Hjelt, S.-E., Heikka , J.V., Lakanen, E., Pelkonen, R. & , Pietilä , R., 1990. The audiomagnetotelluric (AMT) method and its use in ore prospecting and structural research. GeoJogian tutkimuskeskus, Tutkimusraportti 95, 69· 86, 13 figs . When the audiomagnetotelluric (AMT) method is used, several components of the Earth's natural electromagnetic field are registered simultaneously in the frequency band 1 to 10000 Hz. The primary field is produced mainly by lightning discharges and the secondary fjelds carry Information about the distribution of the electrical conductivity within the subsurface. In Finland, the scalar variant of the AMT method has been succesfully appl ied both in ore prospecting and structural research. The present paper reviews the main properties of scalar AMT measurements and the basic principles of data processing and interpretation. Three case histories are presented . A general structural study was made at Talvivaara in the Kainuu schlst belt. In the area of the Rautuvaara Iron mine, the scalar AMT technique was successfully used to locate gently dipping, conducting and ore-<:ritical horizons at depths ranging from a few tens of meters to about 500 meters. In a regional survey, covering about 15 km2 in the Hannukainen area, the AMT results were a relevant complement to other geophysical studies. Some results are presented from a study made at POlvljärvi, on the Outokumpu formation, where the scalar AMT data are compared with recordings using a electrical dipolar source (so-<:alled controlled source AMT = CSAMT measurements). The CSAMT technique has superior signal·to·noise ratios, but the interpretation becomes increasingly complex.

Abstract Key results from the BEAR (Baltic Electromagnetic Array Research) soundings and the SVEKALAPKO Seismic Tomography Experiment (SSTE), which explored geophysical properties of the upper mantle beneath the Fennoscandian Shield, are summarized here. These two projects formed the two key geophysical experiments in EUROPROBE's SVEKALAPKO project. The major result obtained from the teleseismic tomography indicates P- and S-wave velocity variations of up to 4%, as compared with the IASP'91 global model. The positive P-wave velocity anomaly seems to extend down to 300 km, without any indication of an asthenospheric low-velocity layer. This is corroborated by surface-wave analysis. The Archaean-Proterozoic suture zone has no continuation at upper mantle depths. Instead, a laterally and vertically heterogeneous structure of the subcontinental lithospheric mantle in the contact zone of Archaean and Proterozoic domains beneath the SVEKALAPKO area has been revealed by both teleseismic and local event studies. Comparison between stacked receiver functions from the TOR (Teleseismic Tomography of TORnquist Zone) and SSTE arrays indicates that the difference between arrival times of the converted P410 and P660 phases increases below the Shield as a result of a cooler upper mantle. The asthenosphere beneath Central Europe terminates at the southern edge of the shield and cannot be identified in the SVEKALAPKO data. The variations in Moho depths from 30 km to 50 km (locally, more), known from earlier deep seismic sounding (DSS) work, were corroborated by the receiver function analysis. The BEAR (Baltic Electromagnetic Array Research) experiment data indicate that an upper mantle conducting layer is required in some places beneath Fennoscandia. In the northern part of the Shield the upper mantle conductor is located at a depth of c. 170 km. Magnetotelluric data exhibit strong anisotropic behaviour, in particular in the central parts of the Fennoscandian Shield. The present state of modelling implies, however, that isotropic 3D crust and upper mantle explain nearly all observed anisotropic features and that very minor anisotropy of the upper mantle is required to explain the data. The crust is strongly variable electrically, ranging from resistive areas to highly conducting elongated structures. The conductances (= conductivity x layer thickness) obtained range from a few Siemens to several tens of thousands of Siemens. The conductive structures delineate boundaries between Archaean and Proterozoic crustal units and are inferred to image relics of subduction processes. In the central parts of the Shield, the lower crust is conducting, with dipping structures extending far to the NE below the Archaean-Proterozoic boundary. Thus, the present interpretation of electromagnetic BEAR and tomographic data from the Fennoscandian Shield demonstrates that the structure of the upper mantle beneath the Shield is much more heterogeneous than was supposed when the project started. Models of the evolution of lithosphere must be revised to accommodate lateral and vertical heterogeneity in the upper mantle.

