Magnetotelluric signal-noise separation method based on SVM–CEEMDWT
Li Jin, Cai Jin, Tang Jing-Tian, Li Guang, Zhang Xian, and Xu Zhi-Min
1.Hunan Provincial Key Laboratory of Intelligent Computing and Language Information Processing, College of Information Science and Engineering, Hunan Normal University, Changsha 410081, China.
2.School of Geosciences and Info-Physics, Key Laboratory of Metallogenic Prediction of Non-Ferrous Metals and Geological Environment Monitor, Ministry of Education, Central South University, Changsha 410083, China.
3.State Key Laboratory Breeding Base of Nuclear Resources and Environment, East China University of Technology, Nanchang 330013, China.
Abstract：
To better retain useful weak low-frequency magnetotelluric (MT) signals with strong interference during MT data processing, we propose a SVM–CEEMDWT based MT data signal-noise separation method, which extracts the weak MT signal affected by strong interference. First, the approximate entropy, fuzzy entropy, sample entropy, and Lempel–Ziv (LZ) complexity are extracted from the magnetotelluric data. Then, four robust parameters are used as the inputs to the support vector machine (SVM) to train the sample library and build a model based on the different complexity of signals. Based on this model, we can only consider time series with strong interference when using the complementary ensemble empirical mode decomposition (CEEMD) and wavelet threshold (WT) for noise suppression. Simulation results suggest that the SVM based on the robust parameters can distinguish the time periods with strong interference well before noise suppression. Compared with the CEEMDWT, the proposed SVM–CEEMDWT method retains more low-frequency low-variability information, and the apparent resistivity curve is smoother and more continuous. Moreover, the results better reflect the deep electrical structure in the field.
2019, Vol. 16 
