Noise removal for airborne time domain electromagnetic data based on minimum noise fraction

Residual noise remains in airborne time domain electromagnetic profiles after preprocessing the data, and this noise affects the exploration of targets. An approach to reduce this noise based on the minimum noise fraction has been proposed. The minimum noise fraction uses a rotation matrix to transform noise-contaminated electromagnetic data into the minimum noise fraction components ordered by signal-to-noise ratio. The rotation matrix is formed based on the use of noise covariance estimation and the data covariance. Noise can be effectively removed when we reconstruct the electromagnetic data using the low-order minimum noise fraction components whose signal-to-noise ratios are sufficiently high. In this work, we discuss the de-noising process based on the minimum noise fraction for two earth models and field data from Ontario Airborne Geophysical Surveys over the Nestor Falls area, Canada. Example applications to synthetic and field data are used to demonstrate the excellent performance of the proposed method.