A modified empirical mode decomposition method for multiperiod time-series detrending and the application in full-waveform induced polarization data. Issue 2 (4th February 2019)

• Record Type: Journal Article
• Authors: Liu, Weiqiang; Lü, Qingtian; Chen, Rujun; Lin, Pinrong; Chen, Chaojian; Yang, Liangyong; Cai, Hongzhu
• Is Part Of: Geophysical journal international. Volume 217:Issue 2(2019)
• Page Start: 1058
• Page End: 1079
• Publication Date: 2019-02-04
• Subjects: Electrical resistivity tomography (ERT) -- Fourier analysis -- Time-series analysis; Geophysics -- Periodicals; 550
• DOI: 10.1093/gji/ggz067
• Languages: English
• ISSNs: 0956-540X

Abstract

SUMMARY: Trend drift is an annoying background interference in induced polarization (IP) exploration, which has great influence on the final calculation of apparent complex resistivity spectrum at low frequency (<0.1 Hz). This paper proposed a modified empirical mode decomposition (EMD) detrending technique for multiperiod IP data. The method uses local extreme values of the rising edges and the falling edges to form multiple envelopes and then to fit and eliminate the trend term. Through comparing with the traditional EMD methods using IP data with simulated trend drift, we find that the modified method can be used to obtain a more accurate fitting trend and the computational cost is only a fraction of that of the conventional one. Additionally, this detrending is little affected by other strong noise. We also used IP data with and without trend interferences to analyse this method, respectively. The results show that, for data without trend drift, the signals remain almost unchanged; however, for data with strong trend drift, the data quality is greatly improved and the calculation error is reduced. This technique is also applied to a large-scale multiperiod full-waveform IP data acquired in Zhegu Zn-Sb-Ag polymetallic deposit in southern Tibet, China. The apparent complex resistivity and phase of a survey line, a planar contour map and a pseudo-section with and without using the modified EMD were compared, respectively. Overall, before EMD detrending, the apparent phase results are rough and full of outliers. After detrending, the profiles are smooth and reasonable, and the outliers disappear. Both the results demonstrated that our proposed method can be adopted to effectively suppress trend drift interference without additional deviation in distributed full-waveform IP exploration.