Designing successful applications of elastic full waveform inversion to 3D land seismic data is challenging. As a preliminary step to investigate such a difficult topic, a realistic target extracted from the SEAM II Foothill benchmark is considered. In such environments with complex near surface (topography and complex near surface properties), three main difficulties are faced: complex and energetic surface-wave masking body waves, unknown source wavelet, and potential trade-offs between multiple parameters describing the subsurface mechanical properties. In this study, we focus on the two first of these difficulties. Relying on the spectral element modeling and inversion code SEM46, we implement an automatic workflow based on a hierarchical data separation designed to introduce surface waves in a later step in the inversion process. We also implement an efficient time-domain source-wavelet estimation allowing to update the source estimation at each iteration of the inversion at almost no cost. These two ingredients make possible the successful reconstruction of both P-wave and S-wave velocity models for the target we consider.