I tested both algorithms in a common midpoint gather from Brazil offshore (Figure a). The subsurface is flat-structured up to 3.1 sec ( 4.6 Km) and have slightly dipping layers below. The data is strongly contaminated by water-layer reverberation that was not effectively removed by conventional deconvolution and dereverberation processes. The relatively long cable allows for the observation of crossing primary-reflections, and some coherent noise with linear stepout is also present. To cope with all these factors I applied a special processing sequence to the data prior to the inversion step.