Denoising of magnetotelluric data using K‐SVD dictionary training

Abstract

Magnetotelluric is one of the mainstream exploration geophysical methods, which plays a vital role in studying deep geological structures and finding deep hidden blind ore bodies. The seriousness of human electromagnetic noise causes a large number of abnormal waveforms in the time series of measured magnetotelluric data, and the data can no longer objectively reflect the underground electrical distribution. In this work, we propose a magnetotelluric time series data processing method based on K singular value decomposition dictionary training. First, a training matrix and a to-beprocessed matrix are built with the pending magnetotelluric signals. Then, let the K singular value decomposition dictionary training process the training matrix to obtain an over-complete dictionary reflecting the characteristics of the pending signal. Lastly, orthogonal matching pursuit is combined with an over-complete dictionary updated in real time to sparsely represent the to-be-processed matrix and remove human electromagnetic interference in the signal. Experimental results show that the method can update the over-complete dictionary in real-time according to the pending magnetotelluric signals, realize the self-learning signal-noise separation of magnetotelluric signals, and effectively retain low-frequency information. Compared with method of directions dictionary learning, remote reference method, and orthogonal matching pursuit method, the reconstructed data of the proposed method can more accurately reflect the underground electrical structure information.

Separation of magnetotelluric signals based on refined composite multiscale dispersion entropy and orthogonal matching pursuit

“…The MP algorithm at each iteration ensures only that the matching residual data are orthogonal to an atom, which is prone to local optimization and results in a low matching accuracy and a large amount of computation (Huang and Makur 2011;Jin et al 2014). The OMP algorithm is based on the MP algorithm, ensuring full backward orthogonality between the matching residual and the selected waveforms at each iteration and ensuring the optimal approximation regarding all the selected subset of the dictionary after any finite number of iterations (Wang et al 2013;Li et al 2021).…”

Long Short-Term Memory Recurrent Network Architectures for Electromagnetic Field Reconstruction Based on Underground Observations

“…The efficacy of supervised learning-based methods depends on available training datasets that encompass the principal signal and noise morphological features. Additionally, unsupervised data-driven deep learning techniques based on sparse coding algorithms are employed in the noise identification, separation, and suppression of EM data, such as K-SVD dictionary learning [48][49][50][51][52][53], improved shift-invariant sparse coding [54], and adaptive sparse representation [55], among others. However, appropriately defining specified coefficients is essential for these methods, often requiring multiple experiments to achieve optimal results.…”

“…However, constructing sufficient sample libraries that encompass various complex noise features of practical data can be challenging. Although data-driven algorithms such as dictionary learning based on sparse coding [48][49][50]54,55] can be used to obtain atoms that match the target signal directly from the observed data, the applicability of such methods when faced with persistent and strong noise interference still requires further investigation.…”

Removal of strong noise in magnetotelluric data using grey wolf optimized wavelet threshold

“…Wang et al (2017) used inter-station transfer functions and time series from a neighbouring site (STIN) to synthesize natural electromagnetic time series to process MT noise data. Li et al (2021) introduced K-singular value decomposition (SVD) dictionary learning into the field of MT data processing, which effectively improved the quality of MT data and solved the insufficient problems of adaptability and low efficiency in the traditional sparse methods. Zhou et al (2021) proposed the use of a denoising method for the detection of noise in MT data based on discrete wavelet transform (DWT) and SVD, with multi-scale dispersion entropy and phase space reconstruction carried out for pretreatment.…”

“…Li et al. (2021) introduced K‐singular value decomposition (SVD) dictionary learning into the field of MT data processing, which effectively improved the quality of MT data and solved the insufficient problems of adaptability and low efficiency in the traditional sparse methods. Zhou et al.…”