Denoising CSEM Data Using Least-Squares Method Based on Mixed Basis of Fourier Series and Legendre Polynomials
Yang Yang,Changyu Zhou,Heng Zhang,Yonghui Peng,Huaifeng Sun
DOI: https://doi.org/10.1109/tgrs.2023.3326345
IF: 8.2
2023-11-17
IEEE Transactions on Geoscience and Remote Sensing
Abstract:Controlled-source electromagnetic (CSEM) surveys are widely used, but due to the influence of instrumental or environmental factors, the data obtained from CSEM surveys are often disturbed by noise. To address this problem, a denoising method based on least-squares inversion has been proposed to effectively deal with the nonperiodic noise in CSEM data. This method selects special reconstruction areas in the time domain and establishes an overdetermined equation to obtain the noise. However, the condition in selecting reconstruction areas is rigorous, according to which only an area merely consisting of signal and white Gaussian noise can be chosen as an ideal reconstruction area, limiting the application of this method. In this article, an improved method was proposed by introducing the Legendre polynomials into the overdetermined equation. Using this improved method, areas containing other kinds of noise can also be chosen as reconstruction areas, extending the scope of application. To avoid the possible calculation error that comes from an ill-conditioned matrix, a regularization factor is introduced into the overdetermined equation as an option. Meanwhile, to quantitatively evaluate the effect of this proposed method, the envelope evaluation method is involved and a "collapse algorithm" is proposed to evaluate the denoising effect through residual noise. Through simulation and real case study, the validity and effectiveness of the proposed denoising and evaluating process are proved.
imaging science & photographic technology,remote sensing,engineering, electrical & electronic,geochemistry & geophysics