Up/Down Wavefield Decomposition by Sparse Inversion

Expressions have been derived for the decomposition of multi-component seismic recordings into up- and downgoing constituents. However, these expressions contain singularities at critical angles and can be sensitive for noise. By interpreting wavefield decomposition as an inverse problem and imposing constraints on the sparseness of the solution, we arrive at a robust formalism that can be applied to noisy data. The method is demonstrated on synthetic data with multi-component receivers in a horizontal borehole, but can also be applied for different configurations, including OBC and dual-sensor streamers.