PARAMETRIC MODELING OF TRANSIENT ELECTROMAGNETIC SYSTEMS

The modeling of a transient electromagnetic system as a single‐input, single‐output, linear, causal process is considered. Two canonical forms are discussed; one for impulse function input and the other for more general input. Expressions are derived for errors in the estimation of the model parameters caused by errors in the process data. It is shown that Prony's method is a special case of the impulse function model and that serious errors occur in attempts to apply the method in a noisy environment. It is also shown that least squares estimation of the model parameters produces biased results and that attempts to overcome this problem have led to nonlinear equation sets. Some recommendations are made concerning noise identification and filtering in order to improve results. Copyright 1979 by the American Geophysical Union.