A reassessment of spectralTeestimation in continental interiors: The case of North America

Abstract

Conventional spectralTestudies use the real part of the admittance between gravity anomalies and topography or, alternatively, the square of the magnitude of the coherency (i.e., coherence). Here we show the utility of treating both the admittance and coherency as complex quantities. Inverting the real parts to estimateTe, we use the imaginary parts to tell if the inversion is biased by noise. One method inverts the square of the real coherency, with the internal‐to‐total load ratioFderived (as a function of wave number) directly from the gravity and topography. The other method inverts the real part of the admittance assuming thatFis wave number‐independent. We test the methods using synthetic elastic plate models loaded at the surface and Moho in such a way that the final relief is the actual North American topography. In some of the models we add gravity noise generated by a model having both surface and internal loads such that the final topography is zero and find that both methods are susceptible to noise. Application of the two methods to North America givesTemaps showing substantial agreement except in regions affected by noise, but these are not a dominant part of the total area. Given the suggested mechanisms by which noise might arise, it is not surprising that it is not a more widespread feature of the North American craton. Importantly, both methods show that large parts of the Canadian Shield are characterized byTe> 100 km.