Kriging Data with Measurement Error: A Review and a Generalized Approach

Abstract

AbstractSamples used in the mining industry are affected by sampling and analytical errors. An approach called filtered kriging with parametric error (FKPE) is proposed. FKPE is defined by the spatial continuity model of the underlying process of interest and the error associated with any number of subgroups of samples with similar behaviour. The proposed method overcomes the limitations of available methods as it deals with correlated grades, non-stationary errors, and errors correlated among samples. FKPE is mathematically compared to different methods and then illustrated in a synthetic example. Results show that FKPE is more general than available techniques and leads to the same results as standardized co-kriging in the known sampling error. FKPE does not require defining a linear model of co-regionalization.