Study On Multiwave Prestack Elastic Parameters Inversion

As the oil and gas seismic exploration goes deeper and more complicated,the prestack inversion,as an effective fluid identification and reservoir predication approach,is widely used in practical seismic exploration.However the prestack inversion also have some problems such as low accuracy and instability because of the observation error and incorrect inversion model.Thus,this paper carries out some research about some existed problems to improve the accuracy and stability of prestack inversion,and the corresponding solutions of the researched problems are proposed,and the multiwave prestack inversion is introduced.If we do not have velocity information when extracting P-wave velocity,S-wave velocity,and density from elastic impedance(EI),the k(ratio of S-wave velocity and P-wave velocity)is always set to a constant,which is different from the real k.So this assumption will leads to errors in the inversion results.For this problem,the influence of k on the inversion results is analyzed firstly by the model resolution matrix,and we found that the P-wave velocity and density are hardly affected by k,whereas the S-wave velocity is very sensitive to k.Based on this point,we design a nonlinear k value iteration method to obtain accurate k value and S-wave velocity.Then we derive the P-SV wave elastic impedance(SEI)formula based on AkiRichards P-SV wave reflection coefficient approximation formula.Finally we propose a joint iteration method of k value to extract P-wave velocity,S-wave velocity and density from EI and SEI.In order to improve the stability and lateral continuity of the AVO(Amplitude Versus Offset)inversion results when lacking constraint of the low frequency model,a new AVO inversion constraint method is proposed by combining the Bayesian principle and Kalman filtering algorithm to constrain parameters in both vertical and lateral directions.Firstly,the mathematical model of the Kalman-filtering-based AVO inversion system is established by combining the vertical Bayesian prior probability constraint and the lateral continuity assumption of inversion parameters,then this mathematical model is exploited in the Kalman filtering algorithm frame to achieve the constraint of parameters in both vertical and lateral directions.Most of the traditional prestack inversion methods construct objective functions based on least square(LS)method and then solve the objective functions using linear or quasi-linear method to get the model parameters.This paper introduces a more straightforward method,the accumulation method,to extract parameters from elastic impedance.The accumulation method does not process the random error,but uses accumulation to estimate parameters.We test the above proposed methods by using layered-model,partial Marmousi2 model and field data.The test results demonstrate that(1)the k iteration method does not rely on the velocity model,and the k value converges to its real value quickly and accurately,thus the inversion error caused by the error of k value is reduced significantly,(2)the integration of converted S-wave can improve the accuracy of the inversion result,(3)the vertical and lateral directions constrained method can obtain inversion results with better lateral continuity,(4)the LS and accumulation methods provide comparative results.