Stochastic Signal Processing > The Theory of regionalized variables

Probability models or Geostatistics is about solving estimation problems related to a regionalized variable. In seismic signal processing, geostatistics is used for optimizing the parameterization of processing operators, speed the computation time, and at the same time provide a quantification of the reliability of the processing operation, that is of uncertainty.

The theory underlying Geostatistics is called the theory of regionalized variables as it has been named and developed by Prof G. MATHERON at School of Mines of Paris in Fontainebleau (France) from 1968 to 1995. It is about applying the theory of probability and statistics to data set with coordinates.

It accounts for the facts that data values are correlated in space, that their spatial distribution depends on the size of the measurement support, which is called the support effect, and on the density of the data sampling which is called the information effect.