A stable sparse pre-stack reflectivity inversion with time-distance relation constraint

摘要

We present a new method for inversing the locations and P-impedance reflectivity series from pre-stack seismic data without NMO-correcting. Compared to the conventional pre-stack inversion, we focus on improving stability and resolution other than inversing those elastic parameters such as S-velocity and density. Hyperbolic time-distance curve is used as a constraint in our inversion based on horizontally layered medium which is consistent in most cases. We make use of the time-distance relation to improve vertical resolution. Because the time locations of a given number of interfaces are unknown and the reflection time-distance curve is nearly hyperbolic, then the problem is nonlinear and we adopt the global optimization algorithm known as simulated annealing. Two synthetic examples show the result is stable with high resolution and robust under noisy conditions.