An investigation of the magnetotelluric method for determining subsurface resistivities.

The magnetotelluric method, which depends upon the determination of impedance values over a wide frequency range (0.001-1 cps) from a pair of orthogonal electric and magnetic field components, has been used in the past by several investigators to determine the resistivity of the crust and upper mantle. Because of the diversity of the results obtained by the magnetotelluric method it was felt necessary to examine the method critically in order to obtain useful and unambiguous results. To carry this out an investigation was made of the magnetotelluric field recorded simultaneously at six stations in central Alberta during August 1961. The investigation is divided into five main sections; the recording of the magnetotelluric field, the analysis of the field records by various methods, the evaluation of the validity of the different assumptions made in the magnetotelluric method, the determination of subsurface resistivities, and the investigation of inhomogeneous and anisotropic bodies. Continuous recordings of Ey, Hx and Hz were made during August 1961 for two weeks at six stations, each approximately 100 km apart and oriented in a north-south direction (113.5° W longitude). In addition two extra components Ex, Hy were recorded at the central station, Beiseker. Estimates of the wave impedance Ey/Hx were obtained by inspection of quasi-sinusoidal events on the records from Meanook and Cardston. Using Cagniard's method an estimate of the subsurface resistivity ρ was made at Meanook. No estimate could be made at Cardston because of the large scatter of points in the plot of Ey/Hx against period T, Subsurface inhomogeneities near Cardston are believed to be the main cause of this scattering. At Beiseker, power spectra of selected lengths of records were computed and from them the ratios Ey/Hx and Ex/Hy were obtained in order to estimate subsurface resistivities. In addition a method for interpreting anisotropic bodies has been suggested and used at Beiseker to explain the differences which exist between the ratios EV/HL and Ex/Hy. A comparison between the various methods suggested by different investigators to interpret magnetotelluric data has been made and it has been shown with the help of theoretical models that these methods have no advantage over the curve matching method suggested by Cagniard. Moreover, it has been shown that such methods may give ambiguous results if applied to the interpretation of high frequency ( > 0.005 cps) magnetotelluric data. In order to judge the validity of the basic assumption of Cagniard's method, viz. that the horizontal gradients of the field vectors are negligible compared to vertical gradients, power spectra of corresponding lengths of records, used for the estimation of the resistivities, were computed at all six stations for the components Hx and Hz. Micro-pulsation activity which exhibited high coherence of Hx at all six stations yielded the least scatter in the ρ vs T plot as was expected. By carefully selecting data on the basis of this and other coherence criteria it is believed that a reliable indication has been obtained of a marked decrease in the resistivity in the upper part of the Earth's mantle.Science, Faculty ofPhysics and Astronomy, Department ofGraduat