https://brgm.hal.science/hal-01080987Penz, SébastienSébastienPenzBRGM - Bureau de Recherches Géologiques et Minières (BRGM)Girard, Jean-FrançoisJean-FrançoisGirardBRGM - Bureau de Recherches Géologiques et Minières (BRGM)Powerline harmonic noise in MRS data cancellation by sinusoidal subtraction An efficient solution to cancel powerline interferencesHAL CCSD2014[SDU.STU.GP] Sciences of the Universe [physics]/Earth Sciences/Geophysics [physics.geo-ph]Penz, Sébastien2014-11-06 15:59:362022-08-03 04:02:582014-11-27 14:06:43enConference posterapplication/pdf1The surface magnetic resonance method (SNMR) is a valuable technique for hydrogeological studies, since it provides information on the porosity, hydraulic conductivity and water content. However the bad signal-to-noise ratio often encountered limits its application range. In particular, in suburban areas, strong powerline harmonics severely degrade the SNMR signals. The powerline fundamental frequency is subject to small variations in time that make high order harmonics filtering difficult using a classical notch filter without distorting the MRS signal. Harmonics cancellation using sinusoidal subtraction is an alternative, but requires a high accuracy of the instantaneous fundamental frequency value to be efficient. The power grid frequency being regulated we expect that the instantaneous frequency could be monitored with a good S/N ratio using a remote frequencemeter, synchronised with the SNMR device, and later used for processing. Based on this accurate instantaneous frequency measurement we developed a processing method based on the subtraction of a powerline harmonics model of constant amplitudes. We performed numerical experiments to quantify the efficiency of this method, and compare it with classical notch or multi-channel filtering. Studying real noise measurements, we encountered cases where harmonics amplitudes do not remains stable over time. We therefore proposed a processing method that include time-variant amplitudes in the powerline harmonics model. Amplitudes are modeled using a b-spline decomposition and estimated through an inversion procedure. A test on synthetic data is presented to assess the theoretical applicability of the method.