8.2 General Considerations

• Auto-Analysis uses only time-independent flat-field corrections as available in the CAL-G libraries, leading to the possibility of less accurate fluxes compared with those calculated using optimised flat-field corrections as currently done within CIA. The likely outcome is poor flux calibration at the edges of the field of view.
• Auto-Analysis uses the Fouks-Schubert method to correct for transient effects. The method is applied only to LW data, excluding CAM parallel observations. The algorithm is necessarily applied automatically, and adopts as reference the quiescent level at the beginning of the observations when individual inspection could have identified some more appropriate value. In some extreme cases this can lead to completely wrong predictions of the corrected data. Note also that the current algorithm breaks down for point-like sources and is thus regularly disabled in small areas nearby, although in some circumstances contaminated fluxes or spurious hot spots have appeared.
• A changing CVF effective wavelength away from the optical centre leads to wavelength errors as big as the equivalent of one CVF step.
• Observations performed near the extremes of the CVF ranges, namely outside the range recommended in the ISOCAM Observer's Manual, are less accurately calibrated.
• Auto-Analysis makes no straylight or ghost corrections leading to the possibility of flux errors and the detection of spurious sources.
• Flux errors reported, for example, in the CPSL are only statistical and take no account of any systematic effects.
• Non-linearity of the detectors is not taken into account. This affects in particular the SW detector when a very bright point source was observed.