Both ground-based and in-flight measurements contributed to the encapsulation of the instrument model in a set of calibration files (CAL-G) required for analysis of ISOCAM data. The CAM Calibration Explanatory Document (Altieri et al. 1998, [2]) describes the limited time available for calibration measurements during the mission and how this was used. The data here presented form part of the general CAL-G releases for all instruments that were made from time to time, usually coincident with OLP software releases. CAL-G files were used by the pipeline to produce archive material including, in particular, the highest-level products, which include measurements in physical units of images or point sources.

ISOCAM CAL-G filenames usually start with `CCGLW' or `CCGSW' and the appropriate detector except the few that have kept historical names designated long ago. The set of operational files is shown in Table 6.1 along with a list of obsolete files. The early design of the ISOCAM pipeline made room for files that ultimately were never implemented. Users may come across dummy versions which should be ignored.

Operational Calibration Files
CCGLWDEAD	CAM LW Dead Pixel Map
CCGLWDARK	CAM LW Dark Current Exposures
CCGLWDFLT	CAM LW Detector Flat-Field Library
CCGLWOFLT	CAM LW Optical Flat-Field Library
CCGLWDMOD	CAM LW Time dependent dark current model parameters
CCGLWSPEC	CAM LW Filter & CVF spectral data
CCGLWPSF	CAM LW Point Spread Function Library
CCGLWTRANS	CAM LW Fouks-Schubert transient model coeffients
CCGLWRESET	CAM LW RESET value statistics
CCGLWSAT	CAM LW Saturation thresholds
CCGLWLOSS	CAM LW Sensitivity loss parameters
CLWCVF1	CAM LW CVF1 description
CLWCVF2	CAM LW CVF2 description
CCGSWDEAD	CAM SW Dead Pixel Map
CCGSWDARK	CAM SW Dark Current Exposures
CCGSWDFLT	CAM SW Detector Flat-Field Library
CCGSWOFLT	CAM SW Optical Flat-Field Library
CCGSWSPEC	CAM SW Filter & CVF spectral data
CCGSWPSF	CAM SW Point Spread Function Library
CCGSWSAT	CAM SW Saturation thresholds
CSWCVF	CAM SW CVF description
CSCGCROSS	CAM SW Cross-talk correlation matrices
CHCGCONV	CAM Housekeeping conversion factors
CWHEELS	CAM Wheel Information Table
Obsolete Calibration Files
CCGLWSLP	CAM LW CVF spectral line profile
CCGLWLINEAR	CAM LW linearity correction
CCGLWFRAME	CAM LW detector astrometric corrections
CCGLWGLITCH	CAM LW glitch model
CCGLWSTRAY	CAM LW straylight model
CCGSWSLP	CAM SW CVF spectral line profile
CCGSWLINEAR	CAM SW linearity correction
CCGSWFRAME	CAM SW detector astrometric corrections
CCGSWGLITCH	CAM SW glitch model
CCGSWSTRAY	CAM SW straylight model
CCGSWTRANS	CAM SW transient model

Files conform to the same simple FITS scheme used for all products, as described in Chapter 5, of either a single HDU=0 PRIMARY image (empty except for the dead-pixel maps) and associated header or a single HDU=1 BINARY-TABLE extension. Within the overall CAL-G release, as given by the HEADER[0] CALGVERS keyword in pipeline data products, ISOCAM maintained its own calibration version, known as CCD for CAM Calibration Data. CCGLWSPEC[0].CALIBRAT, for example, shows in which CCD release the file appeared. Extensive use is made of the standard FITS multidimensional array-coding mechanisms up to NAXIS=3 often within rows of TABLE[1] when many different calibration components are contained within a single file. For example, the CAL-G Version 7.0 release of CCGLWPSF[1].TABLE contained 121 different normalised point source images.

Most calibration components, such as dark images, flat-field corrections and point spread functions, depend on instrumental configuration as specified by accompanying TABLE[1] parameters of which usually a small subset are significant for selecting the correct component to use -- provision was made in the files to accommodate variations with temperature and voltage settings although these proved redundant and the corresponding columns are filled with zero values which are also to be ignored. Dark exposures, for example, are labelled with several parameters showing how they were derived, although the only significant one for selection purposes is the integration time, although scaling for the instrumental gain is also required. It was not possible to give all calibration components for all possible configurations, though straightforward procedures are available for how to choose the most suitable data in the absence of a perfect match as described below. Each of the calibration files is described in the following subsections showing the parameters on which the selection depends in the form `component=ccg_ component(p1,p2,...)' where the parameter order p1,p2,... is significant in the absence of a perfect match.

