CONDITIONAL SIMULATION OF NON-GAUSSIAN RANDOM-FIELDS FOR EARTHQUAKE MONITORING SYSTEMS

The problem of conditional simulation of random fields gained a signif icant interest recently due to its applications to urban earthquake mo nitoring. In this paper, for the first time in the literature, the met hod of conditional simulation of non-Gaussian random fields is develop ed. It combines previous techniques of iterative procedure of uncondit ional simulation of non-Gaussian fields, and the procedure of conditio nal simulation of Gaussian random fields. To contrast the agreement be tween the simulated correlation function and targeted correlation func tion, the numerical error is decomposed into two parts, namely, into s imulation error and mapping error. Simulation error can be reduced by increasing number of samples while mapping error is eliminated by the suitable iteration procedure. In this paper univariate and time-indepe ndent random fields are considered. Numerical example shows that the c orrelation structure and probability distribution of the simulated ran dom field have excellent agreements with given correlation structure a nd probability distribution, respectively.