Co-Kriging — New Developments

Abstract

Co-Kriging or Joint Estimation utilizes data from correlated variables to improve the estimation of all variables or to compensate for missing data on some variables. The general formulation of Co-Kriging in matrix form was given by the author. The matrix form emphasizes the analogy with Kriging of one variable utilizing only spatial dependence. General conditions are obtained for covariance matrix functions and variogram matrix functions. The extension to block co- kriging is delineated including the Co-Kriging variance. A simple algorithm is given for obtaining the ``under-sampled'' case from the general matrix formulation. Finally a method for reducing the size of the system of equations is given and a iterative method provided which allows solution of even singular systems in which entries are matrices.