6.1.1 CCGLWDEAD & CCGSWDEAD - ISOCAM dead pixel maps

There are PRIMARY $32\times 32$ images for each detector showing LIVE=0 or DEAD=1 pixels. No new pixels died during the mission so that these files remained unchanged. The LW detector's column 24 was dead. The SW detector had 4 dead pixels.

6.1.2 CCGLWDARK & CCGSWDARK - ISOCAM dark current exposures

One of the most important calibration components is the signal detected in the absence of any illumination called the `dark current'. The files contain libraries of dark-current exposures (AXIS3=1) and errors (AXIS3=2) in units of ADU/G/s for GAIN=1 dependent on TINT only, accurate to about 0.3 and 0.5 ADU/G/s for the LW and SW detectors, respectively. Conversion to units of ADU/s for the detector GAIN is necessary using the correction factor ${\rm g}=2^{\rm GAIN}$ .

6.1.3 CCGLWDMOD - ISOCAM LW dark current model

Analysis of the evolution of the LW dark current (Section 4.2.2) showed there were systematic drifts both over the course of the mission and during individual revolutions (Biviano et al. 1998a , [5]; 2000, [9]; Román & Ott 1999, [50]). The drift is linear over the mission and quadratic over the revolution. For revolution number

, and time since activation in seconds

(since the instrument was switched on at the beginning of the revolution), the dark current in pixel

is given by:

The coefficients $ir_{yz}, sr_{yz}, io_{yz}, so_{yz}, po_{yz}$ are given in CCGLWDMOD for the 5 values of TINT at GAIN=0 rather than at GAIN=1, used in CCGLWDARK. These models give the best knowledge of the dark current and should be used in preference to the static data of CCGLWDARK.

6.1.4 CCGLWDFLT & CCGSWDFLT - ISOCAM detector flat-field library

For practical reasons, the overall flat-field correction was decomposed into the product of detector and optical flat-fields (see Section 4.5). Both flat-field components were estimated by exploiting the uniform illumination of the detector provided by the zodiacal light except for those optical configurations, including the SW channel in particular, where it was not bright enough. In such cases the Internal Calibration Device (ICD) was used instead despite the presence of significant residual spatial structure. For details of these procedures, see the ISOCAM Flat-field Calibration Report (Biviano et al. 1998c, [7]).

Under the further assumption that the optical flat-field at the highest 1.5 $^{\prime \prime }$ resolution is identically unity, the detector flat-field was determined from these highest resolution observations. After deglitching, dark correction and stabilisation, exposures were scaled to a median of unity in the central $12\times12$ pixel area to generate the detector flat-fields available in CCGLWDFLT and CCGSWDFLT, which have an accuracy of better than 1% as discussed in the ISOCAM Calibration Error Budget Report (Biviano et al. 1998d, [8]). The DFLT data are probably the CAL-G files in which the optical configuration parameter space was least satisfactorily explored.

6.1.5 CCGLWOFLT & CCGSWOFLT - ISOCAM optical flat-field library

Uniform illuminated observations at resolutions other than 1.5 $^{\prime \prime }$ in conjunction with the detector flat-fields discussed above were used to derive the optical flat-fields reported in CCGLWOFLT and CCGSWOFLT, which include those identically unity at 1.5 $^{\prime \prime }$ .

6.1.6 CCGLWSPEC & CCGSWSPEC - ISOCAM filter and CVF spectral data

After dark and flat-field corrections have been performed in detector units, it is necessary to convert to physical units in order to draw physical conclusions. Observations of calibration standard stars whose fluxes were already known or which had reliable models were made in a variety of optical configurations to derive the information stored in CCGLWSPEC and CCGSWSPEC, where the photometric relationship between detector units (ADU/G/s) and astronomical flux units (Jy) is most clearly encapsulated in the SENSITIV parameter in units of (ADU/G/s) / Jy. Values of SENSITIV are also given for polarisation measurements, as distinguished from normal open aperture measurements by the appropriate values of EWHL, in a number of relevant filters.

Although a detailed report on the approach and methods of the ISOCAM photometric programme is available elsewhere (Blommaert 1998, [10]; Blommaert et al. 2000, [11]) it is worth describing here briefly how these important data were derived. After the normal glitch, dark and flat-field corrections, well-stabilised data of the calibration image were background subtracted and the strength of the source in detector units of ADU/G/s was derived using a circular aperture before the application of a further correction factor calculated from the relevant point spread function to compensate for flux loss outside the aperture. The corresponding flux density for the broad-band filters was integrated over the relevant transmission curve assuming a $\lambda F_{\lambda}(\lambda) = constant$ spectral shape while for the relatively narrow bandwidth CVF scans, no integration was performed and the known spectrum was used directly. The ratio of these two quantities gives the SENSITIV parameter. For details concerning the CVF spectral response function see the ISOCAM CVF Calibration Report (Biviano et al. 1998b, [6]) and the ISOCAM CVF Photometry Report (Blommaert et al. 2001a, [12]). The spectral transmission curve, $R(\lambda)= T(\lambda) \times Q(\lambda)$ , was also derived from ground-based measurements, where $T(\lambda)$ is the filter transmission and $Q(\lambda)$ is the detector quantum efficiency (see Appendix A). Comparison of observed flux and model spectrum requires application of Equation A.

