FCOVARIOGRAM directive

Forms a covariogram structure containing auto-variograms of individual variates and cross-variograms for pairs from a list of variates.


Options

PRINT = string
Controls printed output (statistics, variograms, autovariograms); default stat

METHOD = string
Specifies what to do when the measurements are not all made at the same locations (allwithcrossnugget, allnocrossnugget, commonpoints); default comm

COVARIOGRAM = pointer
Pointer to store the variograms, cross-variograms and associated information for use in MCOVARIOGRAM

MAXLAG = scalar
Maximum lag in all directions

STEPLENGTHS = scalar or variate
Length of the step or steps in which lag is incremented

DIRECTIONS = scalar or variates
Directions along which to form the variogram, scalar for a single direction in 2 dimensions, variate for several directions in 2 dimensions, and pairs of variates for 3 dimensional data

SEGMENTS = scalar or variates
Angles subtended by the segments, defined similarly to DIRECTIONS

COORDSYSTEM = string
Coordinate system used for the geometry for discretizing the lag (mathematical, geographical); default math

MAXCONEDIAMETER = scalar
Diameter at which the segments over which averaging is to be done should cease to expand; default * implies no limit

MINCOUNT = scalar
Minimum number of points required at a particular lag point for the cross-variogram to be estimated there; default 1

DRIFT = string
Mean function (constant, linear, quadratic); default cons


Parameters

DATA = variates
Measurements as a variate

X1 = variates
Locations of each set of measurements in the first dimension

X2 = variates
Locations of each set of measurements in the second dimension (if recorded in more than 1 dimension)

X3 = variates
Locations of each set of measurements in the third dimension (if recorded in 3 dimensions)


Description

The FCOVARIOGRAM directive forms a covariogram structure containing auto- and cross-variograms for pairs from a list of variates.

   The data are supplied as a list of variates using the DATA parameter, where each variate contains the measurements for each variable. The locations of the measurements are supplied using parameters X1, X2 (for two or three dimensions) and X3 (for three dimensions) parameters.

   The METHOD option specifies how to calculate the cross-variograms. The setting commonpoints specifies that only those points observed in common in every sample are to be included; the method described in Section 8.3.4 of Part 2 of the Guide to GenStat are then used. Alternatively, the setting allwithnocrossnugget can be used when the sampling locations do not match. This uses an algorithm outlined in Künsch, Papritz & Bassi (1997) that performs least-squares fitting of the cloud of products of differences to estimate the expected value of these products. If there are no common points, the nugget variance cannot be calculated. However, if there is partial sampling (i.e. some common points), the setting allwithcrossnugget can be used to shift the cross-variograms by the semivariance at the origin to estimate the nugget effect.

   The maximum lag distance in all directions to which the variograms are calculated is set by the MAXLAG option. The increment in distance is set by the STEPLENGTHS option, where you can supply a scalar specifying the length of each step or a variate containing each step. The directions along which to form the variograms are supplied in degrees using the DIRECTIONS option. The geometry used for the directions is given by the COORDSYSTEM option: the setting mathematical specifies directions counter-clockwise from east, and geographical specifies clockwise from north (for the first direction only in three dimensions). Each direction is at the centre of an angular range, which is defined by the SEGMENTS option. For a single direction in two dimensions the DIRECTIONS and SEGMENTS options should be set to scalars, while for several directions they should be set to variates. For directions in three dimensions, the DIRECTIONS and SEGMENTS should specify pairs of variates. The MAXCONEDIATMETER option can be used to specify a diameter at which the segments cease to expand. For cross-variograms that are formed using all points the minimum number of points required at each lag can be specified using the MINCOUNT option.

   The DRIFT option can be used to calculate the variograms after removing a systematic component. Setting the DRIFT option to linear or quadratic will fit a regression to the observations and then form the variograms on the residuals.

   The COVARIOGRAM option allows you to specify pointer to save the auto-variograms, cross-variograms and associated information. Its elements contain:

   1    a matrix with columns of variograms and cross-variograms and rows indexed by lags within directions;

   2    a variate of counts at the lags in each direction;

   3    distances of the lags in each direction;

   4    horizontal angles;

   5    vertical angles;

   6    variances;

   7    distance classes;

   8    method;

   9    pointer containing identifiers of the DATA variates;

   10  number of dimensions.

This structure provides the information required to fit models to the covariogram using the directive MCOVARIOGRAM.

   The PRINT option can be set to statistics to display statistics for each of the variates. The setting variograms displays each of the auto- and cross-variograms, while the setting autovariogram displays only the auto-variograms.

 

Options: PRINT, METHOD, COVARIOGRAM, MAXLAG, STEPLENGTHS, DIRECTIONS, SEGMENTS, COORDSYSTEM, MAXCONEDIAMETER, MINCOUNT, DRIFT.

Parameters: DATA, X1, X2, X3.


Action with RESTRICT

Restrictions are ignored.


Reference

Künsch, H.R., Papritz, A. & Bassi, F. (1997) Generalized cross-covariances and their estimation. Mathematical Geology, 29, 779-799.