Input and Output Formats
Text tables
merge1d and ssh1d accept ordinary text tables, in addition to NetCDF
series. The first column is the coordinate and subsequent columns are fields.
A participating set of tables must have the same number of field columns and
strictly increasing coordinates after input scaling. Scaling does not reorder
table rows. merge1d cannot use both text and NetCDF sources in a single
run.
NetCDF variables
GQ reads single and multiparameter NetCDF files. Compatible fields share coordinate dimensions. Though the names of equivalent fields may differ between source files. Coordinate variables are identified through standard axis metadata and conventional x, y, and z names.
merge2d NetCDF inputs can be ordinary grids or selected layers from
three-dimensional cubes. Three-dimensional model variables may store
their dimensions in any order when the coordinate relationship is
unambiguous. However, GQ output cubes use a (z,y,x) order.
When SSH application changes a model lattice through -R, -I, or
-T, scalar ancillary variables in the NetCDF file are unchanged.
Unselected variables that depend on a changed coordinate are omitted
because they no longer match the output lattice.
Output variables
The merge modules may append a shared variable named weight or write only
that variable. Its long_name is merging weight, its units are 1,
and its range is from zero to one.
The SSH modules use fractional perturbations. In application mode, selected
model values are multiplied by 1 + epsilon. A taper requested for output
is also called weight, with long_name set to
heterogeneity taper weight.
GQ writes missing output values as NaN. See Missing Values.