In a companion paper (Egbert et al., this issue) we describe the estimation of very long period (0.16<T<91 days) magnetotelluric (MT) impedances from 11 years of data collected at the Tucson geomagnetic observatory. Here we discuss the implications of these data for mantle conductivity. Using minimum norm (flattest and smoothest) inversions, we find simple one-dimensional models of electrical conductivity in the depth range 0--1500 km. We use forward modeling, a linearized resolution analysis, and constrained one-dimensional inversions to delineate the range of models which are consistent with the estimated impedances. Although the MT data have limited resolution, large-scale vertical averages of mantle conductivity are well constrained. We reach the following conclusions concerning mantle conductivity beneath Tucson: (1) The upper 200 km has a conductance of order 104 Siemens (S). This anomalously high conductance may be concentrated in an aesthenospheric high conductivity layer, but the geometry of the conductive zone is not constrained. (2) Typical conductivities in the transition zone (400--700 km) are ≈0.1--0.3 S m-1. A step increase to reach this value at or near the 400 km olivine-spinel phase transition is consistent with, but not required by, the data. An upper mantle which is resistive throughout (0.05 S m-1 or less) is not allowed by the data. (3) Resolvable large-scale averages of conductivity increase from ≈0.2 S m-1 to ≈1.0 S m-1 between 600 and 900 km depth.

A range of models, including those with step increases, and step decreases, at the 670 km seismic discontinuity are consistent with the data. (4) Between 900--1500 km, conductivity increases slowly. Average conductivities in this region are of the order of 1 S m-1, to within a factor of 2 or 3. While limited zones of highly resistive mantle are consistent with the data, a lower mantle which is resistive throughout is not. Models in which conductivity is always above 5 S m-1 below 1000 km can also be ruled out. In conjunction with improved laboratory estimates of electrical conductivities of mantle minerals at high temperatures and pressures, these constraints can provide important clues to the composition and physical state of the mantle. ¿ American Geophysical Union 1992