Identification and removal of man-made transients from geomagnetic array time series: a wavelet transform based approach

We address the problem of identifying and removing man-made transients from geomagnetic time series acquired with an array of magnetometers. We model the transients as scaling functions of unknown scales, amplitudes, and delays. We then use an undecimated discrete wavelet transform to identify the transients in the presence of the natural 1/f geomagnetic background. The identification criteria incorporate the temporal and spatial characteristics of the transients. Using maximum likelihood estimates of the transient parameters, we subtract the detected scaling functions from the original time series. The efficacy of the method is demonstrated with experimental data.