Instantaneous Phase Discontinuity: An Innovative Method for Time-Frequency Overlap Signal Separation in Denoising In Situ Magnetic Field Data

Magnetic fields are essential for understanding wave-particle interactions and energy transfer in space physics. However, in situ magnetic field measurements are often disturbed by stray magnetic fields from the host satellite platform. The reaction wheels generate complex magnetic harmonic interferences, often overlapping with natural signals in the time-frequency domain. Extracting the natural components from time-frequency overlapping signals remains a challenging issue. Most existing methods pay little attention to maintaining the integrity of natural signals. In this paper, Instantaneous Phase Discontinuity (IPD) is proposed to separate the overlapping natural magnetic field data and local noise in the time-frequency domain. First, IPD extracts the objective interference from original signals, identifies and marks overlapping regions via the discontinuities in the instantaneous phase function, which act as indicators of overlapping regions. Second, IPD reconstructs interference and valuable components within the overlapping regions through iterative frequency band contraction and envelope correction. It is demonstrated by numerical experiments that IPD achieves interference separation while minimizing signal loss compared to traditional methods. For example, when separating time-frequency overlapping signals with −3 dB white noise, IPD exhibits an exceptionally low signal loss rate, significantly lower than that of the band-stop filter, singular spectrum analysis, and variational mode decomposition. Applied to the China Seismo-Electromagnetic Satellite mission, it is shown that IPD allows non-destructive separation of natural magnetic field data and local reaction wheel interferences. This method first introduces a new approach to address the challenge of separating time-frequency overlap signals in denoising in situ magnetic field data.