Spike noise removal from aeromagnetic data using hybrid nonlinear filtering and fuzzy C-means clustering

Abstract

Aeromagnetic data are often affected by spike noise caused by instrumental errors and cultural interference, which can degrade geological interpretation. This study presents an automated denoising framework combining nonlinear (NL) filtering and fuzzy C-means (FCM) clustering for spike detection and removal. The method incorporates adaptive masking and interpolation based on a minimum curvature approach, with parameter optimization guided by an unsupervised cost function that balances reconstruction error, mask extent, and signal smoothness. The algorithm was validated using synthetic magnetic profiles generated from forward modeling and applied to real aeromagnetic data from the Abu Mehrek area, Western Desert, Egypt. Results show that the proposed approach effectively suppresses high-frequency cultural spikes while preserving geological anomalies. An optional interactive refinement tool is included to allow user adjustment in complex cases. The proposed method provides efficient alternative to manual editing for aeromagnetic data preprocessing.