On Simplifications of Cokriging
Due to the number of variables or of data, cokriging can be a heavy operation, requiring simplifications. Two basic types of simplications, with no loss of information, are considered in this comprehensive paper. The first type of simplifications consists, in the isotopic case, in reducing cokriging to kriging, either of one or several target variables, or of spatially uncorrelated factors. The example of variables linked by a closure relation (e.g. constant sum, such as the indicators of disjoint sets) is in particular considered. The other type of simplifications is related to some particular models that, in given configurations, screen out a possibly large part of data. This results in simplified and various types of heterotopic neighborhoods, such as collocated, dislocated, or transferred.
KeywordsAutocorrelation Cross Correlation Kriging Cokriging
Unable to display preview. Download preview PDF.