STRAIGHTFORWARD INVERSION SCHEME (SIS) FOR ONE-DIMENSIONAL MAGNETOTELLURIC DATA

This paper presents a Straightforward Inversion Scheme (SIS) for inter preting one-dimensional magnetotelluric sounding data. The basic steps of SIS are (i) parameterization of the layered model such that the la yer thickness, expressed in units of its skin depth, is a constant (al pha); (ii) expansion of the reflection function at each interface as a power series in parameter u = exp(-2(1 + j)alpha root (f) over bar);( iii) development of a recurrence relation between the coefficients of the same powers of u in the power series of reflection functions of an y two successive layers; (iv) estimation of the impedance power series coefficients using regressed minimum norm estimator; and (v) evaluati on of layer resistivities and thicknesses using the inverse recurrence relation. The power of SIS is established by inverting four synthetic data sets and two field data sets. The effect of noise is extensively studied on a synthetic data set, deliberately corrupted with increasi ng levels of Gaussian random noise up to 25%. It is found that the sch eme can retrieve broad features of the true model even with noise leve ls as high as 25%. On the basis of findings of different experiments c onducted on SIS, it is concluded that SIS is an efficient, robust algo rithm with high resolving power. Further, being linear, it is non-iter ative and it dispenses with the requirement of having to choose an ini tial guess model.