What is Geostatistics used for?

Geostatistics has been defined by G. Matheron as "the application of probabilistic methods to regionalized variables", which designates any function displayed in a real space. At the difference of conventional statistics, whatever the complexity and the irregularity of the real phenomenon, geostatistics search to exhibit a structure of spatial correlation. This accounts for the intuitive idea that points close in the space should be likely close in values. In other words randomness does not mean independence.

What makes geostatistics powerful is its capability to characterize by means of a consistent probabilistic model that spatial structure. Therefore the predictions made using the geostatistical methods are tailored to the intrinsic structure of the variable and not only to the samples numbers and sampling patterns. This spatial structure is characterized by the variogram.

Because of its probabilistic framework, the geostatistical approach is claiming that the descriptions of the reality are subject to uncertainty, which can be quantified and provide efficient decision tools for practitioners and managers.

The domain of application of geostatistics is practically unlimited. Each time experiments are made in a defined space ( i.e. data with coordinates and values), a geostatistical approach may be explored.

Because of the large variety of domains and the related specific problems, many methods are now proposed in literature and software. Basically 2 groups of methods all based on variogram or covariance are available:

 for mapping or estimating, the variogram is used to interpolate between the data points, this is the kriging.

 for characterizing the uncertainty on estimates (oil volumes, grade above cut-off, risk of pollution), the same variogram can be used in a different way for making simulations of the unknown reality.