6.1.7 CCGLWSAT & CCGSWSAT - ISOCAM saturation thresholds

Very bright sources occasionally exceeded the dynamic range of the ISOCAM detectors, saturating one or more pixels and rendering physical flux estimates highly unreliable. CCGLWSAT and CCGSWSAT report the GAIN- and PROC-dependent saturation thresholds against which EOI pixel values in individual exposures should always be checked, as they are as a matter of course in AAC.

6.1.8 CCGLWRESET - ISOCAM LW detector RESET value statistics

In order to be able to check for pixel saturation those on-board processing modes for which only (EOI

RESET) values are available in telemetry rather than individual EOI and RESET values, it is useful to be able to interrogate the RESET statistics tabulated in CCGLWRESET against revolution number for all the spacecraft orbits in which any CAM data were taken.

6.1.9 CCGLWPSF & CCGSWPSF - ISOCAM point spread functions

CCGLWPSF and CCGSWPSF contain sets of normalised point source images that serve as point spread functions (PSFs) as measured at different optical configurations and detector locations specified in (y,z)=(TABLE[1].CRPIX1, TABLE[1].CRPIX2) (see Section 4.6.1 or Okumura 1998, [44]). The PSFs are intended for the usual purposes of source detection, as in AAC, and aperture photometry corrections.

6.1.10 CCGLWTRANS - ISOCAM LW FS transient model coefficients

CCGLWTRANS contains $32\times 32$ tables of $\lambda$ and $\beta$ coefficients (See Section 4.4.2 or Coulais & Abergel 2000, [21]) for each pixel used in the Fouks-Schubert transient correction model applied during AAC on LW data.

6.1.11 CCGLWPOL - ISOCAM polarisation weight factors

CCGLWPOL reports the orientation angles of the 3 polarisers in the entrance wheel.

6.1.12 CSCGCROSS - ISOCAM SW cross-talk correlation matrices

SW detector data require noise and capactive cross-talk corrections using coefficients held in CSCGCROSS in conjuction with the auxiliary SW columns 33 and 34. In the ISOCAM pipeline, these corrections are made during Derive_SPD, so that users who follow the recommendation to use SPD as the starting point for their analysis should make no use of this file. Otherwise, advice and code are available from the ISO Data Centre.

6.1.13 CHCGCONV - ISOCAM housekeeping conversion factors

CHCGCONV contains tabulations of physical units against coded digital reference values of 31 named housekeeping parameters for the interpolation of the values found in telemetry. Linear schemes defined by two points apply to 22 parameters; 2 parameters have 6-point schemes; 1 parameter has an 8-point scheme; and 6 parameters have 16-point schemes. In the ISOCAM pipeline, the conversion from digital to physical units is performed during Derive_SPD, so that users who follow the recommendation to use SPD as the starting point for their analysis should make no use of these data.

6.1.14 CWHEELS - ISOCAM wheel information table

ISOCAM's optical path is defined by the settings of 4 of 6 wheels. CWHEELS uses the WHLNB index to relate the motor step number used almost exclusively to label imaging data of the:

with descriptive names widely used among observers of the optical element in the path such as `HOLE', `LW10', etc. For the filter wheels, values of central wavelength and filter bandwidth are specified which may have become obsolete. Instead, it is recommended to use the rigorously maintained data in CCGLWSPEC and CCGSWSPEC.

6.1.15 CLWCVF1 & CLWCVF2 - ISOCAM LW CVF descriptions

CLWCVF1 and CLWCVF2 give wavelength in microns and sensitivity and error in (ADU/G/s) / Jy for the LW detector as a function of wheel position, known here as MOTOSTEP instead of the more common FCVF, of the two CVF segments of the LW filter wheel. These data among others have also been incorporated into CCGLWSPEC.

6.1.16 CSWCVF - ISOCAM SW CVF description

CSWCVF gives wavelength in microns and sensitivity and error in (ADU/G/s) / Jy for the SW detector as a function of the CVF wheel position, also known here as MOTOSTEP instead of the more common FCVF. These data among others have also been incorporated into CCGSWSPEC.

6.1 The ISOCAM Instrument Model