This paper addresses the problem of estimating the shape parameters of seismic wavefields in linear arrays. The purpose is to separate waves by exploiting the diversity of the polarization state and propagation velocity between waveflelds. In this paper, we propose a method based on the shift-invariance properties of multiple waveflelds impinging on a uniform linear array of sensors to separate different waves by jointly estimating the velocity and polarization parameters. The surface waves are removed using a spatial filter specifically designed to exploit velocity and polarization of different waves. We introduce a model for wideband polarized signals received by an array of three-component vector-sensors and use it as a framework for implementing the proposed algorithms. Moreover, we analytically derived Cramér-Rao lower bounds on the performance of the proposed model parameters. Examples using simulated and experimental data illustrate the applicability of the proposed methods.