Development of new processing schemes for the removal of harmonic noise from SNMR acquisition signals in complex, multi-frequency noise environments.

In SNMR surveys, the electromagnetic (EM) perturbations originating from powerlines infrastructures is often the main EM noise source that contaminates the SNMR acquisition signal. The removal of these harmonic signals is therefore critical to extract the data and carry out the study. To date, some techniques exist that give satisfying results in many cases, provided that the assumptions of a harmonic noise based on a single and constant fundamental frequency are valid. However, when these conditions are not met and the characteristics of the harmonic noise become more complex, these methods will fail to remove it correctly. In this paper, we show how such complex conditions will affect the frequency spectra of the SNMR acquisition signal. Then, we address the issue of a harmonic noise composition based on two different fundamental frequencies, by solving the non-linear optimization problem simultaneously for the two frequency values, using a 2D grid-search. Then we investigate the case where the fundamental frequency of the harmonic signal is not constant but variates within the signal duration. We expose a new approach based on the Nyman and Gaiser estimator, which, associated with a signal bootstrapping strategy, efficiently handles this situation.