Context

• Use residuals: this option is only available if residuals have been calculated. If the toggle is activated the variogram will be calculated on residuals.

• Model option: this option allows you to specify on which variable(s), variogram(s) will be computed.

  • Compute raw model only: the modeled variogram is based on the raw variable.
  • Compute Gaussian model only: the modeled variogram is based on the gaussian variable. This option is only available if a gaussian variable has been calculated.
  • Compute raw & Gaussian models: two variograms are modeled: a variogram based on the raw variable and a variogram based on the gaussian variable. This option is only available if a gaussian variable has been calculated.
  • Compute residuals model only: the modeled variogram is based on the residual. This option is only available if residuals have been calculated.
  • Compute residuals & Gaussian models: two variograms are modeled: a variogram based on the residuals and a variogram based on the gaussian variable. This option is only available if residuals with a known drift have been calculated and if these ones have been transformed via a gaussian anamorphosis.

• Stationarity option: these stationarity options and their possible associated parameters will be used during kriging and simulation processes. This information will be saved in the Geostatistical Set.

  • Intrinsic stationarity (Ordinary Kriging): Select this option if you wish to perform Ordinary Kriging. No additional parameters are required.

    The Ordinary Kriging method is based on the minimization of the error variance linear estimate at a location. It does not make any prior assumptions about the mean (contrary to the Simple Kriging). As Ordinary Kriging requires the intrinsic stationarity hypothesis, it can be deal with variograms with or without sills.

  • Strict stationarity (Simple Kriging / Rescaled Cokriging): Select this option if you wish to perform Simple Kriging and define global means for the different data variables. If the variables you are working on are the gaussian transformations or residuals (calculated from a known drift), no mean is asked because it is set to 0.

    The Simple Kriging method can be applied once the mean of each data variable is assumed to be known precisely (equal to a constant or defined by a variable). No constraints are attached to the kriging weights. It implies that the model must be strictly stationary, the only basic structures that are allowed being variograms with sills.

    This option is also required to perform a Rescaled Cokriging on a multivariate model. It aims at reducing the risk of negative estimates because of negative weights assigned by Ordinary Cokriging to the secondary variables. To apply this method, the values of the means of all variables must be known.

  • Non stationary (Kriging with Drifts): This option is only available if you are working on residuals calculated with Inferred Drift(s). It enables you to perform Kriging with External Drift. The variogram used for the kriging is fitted on the residuals.