Magnetotelluric data analysis: removal of bias

Two new techniques for analyzing 4‐channel magnetotelluric (MT) data are described. These techniques produce estimates of the elements [Formula: see text] of the impedance tensor that are unbiased by noise in the autopowers of the electric and magnetic fields. Effectively, each technique uses one field channel as a reference signal that can be correlated with the other three channels. Method 1 obtains estimates for the [Formula: see text] in terms of crosspowers of the Fourier components of the electric and magnetic fields [Formula: see text], [Formula: see text], [Formula: see text], and [Formula: see text]. Method 2 is a generalization of method 1, and obtains estimates for [Formula: see text] in terms of weighted crosspowers. Both methods fail when the geology is one‐dimensional, or two‐dimensional with one electrode oriented along the strike direction. To obtain results that are stable for any geology and that are unbiased by autopower noise, at least five channels of data are required. To also minimize bias by correlated noises, one needs six channels of data, two channels of which are for fields measured at a site that is remote from the base MT station. The analysis of MT data using a remote magnetometer as a reference is discussed.