declustering {RGeostats}R Documentation



Calculating the Declustering Weights


declustering(db, method = 1, radius = NA, model = NA, neigh = NA,
             dbgrid = NA, ndisc = NA, flag.sel=method!=1, verbose=FALSE,
             radix = "Decluster", = db.locmod())



The db-class where the Declustering Weight must be stored.


This argument defines the method used for calculating the declustering weights:

  • 1 For each sample, the declustering weight is proportional to the number of samples contained in its neighborhood.

  • 2 The declustering weights are the kriging weights when estimating the drift for the IRF-0 in Unique neighborhood

  • 3 The declustering weights are the cumulative value of the kriging weights for the bloc estimation of the input variable on a grid.


A numeric vector containing the radii of the search ellipse. The length of this vector should be equal to the space dimension. If a single value is provided, it is considered as the radius of a search sphere. If 'radius' is missing, the radii of the defaulted ellipsoid are calculated as 1/5 of the field extension.


The model-class which describes the geostatistical structure to be used in the computation of the clutering weights using the methods based on Kriging ('method' = 2 or 3). If not provided a linear model is defaulted.


The neigh-class which describes the neighborhood parameters used for the estimation ('method' = 3). If 'neigh' is not defined, it is generated by default as a Moving Neighborhood, with an optimum number of samples equal to 100 and ellipsoid radii equal to 1/5 of the field extension.


The db-class object which contains the blocks to be used for the computation of the declustering weights as the cumulative value of the kriging weights ('method' = 3). The ordinary kriging is performed only on the active blocks if a selection is defined.


A numeric vector containing the discretization count along each direction of the space. This parameter is used for the calculation of the declustering weights using 'method' = 3. If 'ndisc'is not provided, this value is set to 3 in each space direction.


Create a selection masking off the samples whose declustered weight is 0. In case 'method' is set to 1, this selection is not created (by default) as no weight can be set to 0.


TRUE for a verbose output (statistics on the calculated weights).


Radix of the name given to the variables storing the declustering weights in the target Db.

Decides whether or not the newly created variable will have its locator defined or not. For more information, see db.locmod.


The input Db where one variable has been added which contains the declustering weights. This variable is set to the locator "w" for future usage.

[Package RGeostats version 11.1.2 Index]