Stochastic Simulation of Microseisms Using Theory of Conditional Random Fields

We examine the applicability of conditional stochastic simulation to interpretation of microseisms observed on soft soil sediments at Kushiro, Hokkaido, Japan. The theory of conditional random fields developed by Kameda and Morikawa (1994) is used, which allows one to perform interpolation of a Gaussian stochastic time-space field that is conditioned by realized values of time functions specified at some discrete locations. The applicability is examined by a blind test, that is, by comparing a set of simulated seismograms and recorded ones obtained from three-point array observa tions. A test of fitness was performed by means of the sign test. It is concluded that the method is applicable to interpretation of microseisms, and that the wave field of microseisms can be treated as Gaussian random fields both in time and space.