Method For Synthesizing The Magnetotelluric Time Series With Forward Modeling And Its Application

Magnetotellurics(MT)is a crucial geophysical exploration technique widely employed in fields such as petroleum and mineral exploration,seismic hazard detection,and deep structural research.Time series processing technology is an essential step in MT applications.However,with the progression of social civilization,electromagnetic noise generated by human activities has become increasingly severe.Conventional MT time series processing techniques,such as robust estimation techniques,are inadequate when processing field data in areas with strong interference.To address this issue,researchers have developed numerous novel time series data processing methods.Nevertheless,the practical effectiveness of these methods remains uncertain due to the absence of standard MT time series for evaluating their performance.The insufficient understanding of numerous issues encountered during the processing of field data presents a bottleneck impeding the advancement of data processing technology.These issues include: Why do robust estimation techniques developed for noisy data fail in areas with intense interference? Why is the reference effect of the electric field inferior to that of the magnetic field in the classical remote reference method? Why is the high-frequency electric field remote reference effective? What are the characteristics of strong man-made interference signals? What influence do they exert on MT responses? How can these characteristics be employed to suppress noise and enhance data processing quality? During the era of two-dimensional inversion,it was commonly assumed that phase quadrature resulted from noise.However,three-dimensional inversion can now also accommodate phase quadrature.Is phase quadrature induced by noise or structure?Since the true response of field data is unknown,as are the signal components and the noise components contained therein,it is challenging to answer these questions through field data.The synthesis of standard MT time series can serve not only as a research sample for the above questions but also as a test standard for novel methods.It holds significant theoretical and practical value for the advancement of MT data processing technology.This paper presents a forward-based MT time series synthesis technology that leverages the relationship between time-domain electromagnetic fields and frequencydomain electromagnetic fields to conduct research on the aforementioned issues.The specific contributions and discoveries of this paper are as follows:(1)The relationship between the time-domain signal and the frequency-domain signal of a single-frequency electromagnetic field is theoretically derived.Utilizing this relationship,the field values computed by frequency-domain MT forward modeling are transformed into time series,augmented by the simulation of natural field sources,forming a set of MT time series synthesis technology.This technology can synthesize MT time series at any position within any model.The characteristics of the time series align with both the natural field and the given forward model and can serve as a standard for future research.To facilitate research on time series processing technology,MT data synthesis processing software has been developed.In the analysis of synthesized time series,pulse noise,step noise,triangular wave noise,and sine wave noise were individually introduced and analyzed.It was discovered that the MT responses varied among different types of noise.(2)To investigate the differences between electric field remote reference and magnetic field remote reference,as well as the issue of electric field remote reference typically being effective at high frequencies,the inversion results of some field data were employed as a model to synthesize MT standard time series above various electrical structures and conduct research on the remote reference method.It was discovered that the magnetic field remote reference is less influenced by underground electrical structures,while the electric field remote reference is more affected.The electric field above complex structures is altered by the induction field generated by underground anomalies,causing its correlation with the natural field source background to decrease and resulting in a deterioration of the reference effect.Locations with simple underground structures are suitable for remote reference stations.The two-dimensional deviation of a large quantity of field data was statistically analyzed and it was determined that the high-frequency two-dimensional skew of field data is generally small.The detection range of high-frequency electromagnetic fields is limited and the underground structure is relatively simple,resulting in a good electric field reference effect.Based on this finding,an observation processing scheme utilizing inter-group synchronous observation to enhance data quality by referencing electric fields with one another in audio magnetotellurics was proposed.The practical application effect is excellent and warrants promotion.(3)Time series synthesis technology was employed to synthesize the time series of dipole source and line source electromagnetic methods,and research was conducted on strong stable interference sources.It was discovered that dipole source noise can influence the near-field shape of MT responses.Strong stable interference primarily impacts the mid-frequency band where the natural field is weak,and the response within this frequency band approximates the scalar response of the noise source.Noisy data and natural field data can be identified and eliminated using frequency domain parameters such as power spectrum,transfer function,and electromagnetic field polarization direction.In the simulation of line source noise,it was determined that asymmetric transmission line source noise may induce phase quadrature phenomena.Phase quadrature caused by noise and structure can be distinguished using parameters such as amplitude spectrum,coherence,and deviation.(4)Finally,two MT time series processing techniques were investigated using synthesized noisy MT time series as samples.The first is a robust estimation weighted multi-parameter data screening method that integrates robust estimation methods with data screening methods.Through repeated weighting and screening,in conjunction with the constraints of Rhoplus inversion,the quality of field data processing results can be effectively enhanced.The second is a self-power spectrum inverse weighting method that employs man-made noise signals to decrease the weight of noisy time period data.This method exerts a certain suppression effect on strong interference noise and possesses promising development potential.