Inversion Uncertainty in Surface Wave Analysis

摘要

This study deals with uncertainties associated to solution non-uniqueness in surface wave analysis, with special attention to the influence of maximum available wavelength in experimental data. Typically, it is assumed that the investigation depth is about half the maximum available wavelength in the experimental dispersion curve. In order to assess the validity of this assumption, three different datasets (one synthetic and two real world case histories) have been analyzed with a Monte Carlo approach. Low frequency experimental data have been progressively scalped to assess the consequences on the reliability of the solution. For the selected case histories, a reasonable estimate of average parameters is obtained with a limited amount of experimental information. Nevertheless, the risk of reaching a local minimum in the inversion process is associated to the availability of experimental data. The use of global search method has to be preferred as they can provide an estimate of the reliability of the solution.