پيشگويي فضاي بيزي براي يك ميدان تضاد في گوسي با مشاهدات داراي نوفه

Bayesian Spatial Prediction for a Gaussian Random Field with Noisy Observations

Spatial prediction of a Gaussian random field in unmeasured sites based on precise observations is easily carried out. But, in practice, because of measurement errors, data contain noise. We assume that noises are independent random variables with distribution N(e, z2) and they are also independent of the interest random field. If parameters of the mean, covariance function and rz are known, the optimal predictor and its MSE could be determined by usual methods. But, these methods are not desirable when some of the model parameters are random variables. We use the Bayesian approach to determine the optimal predictor and its MSE.