Characteristics And Signal Detection Method Of Low-frequency Vibroseis

Low-frequency vibroseis is the artificial source with the low-end frequency down to 1.5Hz.Due to its highly stability,long distance and environmental friendly,Low-frequency vibroseis could be used for seismic exploration in regional scale as tens to hundreds kilometers,which provides a solution for exploring some of the basic questions in geoscience,and became a new focus in the research of seismic exploration.This thesis focuses on test the latest innovated low-frequency vibroseis of BGP in the current world,KZ-28 LF and EV-56 which has a low-end frequency down to3Hz(minimum 1.5Hz).We studied the signal characteristics and stability based on four field experimental data,and developed a new method to eliminate bad excitation catalog and the accurate source signal reference zero time using cross correlation.We also present a method,which is based on similarity distance matrix and hierarchical clustering algorithm,to get a perfect catalog without bad shots.We compared three methods used to compress the vibroseis signals.at last,we show the possibility to explore the shallow velocity structure by extracting the surface wave dispersion curve.The new understandings are listed as following:(1)The low-frequency vibroseis of BGP can steadily excite very low frequency signal down to 1.5 to 12 Hz.Among them,the 1.5~9Hz`s energy is weak and couples a higher energy harmonic components due to the current limitation of mechanism and electricity.the result shows EV-56 is better than KZ-28 LF on suppression the harmonic components.Pseudo random encoding signals are encoded unevenly in1.5~12Hz frequency band,with two energy peaks centered near 4Hz and 11 Hz.(2)Cross correlation method can effectively get a very high precision zero reference shot time,and can perfectly eliminate the bad shots.(3)Cross correlation method and hierarchical clustering algorithm can effectively evaluate the source signal stability and repeatability,and eliminate the bad shots at the same time.The analysis showed that the single frequency scanning method is worst in waveform repeatability,so cannot be used to detect the velocitychange of shallow medium,which demand a very high repeating source signal.Pseudo random encoding scanning method can generate a highly repeatable coding signal for a certain time,and failed perhaps due to some drawbacks of mechanism and electricity.The linear frequency sweep scanning method`s results are the best in signal stability,and can be used for monitoring the velocity change of shallow medium.(4)Cross-coherence and deconvolution methods can be used to compress all kinds of sweeping source signal from signal scanning to pseudo random encoding scanning method.The result shows the cross-coherence method is best,which can get a very high signal to noise ratio.Only corss-coherence method and deconvolution method can be used to compress the source signal when the low-frequency vibroseis is managed with a single frequency sweeping scanning method.cross-correlation method and moving window cross-correlation method always get a larger amplitude of first arrival body waves than surface waves.And the cross-coherence method apt to achieve a higher body waves than others when the source function is in line sweeping,but the amplitude of surface wave is quite the same with deconvolution method.When using pseudo random encoding scanning method,the signal to noise ratio of surface wave recovered by cross-correlation and moving window cross-correlation method is very low,and the amplitude of body wave obtained by cross-coherence method is larger,and the amplitude of surface wave is equal to deconvolution.(5)The largest energy of YIX seismic station is among 2.0 to 5.0 frequency band,which may come from the low-frequency vibroseis located 40 km away.The very weak source signal can be enhanced by stacking based on their highly similarity.The calculation result shows,direct linear stacking method can detect the source signal within 40 km.and the weighted stacking method and rms sorted weighted stacking method utilizes coherence of signal in multichannel waves to suppress noise and enhance the signal to noise ratio,which can retrieve the source signal within100 km.(6)The shallow velocity model of earth structure can be inversed using the signal excited by a low-frequency vibroseis by retrieve the dispersion curves.The results obtained in this paper have some important implications.For example,it can provide some suggestions in field experiment design,and reasonable processing methods in laboratory low-frequency signal processing.