing all of cosmochemical constraints with least squares, and (3) mapping scatters in original data into the variance of the final model through the bootstrap resampling technique. Whereas our model is similar to previous models in terms of Mg, Si, and Fe abundances, the RLE contents are at similar to 2.16 +/- 0.37 times the CI chondrite concentration, which is lower than most of previous estimates. The new model is depleted by > 20% in a number of incompatible elements including heat-producing elements, U, Th, and K, and this depleted nature is further amplified (up to 60%) in terms of predicted composition for the present-day mantle.