UBC Theses and Dissertations
Mineral physics constraints on the chemical composition and temperature of the earth’s mantle Calderwood, Arthur Robertson
Numerical mineral physics modeling is used in seven studies to constrain the composition and dynamics of the Earth's mantle. First, using updated, published experimental data, I evaluate two competing mineralogical models for the composition of the mantle by calculating and comparing one dimensional profiles of density (ρ) and bulk sound velocity (V[sub Φ]) against seismologically inferred profiles. I find that the mineralogy of a uniform bulk composition provides an excellent fit to the seismic properties of the whole mantle along a self-consistent adiabatic geotherm which lacks any thermal boundary layers internal to the mantle. In contrast, the mineralogy of a chemically layered bulk composition results in predicted profiles which cannot satisfy the observations. Second, to further test the predicted properties of the uniform composition, pyrolite model, the mineral physics calculations are extended to predict one dimensional profiles of shear (V[sub S]) and compression (V[sub P]) velocity across the upper mantle and transition zone. Previous studies that modeled V[sub S] and V[sub P] suffered from having to estimate many of the necessary temperature and pressure derivatives required in the calculations and hence, the results were ambiguous. Fortuitously, many of these critical material properties (μ, (მμ/მT)[sub P,] (მμ/მP)[sub T],) are now experimentally measured and this study incorporates the latest results. I find that the predicted profiles of V[sub S] and V[sub P] are in excellent agreement in terms of both velocity gradient and absolute velocity with the latest, high-resolution regional seismic profiles. The predicted mineral physics profiles also give a mutually consistent estimate of the olivine content of the mantle from the velocity jumps at the 410 km seismic discontinuity. Lastly, the predicted V[sub S] and V[sub P] normalincidence reflectivity of the 410 km, 520 km and 660 km discontinuities matches a number of independent seismic estimates. Third, I estimate a high temperature value for [მμ/მT)[sup Mg-Pv][sub p] of Mg-perovskite via constrained adiabatic decompression of the lower mantle PREM seismic properties. A simplified pyrolite lower mantle mineralogy of magnesiowustite and Mg-Fe perovskite is used to provide apriori constraints for some of the unknown high temperature decompressed properties thereby restricting the possible parameter space for the decompressed hot shear modulus (μ[sub O]). This property is used to constrain the value of [მμ/მT)[sup Mg-Pv][sub p] when the room temperature properties are corrected up to the decompressed, but hot lower mantle elastic properties. Fourth, adopting this estimate of [მμ/მT)[sup Mg-Pv][sub p],and estimating ([მμ/მT)[sup Ca-Pv][sub p]) via a high temperature thermodynamic approximation, I then iteratively adjust the unmeasured pressure derivative, [მμ/მP)[sup Mg-Pv][sub T] , under the first order assumption [მμ/მP)[sup Mg-Pv][sub T]= [მμ/მP)[sup Ca-Pv][sub T] , until predicted one dimensional profiles of V[sub S] and V[sub P] for a pyrolite mineralogy match the seismic profiles across the lower mantle. I justify "tuning" the predicted profiles to the observed ones on the basis of the results of chapter two where it was demonstrated with ρ and V[sub Φ] profiles that a pyrolite mineralogy could match the lower mantle properties along a self-consistent adiabatic geotherm. Because the selfconsistent adiabatic geotherm and all of the thermoelastic properties used earlier in calculating ρ and V[sub Φ] are also used to model shear and compression velocity, the predicted V[sub S] and V[sub P] profiles must also agree with the observed profiles. Fifth, I incorporate the recent experimental (Sinelnikov et al., 1998) and theoretical (Karki et al., 1997) values for Mgperovskite and compare these with the inferences deduced in the previous chapters. While the estimated temperature derivatives compare favorably with the with the recent experimental values, I find that to satisfy both the experimental and theoretical pressure dependence, Mg-perovskite must be fit by a 4th order, rather than 3rd order, Birch-Murnaghan equation of state. Sixth, I utilize the pyrolite V[sub S] and V[sub P] profiles to evaluate the ratio of relative V[sub S] and V[sub P] velocity heterogeneity in the mantle via the parameter (მlnV[sub S]/ მlnV[sub P]) . I find that the predicted profile of (მlnV[sub S]/ მlnV[sub P])[sub p] for the pyrolite mantle is within error of the various seismic estimates, and hence, there is no discrepancy between the tomographic models and a melt-free, uniform bulk composition mantle. Seventh, I estimate new extremal bounds for the lateral temperature anomalies in the mantle using the pyrolite mineral physics model of V[sub S] and V[sub P]. I find that the inferred magnitude of the lateral temperature anomalies are significantly reduced compared to previous inferences with δT now ranging from ~ 100 °K at 660 km to ~ ± 480 °K at the top of the D" layer. I conclude that away from chemical and thermal boundary layers, the lateral seismic anomalies resolved in tomographic studies are largely thermal in origin and consequently, the dominant force driving mantle convection is indeed thermal buoyancy.
Item Citations and Data