Statistical Correlation Method Based On Frequency Domain For Casing Wave Suppression In VSP Data

Under the background of high exploitation of oil,natural gas and other resources,the development trend of seismic exploration technology is becoming more and more severe.Vertical seismic profile technology is an extension of conventional ground seismic profile technology and a more refined seismic exploration technology.Using VSP technology,seismic exploration workers can obtain a large number of effective geological information near wells,which can be used to carry out VSP imaging,and the purpose of clear interpretation for formation structure can be achieved.However,due to the special data acquisition method of VSP technology,the data are often affected by various interference waves,which leads to incorrect or even impossible VSP imaging.In this paper,the casing wave is mainly studied and suppressed by using filter designing method and statistical correlation method based on frequency domain.Suppressing interference waves with filters is the most common method for signal denoising.Noise-containing seismic data are selected from zero offset VSP seismic data of Xushen 21-1 well for spectrum analysis,then the design indexes of low-pass,high-pass and band-pass FIR filters are pointed out and applied to VSP data casing interference wave suppression by low-pass,high-pass hybrid filtering and band-pass filtering respectively.The results show that the filtering method designed for singlechannel seismic data can suppress interference waves,but the noise frequency characteristics of different seismic record data are different,and the filter design index must change with the change of the noise frequency characteristics and cannot be unified,which is unrealistic for batch processing of a large number of multi-channel seismic data and cannot meet the requirements of data processing efficiency.This paper proposes a noise suppression method based on statistical correlation in frequency domain,whose starting point is still to eliminate the noise frequency and retain the frequency of the effective signal.The difference is that the research subject no longer designs the filter by analyzing the frequency characteristics of the noise,but analyzes the main frequency distribution characteristics of the effective signal to retain the effective signal in frequency domain and further realize the suppression of the noise signal.Using the basic principle that the input signals are the same when the single shot is excited,and the distance between detection points receiving the signals is 20 meters,and the total distance is not more than 160 meters,so the underground media involved can be regarded as approximately the same and system-related,thus the VSP data of the output signals are also related,the method carries out statistical analysis on the data frequency spectrum of each channel of the same excited signals of the zero offset VSP,and the calculated statistical model can effectively suppress the noise frequency while obtaining the effective signal frequency.The statistical model is used to process the abnormal amplitude of the amplitude spectrum of the disturbed seismic records,and finally time domain data is derived to obtain VSP seismic data with casing wave removed.Each step of the above abnormal amplitude processing is realized through simple operation and artificial observation,The final application results show that the casing wave in the zero offset VSP data can be suppressed well,which proves the correctness and effectiveness of the method proposed in this paper,i.e.the method that the statistical frequency model is calculated by using the correlation between each channel data of the same excitation signal of the original zero offset VSP,and used to process the abnormal peak of amplitude spectrum of noisy data can effectively suppress the casing wave.On the basis of demonstrating the correctness and effectiveness of the method,this paper optimizes and improves the implementation of the initial method in view of the limitations of data sorting when creating statistical frequency model and the excessive human subjectivity and other issues: the input signals can be considered to be similar because the excitation signals are all generated by explosives buried at the same depth underground;The depth range of detection points is very large from 700 meters to 3900 meters,but the underground medium involved can be considered to be approximately the same because it is the exploration target area.Therefore,the object of creating statistical frequency model is improved to all VSP data,and an algorithm to identify noisy data is designed to replace artificial observation identification,finally,the improved method is applied to seismic data.The results show that the improved method is more efficient in calculation and simpler in operation while ensuring the effective suppression of casing interference waves,and has very wide practical application value.