Lowry A.R., Velicogna I., Duglas B.

Bulk rheologic properties of the lithosphere remain poorly constrained in studies of Earth deformation. In situ observations have been derived from geodetic strain, from studies of isostatic rebound, and from piezometric analyses of xenoliths, but the spatial coverage by each of these metrics is limited by signal-to-noise ratio and/or sampling. We present a technique for estimation of lithospheric viscosity at relatively high resolution, using a combination of spectral analysis of isostatic response and geotherms derived from surface heat flow measurements. We demonstrate, to very high confidence, that variations in geotherm are not sufficient to explain variations in isostatic response. Hence we parameterize the material property component of viscosity variation using a variable effective activation energy for the mantle lithosphere. Activation energy is just one of the variable material properties that can influence a ductile-regime constitutive relation, and so effective activation energy is a catch-all for properties such as grain size, defect structure, crystal water content, and true variation in activation enthalpy. Uppermost mantle in the western United States ranges from 300--400 kJ/mol in the actively deforming Cordillera to as much as 550 kJ/mol in the stable craton. Estimates of effective activation energy are sensitive to errors in estimates of geotherm from heat flow measurements. Estimates of viscosity are relatively insensitive to temperature errors, but are sensitive to error in the isostatic response function. Uppermost mantle viscosity is <~10^21 Pa s in deforming regions, but significantly higher in stable lithosphere. We compare and contrast estimates of viscosity from this method with other estimates of viscosity in the western U.S. derived from rebound of Pleistocene lakes, geodetic strain analysis, and xenolith recrystallized-grain-size piezometry.