​Inherent physical difficulties associated with the effect of low-velocity anomalies on wave propagation, limited data sets, and restricted illumination angles affect the tomographic assessment of piles, caissons, slurry walls, and other similar geotechnical systems. This study evaluates various inversion methodologies for the tomographic detection of low-velocity anomalies. Travel time and amplitude data are gathered in the laboratory by simulating realistic field conditions. The inversion methodology involves data preprocessing, fuzzy logic constraining, and various

forms of tomographic inversion based on either pixel or parametric representations of the medium. It is shown that the tradeoff between variance and resolution in pixel-based inversions can be overcome by adding information, such as regularized solutions, or by capturing the problem in parametric form for a presumed simple geometry. Results show that amplitude-based inversion may be more advantageous than time-based inversion in the detection of low-velocity anomalies; however, consistent coupling of transducers is required. The most robust inversion method tested in this study for the detection of low-velocity anomalies under standard field situations (i.e., limited data and restricted illumination angles)

involves a combination of fuzzy logic constraining followed by parametric-based inversion.