Electrical resistivity imaging of contiguous United States from three dimensional inversion of EarthScope USArray magnetotelluric data

Abstract:

The USArray magnetotelluric (MT) transportable array (TA) component of EarthScope is a powerful tool for studying the electrical structure of North America on a continental scale. Between 2006-2018, a total of 1116 long periods MT (10 – 20,000 s) sites were occupied, extending from coast-to-coast, and covering much of the continental US. A number of regional-scale inversions of these MT data, each covering a different area, reveal significant large-scale variations in deep resistivity. These have mostly been interpreted as variations in thickness of the lithosphere (e.g., between the Basin and Range, and adjacent cratons), or as suture zones penetrating deep into the lithosphere. There are also significant differences in deeper asthenospheric resistivity, both within individual models, and between different studies. Such significant variations of deep resistivity, if real, are suggestive of significant variations in mantle hydration. Here we explore the robustness of this deep structure by constructing 3D inverse models for all of the continental US where MT TA data are available. Due to computational limitations we are forced to use a relatively coarse model grid (30 km), and to reduce to a carefully selected (and widely spaced) subset of MT TA sites. Despite the lower resolution of both model and data constraints, the main features of our model are quite similar to those from previous studies. To test robustness of deep resistivity variations we modified the model for depths below 200 km by replacing 3D resistivity values by layer averages. We then restarted the inversion with these 1D layers (below 200 km) frozen. For the modified model normalized root-mean-square (nRMS) misfit increased to 2.29 from 2.14 (5% error floors), but after reinverting, nRMS decreased to 2.07. Importantly, the main features shallower than 200 km are almost indistinguishable in models with and without deep structure constrained. Repeating this test with a depth cutoff of 150 km, we find significant degradation of fit when the 1D constraint is imposed, implying that resistivy variations below 200 km are not required by the MT TA data , but shallower structure is. We will also report on other efforts (e.g., use of phase tensor data from denser set of sites), to further explore resolution of deep structure.