Application of adaptive multiscale decomposition in suppressing interference in geomagnetic observation data

ABSTRACT Geomagnetic data contain valuable information from geophysical sources, such as the earth’s magnetosphere, ionosphere, and seismogenic processes. However, these data are increasingly contaminated by anthropogenic noise, particularly from subway operations, posing significant challenges. To address this, a novel algorithm combining adaptive multiresolution multivariate variational modal decomposition (AMMVMD) and adaptive wavelet transform (AWT), referred to as AMMVMD-AWT, is developed to effectively suppress noise caused by subway operations. The algorithm uses AMMVMD to iteratively decompose low-frequency components, separating useful information from noise. High-frequency components from each decomposition are selected and weighted. Subsequently, the AWT processes the reconstructed signal through multiscale decomposition with adaptive selection of the optimal wavelet basis. Simulation results demonstrate the method’s feasibility, with the signal-to-noise ratio increasing from 4.95 to 20.75 dB and the root-mean-square error significantly reduced. Field data experiments further confirm its effectiveness in suppressing subway interference, with the first-order difference of the processed signal remaining consistently within ±0.1 nT. The AMMVMD-AWT algorithm enhances the reliability and application value of geomagnetic data by improving noise suppression and preserving critical geophysical information. This approach provides a robust foundation for studying geomagnetic variations and related phenomena.