OBJECTIVES: to provide tools adapted to phenomena too complex to be represented by stationary or intrinsic models, expanding the hypotheses of stationarity. To study phenomena with drift or partly regular behavior (behavior at low frequencies, for example) for which there is not allowed variogram model compatible with the experimental variograms.


Study of phenomena with a systematic trend or "drift" with two groups of methods:

  • Universal kriging (case of a global drift of simple form): in a first approach, very intuitive, the studied variable is the sum of two components, one (the residual) that can be treated using the methods of linear geostatistics, the other is considered as a drift (or trend). An extension of this KU approach leads to the method of external drift, in which two types of information are present: the variable of interest, measured in a small number of samples, and an auxiliary variable, densely sampled, which is supposed to provide information on the general behavior of the variable of interest.
  • Intrinsic random functions of order k (IRFk, case of a complex drift): functions that transform the data into stationary quantities (analogous to a differentiation of order k), and that filter the more regular components of the phenomenon. The advantage of this technique is that it broadens the range of models by introducing the generalized covariance (generally, the choice of these generalized covariance and their parameters are handled by an automatic module).

The methods differ only in the nature of the stationary objects. The first part of the course focuses on the identification of non-stationary behavior. A numerical example with some regularly spaced 1D data allows us to imagine a simple solution.

Duration: 5 days
Dates: 27/11/2017 - 01/12/2017
Cost: 2 500 €
Including lunch
Language: English
an application for registration
with a CV

Training location : Mines ParisTech, 35 rue Saint Honoré, 77300  Fontainebleau  -  France

modules list                 previous module        next module

Catalog CFSG