Increments for (CO)Kriging with Trend and Pseudo-Covariance Estimation

摘要: This paper presents an integrated account of the interpolation techniques universal kriging and cokriging by way solution a prediction problem in situation akin to linear regression. Polynomial expectations degree k coordinates region are eliminated consideration increments only, followed introduction covariance structure polynomial pseudo-covariance functions. Then best unbiased predictor is derived as well variance ensuing error, both interpretable some analogy regression situation, but essentially different because non-uniqueness functions absence unconditional positivity matrices involved. Estimation essential coefficients accomplished restricted maximum likelihood method which may be helpful deciding about value well.