Header description

The header is divided in sections, as follows. The exact header definition can be found in appendix B.

1. Leading information: vital information describing the file format.

2. DIMENSION: describes the dimensions of the data in the file.

3. BLANKING: the blanking and tolerance values.

4. EXTREMA: the value and position of the minimum and maximum value in the data.

5. COORDINATE: the axes definition for each dimension, encoded as a 2D array of $N_{dim}$ (ref,val,inc) triplets.

6. DESCRIPTION: the data unit and axes names. The first header block ends here.

7. POSITION: the source description. The second header block starts here.

8. PROJECTION: the projection definition.

9. SPECTROSCOPY: the description of the spectroscopic axis.

10. RESOLUTION: the beam characteristics.

11. NOISE: the data noise.

12. ASTROMETRY: proper motion parameters.

13. UV_DATA: support for UV tables, e.g. description of extra-columns specific to this kind of tables.

The leading pseudo-section and thus all the GDF files (V1 or V2) start with 12 specific characters:

• the 6 firsts are always the word GILDAS (uppercase),
• the 7$^{th}$ is a unique character encoding the system type and the version of the GDF used (see Table 1),
• the 5 last are a subsequent word used to recognize the kind of data in the file, namely IMAGE (for standard Gildas images or N-cubes) or UVFIL (for Gildas UV tables).

In GDFV2, this last character string is only a first order information about the file kind. Distinction between Images, Tables, UV Tables, VO-like Tables, is provided by an integer keywords (see Table 2).

Table 1: GDF encoding of the system and version. Most of the recent computers are little endian (IEEE) machines. The obsolete system VAX is not supported with GDFV2. 1: hyphen, 2: underscore.

Version	IEEE	EEEI	VAX
1	$-^1$	.	$\_^2$
2	$<$	$>$	N/A

Table 2: List of kinds available for GDFV2 files (h%gil%type_gdf). The values are integers, but should not be used explicitely.

Name	Value	Comment
code_gdf_image	0	Normal data, i.e. IMAGE...
code_gdf_uvold	1	Old UV Data with wrong weight
code_gdf_uvt	10	UV Data in "visibility" order
code_gdf_tuv	-code_gdf_uvt	The transposed way (channel order)...
code_gdf_table	20	A simple Table, with no information
code_gdf_vo	-code_gdf_table	Virtual Observatory (VO) Tables are ordered in the
		transposed way...
code_gdf_xyt	40	A specific CLASS table (by symmetry with code_gdf_uvt)
code_gdf_txy	-code_gdf_xyt	A specific CLASS table (by symmetry with code_gdf_uvt)
code_gdf_sheet	50	A spread sheet for the SIC Tabler...

Then follows the data format, the various number of blocks in the file, the version of the GDF currently in use, and the kind of GDF file.

Each section has also a parameter describing its length. A zero value means that this section is not present in the file and/or not filled in memory. For the programmer or the end-user, any other value means it is enabled (then all the parameters in the section are expected to be filled).

In details, some of the sections have a fixed length for any file (e.g. BLANKING is $2+2$ 4-bytes words) and some have a length depending on the number of dimensions in the file (e.g. COORDINATE is $2 + 2 \times ( N_{dim} \times 3 )$ words).