The Svecofennian‐Karelian‐Lapland‐Kola Transect (SVEKALAPKO) project is one of the five multidisciplinary key projects of Europrobe, a scientific program of the European Science Foundation (ESF) that studies the tectonic evolution of European continental lithosphere [Gee and Zeyen, 1996]. The SVEKALAPKO project [Hjelt and Daly, 1996] has adopted a multidisciplinary approach that uses geological, penological, and geophysical methods to unravel the evolution of the crust and lower lithosphere in three major crustal segments of the Fennoscandian Shield: the Proterozoic Svecofennian and Lapland‐Kola orogens and the intervening Archaean Karelia craton. Improved knowledge of the structure and evolution of the Fennoscandian Shield should lead to a better understanding of plate‐tectonic processes in the early history of the Earth.

To proceed with any optimization technique, a measure of the distance between the observation and the response calculated from the model is required to be defined. This measure is called 'Objective function' or 'Misfit function' (MF, used in this paper). Several norms (l 1 , l 2 ,...,l p , etc. ) can be used to define this function. The behaviour of MF over a space described by a combination of model parameters can be displayed in the form of a MF topography (MFT). Construction of the MFT requires a thorough scanning of a vast model space. The MFT maps can provide valuable clues on the correlation between the parameters defining the model space. For a dynamic (time-variant) system, such as the transient electromagnetic (TEM) method, the MFT is found to vary as a function of time. Thus it is possible to select suitable time-windows of observation which provide optimal resolution of the model parameters. Since various TEM systems employ different types of exciting pulses (sinusoidal, ramp, triangular, etc.) and recording time channels, the MFT's can also be exploited in comparing the performance of various TEM systems. Also, since MFT's are based on the scanning of a vast model space, in essence they represent a process similar to the grid search technique of optimization commonly used in finding non-linear parameters from geopysical data. The above applications of the MFT are shown considering synthetic responses of i) a perfectly conducting half-plane (minimal) model in frequency domain and ii) a conducting finite plate model to commonly employed generic TEM systems. Both these models are assumed to be immersed in a non-conducting medium for the sake of simplicity. A field example from the nickel sulphide deposit, Mt. Keith South, Western Australia is also studied to demonstrate how the MFT maps can be used to obtain the model parameters.

Sensitivity analysis is made for transient electromagnetic (TEM) profile data using a thin plate model. Five commonly used measurement geometries are compared with each other and coincident loop system is studied in detail. The sensitivity analysis is made using the singular value decomposition (SVD) of the partial derivative matrix which is computed numerically. Forward solution for a thin plate model in free space is computed using the PLATE program. The following parameters are studied: strike length, depth extent, x-position (epicentre of the midpoint ofthe top edge of the plate), depth of burial, dip angle and conductance. The SVD, the singular values and the parameter and data eigenvectors, yield information about the importance of each parameter and the correlation between the parameters. The correlation studies are particularly useful in understanding the properties and behaviour of model parameters. This information can also be employed to compare different TEM systems from the inversion point of view.

Inversion of the geophysical data can be made more efficient and effective by incorporating the a priori information about the model parameters as a probability distribution. The mean of this dist ribution is expected to represent the value of the model parameter vector while its shape describes the certainty of this expectation. In this paper it is shown that the topography of the misfit function can provide useful information in selecting the a priori probability distribution. The shape of the misfit function topography also describes the correlations among various model parameters. Thus it is possible to study the strength of linear, non- linear or null dependence of various model parameters on each other. Study of the misfit function topography thus yields vital clues for effective model optimization of geophysical data.

