|
| | Abstract | | Heuristic algorithms, such as GA and SA have been frequently used in the inversion of surface-wave phase velocity in microtremor explorations. One of the advantages of the heuristic inversion methods is the no requirement of calculation of derivatives of an objective functions and matrix inversions. Therefore the inversions based on the algorithms are so robust. Although the heuristic methods can globally search model parameters in parameter spaces, it is sometimes difficult to estimate model sensitivity in these approaches. In this study Markov Chain Monte Carlo method is applied in surface wave phase velocity inversion for microtremor exploration. The algorithm is one of numerical statistical approaches for calculating probability density function of inversion parameters based on likelihood function from observed data. I first explained the computational procedure of the method. Then, the applicability of the method is examined using synthetic phase velocity data of Rayleigh wave. Resolutions for model parameters were estimated using sampled models from the method. We furthermore estimated variations of predominant periods and amplification factors for the sampled models. It is concluded that this method can be used for the phase velocity inversion similar to the other heuristic methods, but can provide resolutions of parameters of inverted models. |
|
|
|
| |
|