Source-field bias in geomagnetic transfer functions: A case history.

Abstract

A magnetometer array study in 1971 discovered the Southern Cape Conductive Belt beneath the Karroo Basin and Cape Folded Mountains of South Africa. A second array in 1977 has mapped this Conductive Belt in some detail, by means of anomalies in maps of Fourier transform amplitudes and phases. Transfer functions from horizontal to vertical components were calculated from data recorded by each array, and used to map induction vectors. The in-phase induction vectors from the 1971 array are in good agreement with the current locations shown by the Fourier transform anomaly maps. The 1977 array data give induction vectors which agree with neither those of 1971, nor the anomaly maps. After other possible sources of systematic error had been eliminated, it has been shown that the problem arises from source-field bias introduced by cross-correlation between the horizontal components of the normal field. Two tests are suggested, in the selection of events for calculation of valid transfer functions. Errors from source-field bias could be important in work with small numbers of magnetometers, in which anomalies cannot be mapped from simultaneous observations of events by a large array.