Studies On High-accuracy Q Inversion Methods

As an inherent parameter of rock,quality factor,Q-value,plays an important role in seismic exploration.In seismic data processing,we can employ accurate Q estimation results during the process of inverse Q filtering to improve seismic resolution;meanwhile,in seismic interpretation,we can use accurate Q estimation results to evaluate the elastoplastic of subsurface rocks.What is more meaningful is that a lot of interesting reservoir properties such as hydrocarbon bearing,the degree of fracture development,rock brittleness and so on are related to the elastoplastic of rock.In this case,we can reasonably evaluate the above reservoir properties by using accurate Q inversion results.So,it's very meaningful to improve the accuracy of Q inversion results.The common used Q inversion approach is spectral ratio method.However,this method has many limitations when it's used to field data inversion.These limitations mainly refer to two aspects: firstly,seismic tuning effect is the most knotty factor in the use of spectral ratio method.Spectral ratio method can't be used because of the spectral overlap of tuning seismic waves,which leads to the difficult of extracting single wavelet's spectrum;secondly,it's also important to improve the stability of spectral ratio method.Natural spectral ratio value is very sensitive to noise.It's difficult for spectral ratio method to obtain stable results when the S/N of seismic data is low.In this thesis,to provide a foundation to the numerical simulating,the author researches the characteristics of the common used viscoelastic model(Kolsky-Futterman model,KF model),and then the author implements the algorithms of viscoelastic VSP and post-stack simulating based on KF model.To tackle the two limitations mentioned in the above,the author implements the following algorithm studies and reformations based on post-and pre-stack seismic data:(1)To deal with the effect of seismic tuning,the author introduces two-parameter generalized S transform(GST)during the procedure of extracting amplitude spectrum.Unfortunately,the amplitude spectra are polluted by the frequency response of Gaussian window function which is employed in GST.To eliminate this fatal effect,the author deduces a novel spectral ratio method inversion formula??GSQI.In this equation,the Fourier spectrum of source wavelet is considered as Gaussian function.Meanwhile,the author defines a new parameter,?,in GSQI formula.What is interesting is that there is a linear relationship between spectral ratio and ?.The slope of this linear relationship is 1/Q.Synthetic tests and field data applications verify that after performing GST to seismic data the relationship between spectral ratio and frequency is non-linear whereas the relationship between spectral ratio and ? is linear,which demonstrates the accuracy of GSQI formula.(2)We have to extract the local maximum amplitude spectra on time-frequency map for the calculating of spectral ratio values,which is decided by GSQI method.However,the data noises on time-frequency map have large effect on the choice of such local maximum amplitude spectra.The author found that if we adopt different threshold values in the procedure of extracting local maximum amplitude spectra different results will we get.To overcome the above instability,the author adopts basic pursuit high-resolution inversion method(BPI)to constrain seismic horizons' location.We can directly extract the local maximum amplitude spectra at our interesting horizons' location.Then the GSQI formula should be used.This novel approach is called BP-GSQI.Numerical tests and field data applications testify BP-GSQI should efficiently improve the accuracy of GSQI.(3)Generally speaking,S/N is relatively low for per-stack seismic data.To improve the anti-noise ability of common used pre-stack Q inversion method,two-step QVO,the author develops a new approach called one-step QVO method.The essence of two-step QVO method is two-step linear fitting.Its robustness can't accommodate the situation of low S/N.In this thesis,the author treats spectral ratio as a surface which is related to frequency and travel-time difference/square of offset.This surface is called spectral ratio surface.The expression of spectral ratio surface should be easily obtained by modifying the two-step QVO equation.Then we can calculate Q-value by one-step surface fitting approach.In fact,two-step QVO method just uses the information along two directions(directions of frequency axis and travel-time difference/square of offset axis),whereas one-step QVO method takes full use of the whole information on spectral ratio surface and the surface distortions should be corrected more sufficiently.Therefore,one-step QVO method is more stable than two-step QVO method.Numerical tests verify the ability of anti-noise of one-step QVO method is superior to traditional two-step QVO method.