The long wavelength component of mantle convection

Abstract

Theoretical models of mantle convection predict that the major temperature fluctuations within the mantle are confined to narrow horizontal boundary layers and vertical plumes. These fluctuations result in heterogeneities in seismic body wave velocities which could, in principle, be detected by seismic tomographic techniques. However, recent tomographic images of the spatial variations of temperature in the mantle are considerably “out of focus” in that only the longest wavelength components can be resolved. To assess this partial recovery of the total tomographic image, theoretical temperature fields have been generated with a numerical model of high Rayleigh number mantle convection and then Fourier analysed in two dimensions. Upon re-synthesizing the model temperature fields, the Fourier series expansions were truncated at various levels of resolution. The truncated expansions, containing only the long wavelength components of the model temperature fields, are compared to both the complete field and the tomographic images of the mantle. At the current level of resolution it seems unlikely that seismic tomography could distinguish between layered and whole-mantle convection. Estimates, based on current tomographic data, of long wavelength fluctuations of temperature and surface topography are predicted, in the case of whole-mantle convection, to represent approximately 10% of the total temperature variation, and approximately 50% of the total topographic relief. Thus topography at the core-mantle boundary may be more accurately inferred from seismic tomography than may the characteristic lateral temperature fluctuation which drives the convective circulation.