Least Squares Elastic Wave Equation Datuming for Multicomponent Seismic Data

Abstract

Wave equation datuming (WED) is necessary in areas with rugged topography and strong variations of near surface velocity. For multicomponent data, we suggest that the datuming should be implemented directly to the multicomponent data instead of the separated data so that the data can be still available for following processing techniques requiring multicomponent data as input. This can be achieved with elastic WED which, however, usually suffers from crosstalk noise. In order to eliminate such noise, we develop a least squares elastic WED strategy. Receiver point datuming and source point datuming are performed respectively, each of which is implemented using an iterative least squares procedure. We use computed Green’s functions instead of the finite-difference extrapolation in the least squares procedure to achieve high efficiency. A numerical test shows that our method is valid and the crosstalk noise can be effectively suppressed.