Tensor CSAMT survey over the Sulphur Springs thermal area, Valles Caldera, New Mexico, United States of America, Part II: Implications for CSAMT methodology

The resistivity model for the Sulphur Springs area in the companion paper (Part I) plus the availability of overlapping controlled‐source audiomagnetotelluric (CSAMT) and magnetotelluric (MT) data has allowed study of far‐field to near‐field transitions, source field geometries over the survey area, and scalar‐tensor impedance discrepancies. The regional setting of conductive Paleozoic sediments over resistive basement seriously reduced depth of exploration within the plane‐wave regime to about 1/20th of the transmitter‐receiver separation, rather than the traditional 1/3rd to 1/5th based on half‐space models. As frequency falls to where skin depth in the sedimentary layer exceeds its thickness, transmitter electromagnetic (EM) fields enter the resistive basement and may diffuse to the receiver with relatively little attenuation, promoting near‐field behavior. Comparisons are made of observed electric (E) and magnetic (H) fields inside and outside the caldera with EM fields computed from layered resistivity models derived from local 1-D inversion of the ρa and θ, and from simple 3-D models. First, the comparisons indicate that small‐scale structure near the transmitter does not lead to overprint effects in the impedance data at the receiver but, instead, acts as an equivalent far‐field source. Second, at both high and low frequencies, the observed E and H fields can depart substantially from those predicted by local layered models. In fact, an effective regional layering appears to control the magnetic field amplitudes and the far‐to near‐field transition in this survey area. The observed electric fields, on the other hand, are controlled by all scales of geology. When heterogeneity is important, significant departures between scalar and tensor CSAMT data can be expected, and are exacerbated when the source field is poorly coupled to the sensors. The problem is much reduced for vector CSAMT measurements where all horizontal field components are measured and the maximally coupled results are defined, but mode identification is more difficult for multidimensional structures.