The set of ISOCAM data products associated with an observation can be categorised, depending on how they are derived or used, as one of the following:
They contain information about the status of the instrument; the status of the telescope; the pointing history; and, last but not least, the scientific data. Their root names begin with the letter `C' for CAM while the rest of the name serves as a mnemonic of the contents. A complete list of ISOCAM data products is shown in Table 5.2. In addition to the discussion of the early stages of the pipeline below, details of the instrument model and of the high and intermediate level products can be found in the dedicated Chapters 6 and 7, respectively.
ISOCAM's raw data files are simply binary tables of chronologically
ordered rows of records in which each column contains some piece of
imaging or housekeeping
data. It is usually the job of the analysis software to read in such data
sequentially and incorporate them into more coherent astronomical structures
with other data in memory or files. For higher level files, wherever
possible,
products have been designed to conform to the conventions described in the
FITS User's Guide, particularly where images are concerned. This also
applies to calibration files. The OLP system
worked under the constraint of producing only one file of a given type per
observation, no matter how many different images were taken. The standard
FITS mechanism for delivering single images uses the PRIMARY array and a
conventional set of keywords to define the coordinate system, wavelength
and other
associated data. In order to put several images into one CMAP or CMOS file
these PRIMARY array data structures were reproduced in the columns of
the binary table
in a series of three consecutive records labelled FLUX, FLUX_ERROR and
EXPOSURE for an individual image. Despite these efforts, it has been brought
to attention that some FITS readers are not able to deal with a standard
image format in this modified context although any deficiencies have an
easy workaround. In other products containing physical
units, such as lists of detected point sources or their spectra, for which
there were no common standards when the design was taking place, we adopted
as many standard column names as possible to make the structures
easily intelligible. Occasionally, NULL values were needed in some binary
tables and their appearance in ISOCAM products does not always conform to
FITS standards because of platform dependencies that it was not possible to
avoid. The most important NULL values, those that appear regularly in
images such as the LW channels dead column 24, do conform via the standard
use of scaled integers to
represent pixel flux and other values. Any NULL integers could similarly be
handled through their associated TNULL keywords. The platform-dependent
problems arose due to a lack of support for IEEE NaN (not-a-number)
conventions. Instead, the user will encounter the value 
1.2E34 to
show NULLs in, for example,
the columns CCIM[1].RA(*), CCIM[1].DEC(*) and CCIM[1].ROLL(*) for those
detector images and instrument modes in which the sky was not
observed.
In addition to mandatory FITS keywords, all ISOCAM products contain a small number of common PRIMARY keywords describing the production system, most of them self-explanatory. The VERSn keywords show the files used to derive the product.
For example:
ORIGIN = 'ESA ' / Not from central ESA archive TELESCOP= 'ISO ' / Infrared Space Observatory INSTRUME= 'CAM ' / Instrument used COMMENT CAM Image Edited Raw Data FILENAME= 'CIER12900907' / File name DATE = '2001-07-02' / Creation date 2001/183 FILEVERS= '2523 ' / Version ID (derived from creation date) OLPVERS = 'OLP_95 ' / SOC OLP system version CALGVERS= 'CALG_65 ' / SOC OLP CAL-G files version USERNAME= 'APOLLOCK' / Product not catalogued VERS1 = '2155/EOHA129' / Version ID of each input file VERS2 = '2155/EOHI129' / Version ID of each input file VERS3 = '2318/APPH129' / Version ID of each input file