Improved shift-invariant sparse coding for noise attenuation of magnetotelluric data

Li, Guang; Liu, Xiaoqiong; Tang, Jingtian; Deng, Juzhi; Hu, Shuanggui; Zhou, Cong; Chen, Chaojian; Tang, Wenwu

doi:10.1186/s40623-020-01173-7

Cited by 30 publications

(10 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Thus, strong interference of MT data will cause the excessive distortion of the apparent resistivity-phase curve and excessive concentration of phase angle in polarization direction. The high quality data obtained after de-noising will provide technical support for the following inversion interpretation (Qi et al 2020;Li et al 2020a).…”

Section: Introductionmentioning

confidence: 90%

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

Zhang

Li²,

et al. 2021

Preprint

View full text Add to dashboard Cite

Magnetotelluric (MT) data processing can increase the reliability of measured data. Traditional MT de-noising methods are usually filtered in entire MT time-series sequence, which result in losing of useful MT signals and the decrease of imaging accuracy of electromagnetic inversion. However, targeted MT noise separation can retain the part of data not affected by strong noise, and enhance the quality of MT data. Thus, we proposed a novel method for MT noise separation, which using refined composite multiscale dispersion entropy (RCMDE) and orthogonal matching pursuit (OMP). Firstly, the RCMDE characteristic parameters are extracted from each segment of the MT time-series. Then, the characteristic parameters are input to the fuzzy c-mean (FCM) clustering for automatic identification of MT signal and noise. Next, OMP method is utilized to remove the identified noise segments independently. Finally, the reconstructed signal consists of the denoised data segments and the identified useful signal segments. We conducted the simulation experiments and algorithm evaluation on the EMTF data, simulated data and measured sites. The results indicate that the RCMDE can improve the stability of multiscale dispersion entropy (MDE) and multiscale entropy (MSE) by analyzing the characteristics of the signal samples library, effectively dividing MT signals and noise. Compared with the existing techniques of the entire time domain de-noising and signal-noise identification, the proposed method used RCMDE and OMP as characteristic parameter and noise separation, simplified the multi-features fusion, and improved the accuracy of signal-noise identification. Moreover, the de-noising efficiency has accelerated, and the MT data quality of low-frequency band has improved greatly.

show abstract

Section: Introductionmentioning

confidence: 90%

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

Zhang

Li²,

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Existing time series processing techniques, such as the least-square method (Sims et al, 1971), remote reference method (Goubau et al, 1978;Gamble et al, 1979b;Gamble et al, 1979a), robust estimation method (Egbert and Booker, 1986;Chave et al, 1987;Larsen et al, 1996;Smirnov, 2003;Chave and Thomson, 2004), maximum likelihood estimation method (Chave, 2014;Chave, 2017), and others methods based on wavelet transform, 10.3389/feart.2023.1086749 Hilbert-Huang transform, variational mode decomposition, and interstation transfer function (Kappler, 2012;Cai, 2014;Campanya et al, 2014;Cai and Chen, 2015;Carbonari et al, 2017;Wang et al, 2017), have different processing performance for different kinds of noise. For example, the least-square method performs poorly in the presence of outliers (Egbert and Booker, 1986), and the remote reference method does not work when noise is correlated at the local and remote sites (Shalivahan and Bhattacharya, 2002;Pomposiello et al, 2009), and the robust estimation method usually fails to work when MT data are contaminated by persistent or coherent noises (Escalas et al, 2013;Carbonari et al, 2017;Li et al, 2020a;Li et al, 2020c;Zhou et al, 2022;Zhang et al, 2022). New time series processing methods for various types of noise are the focus of current research, but there is a lack of criteria to evaluate the effectiveness of various MT time series processing methods in existing studies.…”

Section: Introductionmentioning

confidence: 99%

Synthesizing magnetotelluric time series based on forward modeling

Wang

Chen²,

Zhang³

2023

Front. Earth Sci.

View full text Add to dashboard Cite

The validity of magnetotelluric time-series processing methods has been lacking reasonable testing criteria. Since the time series synthesized by existing techniques are not fully derived from a given model, they are not reliable. In this paper, we present a novel approach to synthesize magnetotelluric time series based on forward modeling and the correspondence between frequency and time domain electromagnetic fields. In this approach, we obtain the electromagnetic response of two orthogonal polarization sources for a given model by magnetotelluric forward modeling, and simulate the randomness of the polarization of the natural field source by a linear combination of the two polarization sources. Based on the correspondence between the frequency and time domain electromagnetic fields, the electromagnetic fields obtained by forward modeling in the frequency domain are transformed into the time domain, and finally the time series are synthesized. The test results on 1D and 3D models validate the effectiveness of the proposed method and the correctness of the procedure. After adding noise to the synthesized time series, we can test the performance of each method by comparing the results of the time series processing methods with the response of the given model. Therefore, the method presented in this paper can be used to construct standard magnetotelluric time series, which can be used as a carrier to construct synthetic data satisfying various noise distributions, and for the study of related methods. This method can also be used to synthesize time series of other frequency-domain electromagnetic methods.

show abstract

“…Therefore, using modern data processing technology to remove the strong interference existing in measured MT signals has become an important research topic in the field of electromagnetic exploration, and the improvement of MT data quality in the strong interference area will provide strong technical support for the subsequent inversion interpretation (Ren et al ., 2013; Qi et al ., 2020). A wide variety of methods have been proposed for solving this problem, such as short‐time Fourier transform (Vozoff, 1972; Kao and Rankin, 1977; Griffin and Lim, 1984), remote reference (RR) method (Goubau et al ., 1978; Gamble et al ., 1979; Clarke et al ., 1983; Kappler, 2012), robust estimation (Egbert and Booker, 1986; Larsen, 1989; Chave and Thomson, 1989, 2004; Larsen et al ., 1996; Egbert, 1997), wavelet transform (Trad and Travassos, 2000; He et al ., 2009; Carbonari et al ., 2017), Hilbert–Huang transform (HHT) and empirical mode decomposition (EMD; Chen et al ., 2012; Cai, 2014; Chen and Fomel, 2018; Liu et al ., 2019), mathematical morphological filtering (MMF; Tang et al ., 2012b), inter‐station transfer functions (Wang et al ., 2017), Self‐organizing Map (SOM) neural networks (Carbonari et al ., 2018), multifractal spectrum and matching pursuit (MP; Li et al ., 2019), Mahalanobis distance and magnetic field constraints (Platz and Weckmann, 2019), shift‐invariant sparse coding (Li et al ., 2020) etc. These methods have certain advantages and promote the development of MT signal–noise separation research to a certain extent, but there are still some shortcomings.…”

Section: Introductionmentioning

confidence: 99%

Denoising of magnetotelluric data using K‐SVD dictionary training

Peng

Tang

et al. 2021

Geophysical Prospecting

Self Cite

View full text Add to dashboard Cite

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.

show abstract

Improved shift-invariant sparse coding for noise attenuation of magnetotelluric data

Cited by 30 publications

References 34 publications

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

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

Synthesizing magnetotelluric time series based on forward modeling

Denoising of magnetotelluric data using K‐SVD dictionary training

Contact Info

Product

Resources

About