***** File IMAGFORM.TXT
Image Data Descriptions
The International Halley Watch agreed early on in the project that
all data would be submitted from the individual disciplines to the Lead
Center using the FITS format. When the decision was made to distribute
this information on CD-ROM, it was determined that the data had to have
even broader accessibility. For this reason, the original FITS files,
with contiguous headers and data, were split into separate files. The
original FITS byte stream could then be recovered by concatenating the
appropriate header and data files. PDS labels were constructed to allow
definition of the data files for the Planetary Data System. For each
data there must always be a FITS header associated.
The convention for naming files on the CD-ROM was proposed by the
IHW Lead Center to include a unique data qualifier for the data. A
specific set of rules was established to identify the net and subnet for
each discipline (given by a letter code). A CD-ROM running number relates
the information contained in the various indices to the files containing
the actual data. A short list of this convention for images follows:
PDS Object FITS Discipline Letter File Ext
(description) NAXIS = Code
image 2 IR Studies IRIM .img
image(compressed) 1 Large Scale Phen LSPN .imq
image(browse) 2 " LSPN .ibg
image 2 Near Nucleus NNSN .img
image(multiple) 3 " RSOC .img
image 2 Spectroscopy SPEC .img
Note that the Subnet coding (explained in the Filenaming Conventions)
is one way to distinguish file extension and content.
The file extensions follow suggestions by the Planetary Data System
(SPIDS v1.1) for tabular and image data. In addition, for IHW FITS, the
original headers and data were split into separate files, with filename
extensions as listed below.
.IBG - data records for subsampled browse image
.IMG - image data records
.IMQ - data records for compressed image
These PDS labels are metadata (as headers describing data submitted to
the archive). There has been no effort to duplicate the documentation
contained in the full FITS headers because the PDS and FITS headers for
a given data file differ only in the filename extension. Instead we
have attempted to use the power of the PDS label syntax to fully
describe the data structures and thus gain access to software by that group.
Standards for the Preparation and Interchange of Data Sets, Martin, T.
Z., et al, Document D-4683, Jet Propulsion Laboratory, California
Institute of Technology, Pasadena, CA document version 1.1 was the
primary reference to the Object Description Language (ODL) necessary to
create the PDS labels. (We thankfully acknowledge the help of M. Martin,
JPL, for assisting the IHW through version 1.1 of the ODL that allows
for SPECTRUM.) The basic PDS descriptors such as SFDU_LABEL, RECORD_TYPE,
RECORD_BYTES, and FILE_RECORDS are explained in the SPIDS document. The
RECORD_TYPE for all data files is FIXED_LENGTH. The PDS labels have
been formed as fixed length (78 byte) with an embedded CR and LF.
In the case of images, we have included a new keyword describing
the byte ordering of the data (MSB_INTEGER) required by FITS. In PDS,
images (.IMG, .IMQ, .IBG) are in terms of LINES (FITS keyword NAXIS2) and
SAMPLES (FITS keyword NAXIS1), given knowledge of the SAMPLE_BITS (FITS
keyword BITPIX), and are easy for the split files. The final form of the
label for compressed images is still under active discussion. Unlike
previous PDS efforts with compressed images, we chose not to compress
the header (or label) and thus have included a keyword to describe the
type of compression (ENCODING_TYPE = "PREVIOUS PIXEL") used. The label
for compressed images also contains information to permit software to
skip over the data if the decoding algorithm is unknown (ITEMS,
ITEM_TYPE, and ITEM_BITS). We use ODL to indicate various subclass
structures for the data objects. An example of this is the DIFFERENCE
modifier applied to IMAGE, yielding the keyword DIFFERENCE_IMAGE, which
indicates that a processing step was applied to the original image.