https://hal-nantes-universite.archives-ouvertes.fr/hal-02481429Kremer, ThomasThomasKremerLPG - Laboratoire de Planétologie et Géodynamique [UMR 6112] - UA - Université d'Angers - UN UFR ST - Université de Nantes - UFR des Sciences et des Techniques - UN - Université de Nantes - INSU - CNRS - Institut national des sciences de l'Univers - CNRS - Centre National de la Recherche ScientifiqueARGENCO - Université de Liège (FNRS, Dpt ARGENCO) - FNRSLarsen, JakobJakobLarsenAarhus School of Engineering - Aarhus University [Aarhus]Nguyen, FredericFredericNguyenARGENCO - Université de Liège (FNRS, Dpt ARGENCO) - FNRSProcessing harmonic EM noise with multiple or unstable frequency content in surface NMR surveysHAL CCSD2019HydrogeophysicsInstrumental noiseTime-series analysisHarmonic EM noiseSurface nuclear magnetic resonance[SDU.STU.GP] Sciences of the Universe [physics]/Earth Sciences/Geophysics [physics.geo-ph]KREMER, Thomas2020-02-17 14:37:582022-05-12 09:56:362020-02-20 14:13:46enJournal articleshttps://hal-nantes-universite.archives-ouvertes.fr/hal-02481429/document10.1093/gji/ggz307application/pdf1The harmonic electromagnetic noise produced by anthropic electrical structures is a critical component of the global noise affecting geophysical signals and increasing data uncertainty. It is composed of a series of harmonic signals whose frequencies are multiple integers of the fundamental frequency specific to the electrical noise source. To date, most model-based noise removal strategies assume that the fundamental frequency constraining the harmonic noise is single and constant over the duration of the geophysical record. In this paper, we demonstrate that classical harmonic processing methods lose efficacy when these assumptions are not valid. We present several surface nuclear magnetic resonance field data sets, which testify the increasing probability of recording the harmonic noise with such multiple or unstable frequency content. For each case (multiple frequencies or unstable frequency) we propose new processing strategies, namely, the 2-D grid-search and the segmentation approach, respectively, which efficiently manage to remove the harmonic noise in these difficult conditions. In the process, we also apply a fast frequency estimator called the Nyman, Gaiser and Saucier estimation method, which shows equivalent performance as classical estimators while allowing a reduction of the computing time by a factor of 2.5.