Conclusion

A robust and fast dispersion curve algorithm for one-dimensional models is developed and tested in representative cases. However, the sensitivity study carried in this chapter is far from being exhaustive. The objective is limited to the determination of the significant parameters which might be inverted. Traditionally, $V_s$ is the only one parameter included in the inversion of dispersion curves. Nevertheless, this work demonstrates that, in some cases, $V_p$ has also a non negligible influence. The ellipticity and the auto-correlation curves can be easily computed as well. For each spectral property, a misfit function is defined. These forward algorithms can be used in a non-linear and stochastic inversion such as the neighbourhood algorithm (chapter 4).