APPLICATION OF HEURISTIC SEARCH METHODS TO PHASEVELOCITY INVERSION IN MICROTREMOR ARRAY EXPLORATION

摘要

Microtremor array exploration has been recognized as one of powerful techniques to know S-wave profileof sedimentary layers. In the exploration, phase velocity of Rayleigh wave is estimated from array recordsof vertical microtremors. The phase velocity is inverted to an S-wave profile using optimizationtechniques. Originally, least square methods developed in seismological community were applied in theinversion. Recently Genetic Algorithms which are regarded as one of the heuristic search methods aretried to be used in inverting phase velocity from microtremor exploration, because of robustness of thealgorithms.In this study we compared performances of heuristic optimization techniques in phase velocity inversion.The heuristic methods examined are Genetic Algorithms, Simulated Annealing, and Tabu Search. Weinverted synthetic phase velocity in numerical experiments. The Genetic Algorithms and SimulatedAnnealing show the rapid convergence of the misfits. However, the Simulated Annealing can find modelswith the smallest misfits. We also found the same conclusions in application of the methods to actual data.