Stochastic Signal Processing > Simulations

Another way to build numerical models of a random function Z is to reproduce its spatial variability as it is known by the stationarity order, average and spatial covariance function or variogram . A number of stochastic processes enable to generate numerical spatial simulations that are equi-probable realizations of Z with same mean and covariance. At any location x, the average of simulations is equal to kriging estimate and the dispersion of simulated values equals the kriging estimation variance. Simulations are said to be conditional when they honour the existing available information.