Information on electrical conductivity of the subsurface at a crustal scale is inferred mainly from magnetotelluric (MT) and magnetometer data. The latter include both magnetometer array studies (MV) and geomagnetic depth soundings (GDS). Locally, controlled source data (e.g. data from d.c. and very low frequency (VLF) resistivity surveys, frequency soundings and airborne electromagnetic mappings) provide information on near-surface structures, e.g. on exposures of deeper conductors. The variety of electromagnetic methods provides an efficient zooming ability for structures from a regional scale of hundreds of kilometres to small local details of some metres. Thus both the mapping of large-scale crustal conductivity structures and a detail study of the anomalous structures are possible.The electrical conductivity of the crust and upper mantle is described within the plate tectonic framework: examples of structures which become electrically conductive at different phases of the process are shown. Using the results of electromagnetic soundings in various tectonic environments in Europe it is shown how the crustal conductors can be interpreted as ancient and modern tectonic markers. These include structures from the Fennoscandian Shield (Precambrian extensional basins and collisional zones), the Carpathians, Scotland and Ireland (suture zones), and the Pannonian Basin (extensional basin and thrust zones), as well as conductors detected along the European Geotraverse (e.g. the terrain boundaries and the decollement of the Variscan thrusting).An example which considers the lower-crustal conductivity from the Fennoscandian Shield shows rather large lateral variations in conductance of the lower crust varying from 1 S to several hundreds of siemens. Estimates of the dept of the asthenospheric conducting layer within the European territory also vary considerably. Shallowest depths are detected beneath extensional regions (the Pannonian Basin and the Rhinegraben; 40-80 km), whereas depths of 200 lan or over (or the absence of a conducting layer) are detected beneath the Alps and old Precambrian cratons.Among the key areas where electromagnetic investigations should play an essential role are the deep structure of the East European platform (boundaries of Archaean and Proterozoic terrains), the structure of the Teisseyre-Tornquist zone, the nature of the present-day continental lower crust, and the characteristics of the electrical asthenosphere beneath Europe.

Large resistive regions (ideally having a horizontal size of about 200 km) are needed to study reliably the conductivity distribution of the lower crust of the Baltic (Scandinavian) Shield by deep electromagnetic (EM) methods. The overburden of the shield consists of Quaternary sediments, and is thin with a low conductance. This makes the shield favourable for EM studies. The numerous conductive zones (e.g., schist belts) in the uppermost crust, however, require careful two-dimensional modelling in order to obtain reliable electrical cross sections of the Earth's crust. Electromagnetic methods, magnetometer array studies and magnetotelluric profiling have proven to be a useful tool in studies of the lithospheric geophysics of the shield. The crust of the Finnish and Soviet Baltic Shield can be divided into geoelectrically different blocks. In Central Finland the blocks are surrounded by narrow, highly conductive band-like formations. They are located in the middle crust, and their upper surface often occurs in the 6-10 km region. The resistivities are typically 5-10 OMEGA-m. The greatest anomalous crustal regions have a conductance of more than 20,000 S. The lower crust is very conductive in most parts of the shield. Two-dimensional models of long magnetotelluric profiles indicate the great variety and complexity of the geoelectric structure of the crust.

The review describes in broad terms the development of regional EM studies during the last five-six years. Large simultaneous magnetometer arrays, broadband and dense profiling with five component instruments, the use of remote reference techniques and in-field data processing have increased both the number and the quality of EM surveys. The increase has been strong all over the world.

Three groups of electromagnetic data have been considered in order to construct a preliminary geoelectrical depth model of the old basement of the Baltic shield: (i) audiomagnetotelluric data in the period range 13700−18 s; (ii) magnetotelluric data in the period range 25–1000 s; and (iii) global magnetovariation-sounding data in the period range >6 h.

An automatic interpretation program package, using least-squares fit, is described. Starting with models provided either automatically by the program or by the user, the parameters of up to 20 plates or prisms can be found in the non-linear optimization option. Alternatively, the susceptibilities or densities of up to 1000 plates or prisms can be computed. Automatic fit is superior to the methods using characteristic points, since all measured values are utilized and the basic models, plates and prisms can be combined to form highly complicated structures. The program package has been used mainly for structural interpretation on the regional scale. The examples show two-dimensional interpretations of several parallel profiles and of the same profile measured by different geophysical methods, as well as three-dimensional models. Automatic interpretation tends to yield bodies which are broad, are located near the earth's surface and have a low value of the material constant. This has led to the use of the program package as a tool for prognostic ore interpretation. The compact, more highly magnetized units yielded by the automatic process are replaced by thinner bodies having the susceptibility of iron ores. A re-interpretation tells us whether this new, prognostic model is able to explain the measured anomalies or not.