AVO Inversion Based On An Improved PP Wave Reflectivity Approximation

Prestack simultaneous inversion as an important technology for evaluating rock physical parameters has been widely used in oil and gas exploration and development.Compared with other inversion methods,such as full waveform inversion,the prestack inversion based on the seismic amplitude varies with offset(AVO)relation can synthesize the multiscale data of logging and rock physics,and also have obvious advantages in the calculation speed.AVO analysis and inversion provide a lot of help for oil and gas identification and reservoir evaluation,so they are very popular in production and application.However,there are still some problems in AVO inversion,which mainly include the following three aspects:(1)there is a relatively large error between the approximate formula of linear reflection coefficient and the exact formula in the case of strong impedance difference.(2)in the process of inversion,the ratio of P-to S-wave velocity should be estimated in advance;(3)the inversion problem is highly ill-posed and requires reasonable regularizations when there are three or more parameters need to be inverted simultaneously.In order to solve the above problems,an improved approximation equation of seismic longitudinal wave(PP wave)reflection coefficient is proposed on the basis of existing research.Compared with the existing linear approximations of reflection coefficient,this equation has higher precision in the medium and small angle range for the interfaces of four different reservoir types.When AVO inversion is performed,it is not necessary to assume that the ratio of P-to S-wave velocity is a constant,which avoids that the inversion results are seriously impact by the error of the constant.The improved AVO equation also reduces the size of the sensitivity matrix conditions in the inversion,and reduces the ill-posedness of the inversion.Based on the improved approximation equation,the corresponding inversion method is developed,and the stability and reliability of the method are demonstrated by model test and practical data application.