An application of stochastic system identification to geo-electromagnetics

摘要

Concepts from modern system identification theory have been applied to the processing of stochastic electromagnetic (EM) signals of geophysical origin in order to separate a desired signal from corrupting background noise for situations where a priori knowledge of the noise statistics is minimal. Parametric identification has been utilized in controlled source electromagnetic methods (CSEM) for the separation of the target signal from the noise (viz., natural fields and geological and topographic noise) by using only information provided by the received signal measured at a base sensor and by the reference inputs measured at the remote or reference sensors. A recursive maximum-likelihood prediction-error algorithm, which permits data processing in either an off-line or an on-line, real-time mode, has been implemented and applied to both synthetic and real CSEM data in order to demonstrate the potential value of the method developed.