| One of the factors, which contributes to errors in shear wave velocity profile obtained from the inversion of surface wave dispersion data is non-uniqueness due to the limited number of field dispersion data. In this research, a new procedure is developed to improve the uniqueness of the shear wave velocity profile resulting from the inversion. A new forward modeling algorithm using the smallest absolute eigenvalue as the screening parameter to generate Rayleigh wave modes from a theoretical model is developed. The theoretical model adopted in this research is the Dynamic Stiffness Matrix. The results indicate that the new technique is more reliable than the traditional method using the determinant as the screening parameter.; The performance of the Broyden-Fletcher-Goldfarb-Shanno and Levenberg-Marquardt methods are evaluated in this research to determine the most suitable gradient method for surface wave inversion. Comparison of the performance of the two methods shows that the Levenberg-Marquardt method produces more accurate results than the Broyden-Fletcher-Goldfarb-Shanno method.; An updated inversion technique which divides the inversion process into a number of stages, with each successive stage utilizing the shear wave velocities obtained from the previous stage as its initial model, is introduced. The number of stages is the same as the highest Rayleigh wave mode number, and the kth stage of the inversion utilizes the dispersion data from the 1st through kth modes. Shear wave velocities obtained from the updated inversion technique are more accurate than those obtained from the inversion procedure using an initial model constructed from fundamental mode dispersion data. |
