Comparison of PSO and GA Metaheuristic methods to invert Rayleigh wave dispersion curves for Vs estimation: a case study

Current techniques of estimating shallow shear velocities for assessment of earthquake site response are too costly to be used at most construction sites. Recently, refraction microtremor (ReMi) data have been frequently used for estimating of dispersion curves and simulating velocity of S-waves. In the current study we applied two metaheuristic optimization approaches, particle swarm optimization (PSO) and Genetic algorithm (GA), for inversion of Rayleigh wave dispersion curves. In geophysical surveys, the application of metaheuristic techniques is quite new. The efficiency of the proposed methods were investigated by synthetic models. At the end, PSO and GA inversion algorithms were tested on a real ReMi data set. Also in this study, resistivity data (Wenner array), as auxiliary information, were carried out. In order to achieve reliable results, processing of ReMi data in particular inversion stage should be done accurately. Similar to other surface wave methods, after the construction of dispersion curves, the next step is to invert of the dispersion curve to obtain a single VS profile. To make reliable geological inference, it is necessary to use accurate models and assess their accuracy. In this study, the reliability of two metaheuristic methods, i.e., GA and PSO for inversion of Rayleigh wave dispersion curves were investigated on two synthetic datasets (i.e., models A and B) and an actual dataset. The inversion was done for S and P-wave velocity by the assumption of 0.2<Poisson's ratio<0.5 and fixing densities to their known values. The procedure was designed to find the global minimum of RMS (root-mean-square) error misfit between the measured and the predicted phase velocities. In this study, the ReMi method was performed using an OYO 12-channal seismograph and 4.5Hz geophones with the receiver spacing of 10m. Also, in this study, resistivity data are used as auxiliary information. The results proved applicability of the proposed inversion algorithms in Rayleigh wave dispersion curve inversion The findings in both synthetic and real dada sets proved that GA and PSO algorithms were suitable strategies for the inversion of surface waves.  Also, comparison of two inversion algorithms showed that PSO algorithm is fast and easy to implement compared to GA, because of the few required parameters adjustment.