Resolution of high-frequency Rayleigh-wave data

It is essential to understand the potential resolving power of surface wave data in interpreting inverted vertical (layered 1D) and horizontal ('pseudo-2D') shear (S)-wave velocity models. The model resolution matrix indicates that the S-wave velocity model of a layered earth model can be perfectly resolved in a least-squares sense if error-free data are inverted. However, we show that errors in real data introduce a smear matrix in the least-squares solution, reducing the resolution of an S-wave velocity model. A field test shows how accurate dispersion curves can resolve a low velocity layer of about 25% contrast at 3.5 m depth below a stiff overburden. The horizontal resolution of surface wave surveys is limited to the aperture (length) of the recording array. In phase velocity measurements, the layered model estimate represents an average over this length. Any lateral discontinuities introduce a systematic error into the inversion, leading to acquisition geometry-dependent estimated models. The multichannel analysis of surface waves in roll-along mode allows construction of 'pseudo 2D-sections' by aligning 1D models at each spread midpoint. However, images from field data between forward and reverse can be inconsistent. We apply a generalized inversion to these 'pseudo 2D-sections' in the horizontal direction, which removes these inconsistencies and provides an improvement in horizontal resolution, allowing a more objective geological appraisal.