The continental tectosphere and Earth's long-wavelength gravity field

Abstract

To estimate the average density contrast associated with the continental tectosphere, we separately project the degree 2–36 non-hydrostatic geoid and free-air gravity anomalies onto several tectonic regionalizations. Because both the regionalizations and the geoid have distinctly red spectra, we do not use conventional statistical analysis, which is based on the assumption of white spectra. Rather, we utilize, a Monte Carlo approach that incorporates the spectral properties of these fields. These simulations reveal that the undulations of Earth's geoid correlate with surface tectonics no better than they would were it randomly oriented with respect to the surface. However, our simulations indicate that free-air gravity anomalies correlate with surface tectonics better, than almost 98% of our trials in which the free-air gravity anomalies were randomly oriented with respect to Earth's surface. The average geoid anomaly, and free-air gravity anomaly over platforms and shields are significant at slightly better than the one-standard-deviation level: −11±8 m and −4±3 mgal, respectively. After removing from the geoid estimated contributions associated with (1) a simple model of the continental crust and oceanic lithosphere, (2) the lower mantle, (3) subducted slabs, and, (4) remnant glacial isostatic disequilibrium, we estimate a platform and shield signal of −8±4 m. We conclude that there is little contribution of platforms and shields to the gravity field, consistent with their, keels having small density contrasts. Using this estimate of the platform and shield signal, and previous estimates of upper-mantle shear-wave travel-time perturbations, we find that the average value of ϖln ρ/ϖln νs within the 140–440 km depth range is 0.04±0.02. A continental tectosphere with an isopycnic (equal-density) structure (ϖln ρ/ϖln νs=0) enforced by compositional variations is consistent with this result at the 2.0σ level. Without compositional buoyancy, the continental tectosphere would have an average ϖlnρ/ϖlnνs≈0.25, exceeding our estimate by 10σ.