Abstract:

Controlled Directional Reception (CDR) is a reflection tomography technique that accepts seismic traveltimes and slopes (traveltime derivatives w.r.t source and receiver coordinates) and returns a velocity model fitting this data. In contrast to other slope-based methods, it uses parsimonious model parametrization and relies on ray tracing thus being computationally efficient and fairly general. However, it is unstable w.r.t. data errors. In this paper we revisit the CDR method and develop a formalism to mitigate its instability. Our approach is based on linearized estimates of ray tracing errors allowing for suboptimal regularization of the inverse problem. We apply our original ray method asymptotics of the pseudodifferential Double Square Root equation to parametrize the wavefield and test our formulation of the CDR method on two benchmark synthetic datasets. We demonstrate that it provides competitive results suitable for depth migration. We restrict ourselves to 2D settings although the approach can be generalized to 3D problems as well.