On the method of multidimensional Q-factor and velocity inversion with VSP data

摘要

Based on the theory of one-way wave propagation in a 1-D viscoelastic medium, we propose an inversion method to estimate the Q-factor and velocity using the Gauss-Newton method. To speed up the computation, we derive an analytic expression of the Jacobi matrix. As the forward simulation is one of the base steps of the inversion problem, we derived the transfer matrix to simulate the forward model with direct downgoing wave and primary reflection (upgoing wave) of zero-offset VSP data. We give the workflow of both the forward and inversion method, and discuss the optimum initial value determination. Test on synthetic VSP data shows that the proposed method doesn't need a priori information of interfaces positions and can eliminates the influence of upgoing wave at the interfaces. By applying to field VSP data, we obtained Q and velocity curves which have almost the same distribution as that obtained from a known well. It's a powerful tool to predict potential high productivity gas well.