A new look at an old problem: magnetotelluric modelling of 1‐D structures

Abstract

Some features of 1‐D magnetotelluric modelling (1‐D/MT) are reviewed from a practical point of view. A well‐known feature of 1‐D modelling is that deeply buried conductors can always be perfectly fitted with ideally thin sheets of finite conductance. Less well known is that a deeply buried resistor sandwiched by two conductive layers can best be modelled with an insulating layer of minimum thickness. These two observations call to mind Parker's theory according to which the best fit to a 1‐D structure is a D+ distribution (i.e. a perfectly insulating half‐space containing a finite number of ideally thin sheets, each with a finite conductance). However, a structure with n layers of alternatingly low and high resistivities cannot be fitted satisfactorily by replacing the low‐resistivity layers with thin sheets and the high‐resistivity ones by insulators, as this would reduce the number of adjustable parameters from an initial 2n ‐ 1 to only n or n ‐ 1 (the positions of thin sheets with their associated conductances). To achieve its good fit Parker's D+ distribution also requires a number of adjustable parameters close to 2n ‐ 1. In other words, it requires about twice as many layers of alternatingly infinite or vanishing resisitivity. But such models do not, in general, bear any relationship to the structures under investigation. This underlines the case in favour of making use of all possible reliable prior information to constrain the family of permissible models. This should steer the 1‐D model search away from thin‐sheet models toward more realistic structures. Copyright © 1991, Wiley Blackwell. All rights reserved