The effects of mantle compressibility on geoid and topographic signatures caused by internal loading in the mantle are studied. For a Cartesian compressible model the Green's functions of these geophysical signatures are determined by solving a two-point boundary value problem with depth for each horizontal wave number associated with the density anomaly. Various types of equations of state have been considered, which include both constant and depth-dependent thermodynamic parameters. Surface topographies are not changed too much by the effects of mantle compressibility, even for long wavelengths. The deformation of bottom boundary is not seriously influenced by mantle compressibility, except for long wavelengths and large viscosity contrasts between the upper and lower mantles. Geoid signals with horizontal wavelengths, exceeding 10,000 km, can be greatly influenced by mantle compressibility. For large viscosity contrasts, differences of up to 100% between incompressible and compressible geoid responses can be found. The admittance function also shows great sensitivity to mantle compressibility at long wavelengths. There are significant differences in the geophysical signatures from the ways in parameterizing the viscosity stratification in the mantle. An exponentially dependent viscosity model produces smaller geoid and surface topographical signals than for a single step function viscosity model.
INTRODUCTION
In the past decade there have been growing efforts in using long-wavelength seismic anomalies in the mantle to interpret geoid signals on the basis of dynamical circulation models driven by internal density heterogeneities [Hager, 1984; Richards and Hager, 1984; Ricard et al., 1984; Hager et al., 1985; Forte and Peltlet 1987; Ricard et al., 1989; Ricard and Vigny, 1989]. It now appears possible to model and explain the long-wavelength portion of the geoid spectrum by using seismic tomographic results [Woodhouse and Dziewonski, 1984; Tanirnoto, 1987]. To a first approximation, a physical understanding of the relationship between seismic and geoid anomaly observables has now been derived. The paradox concerning the negative sign of the predicted long-wavelength geoid [Dziewonski, 1984] can be explained by taking into account the radial dependence of the viscosity structure in dynamical models with internal loading [Richards and Hager, 1984; Ricard et al., 1984; Lago and Rabinowicz, 1984]. However, there are many different ways in which a viscosity model can be parameterized [Hong and Yuen, 1986; Revenaugh and Parsons, 1987], and there exist invariably trade-offs between different viscosity structures, which can lead to nonunique solutions able to explain most of the data equally well [Ricard et al., 1989]. Calculations of geoid anomalies and topographies from internal heating have, up to now, been based on the incompressible (Boussinesq) approximation. However, the density in the mantle increases with depth because of compressibility. Hence it is necessary to evaluate the effects on surface signatures from ...