Commands

General conventions for gstat command files are:

• command files are ascii text files
• each command ends with a ;
• command files start with one or more data( id) commands to define the data, where id, the identifier, is an unique one-word reminder for the variable defined
• regular file names are written between single or double quotes, as
  'file.dat' or "file.dat", special file names include pipes, shell command output substitution or append-to files (section 5.4)
• white space (spaces, tabs, newlines) is ignored, except in file names

• comment is supported as follows: a # may appear anywhere in a line and gstat will ignore the rest of the line. It will not have this effect inside quoted strings (e.g. file names).

The commands that can appear in command files are listed below. Here, id, id1 and id2 refer to three distinct identifiers, and file is a valid file name. (Commands may be abbreviated to the first few characters that make them unique.)

General options

data(id): body;

here, body defines the data to be read for variable id (section 4.2)

variogram(id): body;

here, body defines the variogram model of variable id (section 4.3)

variogram(id1,id2): body;

here, body defines the cross variogram of variables id1 and id2 (section 4.3)

method: body;

here, body specifies the method (section 4.5)

set parameter= value;

assign value to parameter (section 4.4)

Variogram modelling options

bounds: body;

body is either a list with (white-space or comma-separated) strictly increasing interval boundary values, or it is a file name, pointing to a file that contains such a list

Prediction or simulation options

mask: body ;

here, body defines the input mask map(s), with the locations where predictions will be made (the non-missing valued cells), and the values of the user-defined base functions at map prediction locations (for universal kriging or linear models) or the category number (for stratified prediction or simulation); for multiple mask maps body is a comma-separated list of file names.

predictions(id): 'file';

here, file defines the output map with the predictions on variable id

estimates(id): 'file';

synonymous to the predictions command

variances(id): 'file';

here, file defines the output map that will hold the prediction error variances on id

covariances(id1,id2): 'file';

here, file defines the output map that will hold the prediction error covariances on variables id1 and id2

data(): body;

here, body defines the non-gridded prediction locations

blocksize: body;

here, body defines the block size (default 0, see section 3.5)

edges: body;

here, body is a comma-separated list with files containing open or closed polygons. Edges (boundaries) may be used in interpolation to further constrain a neighbourhood definition: only when a point is on the same side of an edge as the prediction location, it will be included for prediction (or simulation). See Appendix B for details and polygon file formats.

area: body;

here, body defines the `block' discretization points (section 3.5, Appendix A.3)

merge id1( i) with id2( j);

In multivariable ordinary or universal kriging (or simulation), by default each variable has it's own set of parameters $\beta_k$. The merge command allows to define a common parameter for two or more variables. Suppose, $Z_1 (x) = m + e_1 (x)$ and $Z_2 (x) = m + e_2 (x)$, where $m$ is the unknown common mean for both variables. Note: the variable numbers i and j start at 0; merge id1 with id2 is the abbreviation of merge id1(0) with id2(0) (see also Appendix A.2).