A New Misfit Function for Multimode Dispersion Curve Inversion of Rayleigh Waves

Inversion of the Rayleigh wave dispersion curve can effectively obtain the shear wave velocity and stratum thickness. And the classical inversion of multimode Rayleigh wave dispersion curve requires correct mode identification. However, there may be "mode-kissing" and "mode-jumping" phenomena in the Rayleigh waves when the stratum contains complex structures such as low-velocity soft intercalations or high-speed stiff sandwich layers. These phenomena can easily lead to mode misidentification, and lead to wrong inversion results. At the same time, the classical dispersion curve inversion method needs to seek the roots, which leads to that the nonlinear inversion of Rayleigh wave is slow and the computation time is long. In view of these, the classical Haskell-Thomson dispersion curve forward modeling algorithm is improved, and a novel and effective misfit function is proposed. The misfit function is directly used to fit the dispersion function surface shape of the iterative updating model by using the measured dispersion curve. It is not necessary to assign the multimode dispersion data to a specific mode, which can effectively avoid the mode misidentification in the inversion of multimode Rayleigh wave dispersion curve. And the misfit function does not require the seeking root operation, thus greatly accelerating the nonlinear inversion speed. In this paper, based on the particle swarm optimization algorithm, three theoretical geological models and a certain roadbed test data often encountered in practical work are used to calculate the theoretical model and analyze the example. And the validity and practicability of the new method of Rayleigh wave multimode dispersion curve inversion are verified.