![]() |
Figure 3:
Temporal evolution after short term transient correction of the median
flux observed in the
![]() |
Long term transient.
Figure 3 presents the temporal evolution
of the median flux observed in the
central square of the detector for the GRB1 (a) and GRB2 (b) observations.
One striking feature of Fig. 3 is the drift in the first observation, not seen
in the second.
This drift in ISOCAM response is observed quite frequently and is called
in the following the Long Term Transient (LTT).
When observed, the amplitude of the LTT is about 5%
of the total sky emission, as in Fig. 3.
One also sees that the first sky image (Fig. 8A) is completely dominated by the
long term drift. This is true for any observation affected by a LTT
of faint extended emission with
contrasts smaller than a few percent of the zodiacal light.
The LTT was identified during the pre-launch measurements. Many calibration observations presented a slow and continuous increase of the response after a positive flux step. This rise gradually attenuates on a variable time scale which can be up to several hours. Unlike the short term transient, no analytical or physical description has been developed so far to correct for this specific effect which seems to be a general behavior of this type of detector. Vinokurov & Fouks (1991) addressed the problem of long term drift in their study of nonlinear response of photo-conductors. Unfortunately, the analytical approximation they propose does not seem to reproduce accurately ISOCAM's LTT.
Slow glitches. Another important problem is related to glitches which modify the response during a significant time. This is illustrated in Fig. 2 where one notices a significant change in the detector response near t=350 s, just after a glitch impact. The glitch impact is removed by the standard deglitching algorithm but the detector response is significantly disrupted for more than 500 s. In this example the signal is depressed after the glitch, but it can also be enhanced. These glitches with a memory effect (called in the following "slow glitches''), which affect the response of one or more pixels for some time after the impact, are believed to be due to heavy ions. They are responsible for most of the periodic patterns seen in Fig. 8A. No model has been developed to correct this instrumental effect. Desert et al. (1999) describe the various types of slow glitches known and a method to correct the response accident. However, it is not suitable for our problem; this method was indeed optimized for point source extraction where diffuse emission is not intended to be restored.
Ghosts. There are also several artificial point sources ("ghosts'' in the following) due to uncorrected memory effects. After seeing a bright point source, the response of a given pixel is significantly affected for some time. We have seen in the previous section that significant memory effects remain after strong point sources even after the short term transient correction. Therefore, as the satellite moves on the sky from one sky position to the other, pixels that have observed bright point sources are affected by a memory effect, and the pattern of each point source appears repetitively in the sky image until the response of the pixels returns to a normal value.
We conclude that the actual data processing does not adequately take into account (1) the long term drift, (2) slow glitches and (3) short term transients on point sources. The long term drift precludes the study of large-scale emission fainter than 10% of the zodiacal light background. All instrumental effects limit the brightness sensitivity for small scale structures. Solving these problems is crucial in order to take advantage of the unique sensitivity and angular resolution of ISOCAM. This is particularly true for small-scale structures since generally the contrast of interstellar medium emission decreases from large to small scales (Gautier 1992).
Copyright The European Southern Observatory (ESO)