Research On Long-period Magnetotelluric Data Processing

Long-period magnetotelluric(LMT) sounding is an important method to study the crust-mantle electrical resistivity structure. As domestic LMT sounding method for a wide range of soundings started in 2008, at present, there are relatively few researches on LMT. To sum up, two aspects of the research on long-period magnetotelluric remain to be studied further: The first is the LMT data processing problem. Compared with the routine broadband magnetotelluric sounding, the signal characteristics, equipment, layout of observation stations, data acquisition and processing of long-period magnetotelluric have the particularity, with a lower frequency, a weaker signal, a longer acquisition time, and more interference. Studying the LMT data processing problem can improve the processing precision of long period magnetotelluric. And the second is the LMT forward modeling and inversion. For example, the traditional one-dimension horizontal layered medium model, two-dimensional rectanqular boundary model and three-dimensional rectanqular solid boundary model are not appropriate for the long-range profile and large-depth long-period magnetotelluric. For the improvement of MT sounding effect, top priority should be given to the research on data processing.This paper mainly focuses on the first one, that is, to research the long-period magnetotelluric data processing, the measures of which mainly include three aspects: The first is to study the signal characteristics and noise separation techniques; the second is to contrast and summarize current impedance tensor estimation methods, and the third is to propose a remote reference magnetotelluric method based on magnetic filed correlation. On the basis of above methods, a set of LMT data processing process module with proprietary intellectual right is established. With above-mentioned modules adopted to process the real LMT data, the crust-mantle electrical resistivity structure of Panxi has been obtained, which can provide scientific evidence for the research on the occurrence mechanism and dynamics problems of strong earthquakes in this area.This paper analyzes the signal characteristics of long-period magnetotelluric, and noise separation techniques, and it can be found that the electromagnetic polarization direction of natural LMT signal is arbitrary with a high degree of coherence of nearly 1 as the frequency changes, and the time-frequency characteristics indicate that energy spectra at the same frequency have uniform distribution and slow changes. Moreover, it can be discovered that the electromagnetic field intensity of the low frequency(~10000s) is nearly two orders higher than that of “dead bond”(~4s-~20s), and the electric field intensity of low frequencies is apparently higher than the magnetic field intensity; Furthermore, the LMT field source is mainly influenced by solar diurnal variation,magnetic storm,magnetic substorm and geomagnetic pulsation; the noise source of the LMT can be divided into field-source noise, geological noise and human noise, and in terms of the human noise, seven wavelet basis functions are used to compare the suppression effects of simulation square wave and step noise, through which it can be found that db wavelet has a stronger suppression effect on noise, and the median filter has abetter suppression effect on impulse noise than that of db wavelet.A set of LMT data processing process module with proprietary intellectual right is established. The screening processes of comprehensive-information vector coherence and regular degree of coherence for the data are systematically analyzed, and it is believed that the two methods have different degrees of improvement for the processing effect of sounding data, and during the real data processing, these two methods are suggested to achieve a combination. The least square method, M estimation Robust algorithm, and Robust algorithm based on median repetition have been achieved, and with the three estimation methods used for measured LMT data processing and compared with each other, it can be found that the least square method has poor data processing results, and the estimation Robust algorithm based on median repetition shows good processing results, while M estimation Robust algorithm falls between the two; When the written programming module is applied to the standard time series and the actual long-period data processing, the comparison between its processing result and that of the open source software EMTF and the commercial software PRC_MTMV, the calculation of the written programming module turns out to be correct.A remote reference magnetotelluric method based on magnetic filed correlation is proposed and long-period magnetotelluric sounding curve is optimized, thus improving the effect of long-period magnetotelluric sounding. Based on the analysis on simulation data, the influences of the remote reference magnetotelluric on sounding curves are simulated under different reference traces, signal-noise ratios or only under reference traces or base stations. As the regular remote reference magnetotelluric method focuses on the non-correlation between noise and signal or between the base station noise and reference station noise, but attaches little importance to the correlation among magnetic signals in base station and reference station, the remote reference magnetotelluric method based on magnetic filed correlation has been proposed. With simulation triangular wave, square wave and sine wave noise added into the original data segment of the base station, and according to the analysis on the middle data of impedance tensor estimation process, it is believed that the remote reference magnetotelluric method based on magnetic field correlation can get better sounding curve than the regular remote reference magnetotelluric method, and the processing result of measured data also indicates that the remote reference magnetotelluric method based on magnetic field correlation is superior to the regular remote reference magnetotelluric method.The characteristics of crust-mantle electrical resistivity structure of Panxi As above algorithm modules are applied to the real data processing of the Yanyuan-Yongshan LMT profile, the characteristics of electrical resistivity structure have been got through the relative inversion software. With reference to other geophysics and surface geology data, it is thought that thickened basic rocks and ultra-basic rocks may heap up in High Resistivity Column of Panxi Lithosphere, and then it can be inferred that current High Resistivity Column of Panxi Lithosphere may be the eruption channel of Permian Emei Mountain basalts, as well as the deposition channel of thickened basic rocks and ultra-basic rocks. And the relation between electrical structure characteristics and seismic sources of strong earthquakes has been discussed.