CGLL
CJAM
CUFF
CUFF
CCIM
Once the SCD data structures have been established a series of instrumental procedures is applied to ensure the reliability and integrity of the imaging data. Each step is discussed below. Many of them have the effect of causing a bit to set in the SCD pixel mask that ensures exclusion of the datum from photometric calculations of any type of which the CMAP is the most important. More generally within SCDs, data are also excluded while the instrument is under adjustment as shown by non-zero values of the CISP[1].CISPDISC flags that show a configuration other than that commanded. Occasional SCDs have all their data excluded as reported in the CUFF:
No useful data in CSCD3610030101AX
No useful data in CSCD3610030101AY
Inspection of the raw images reported in the CCIM file gives some idea of the importance of these steps. There are 3-row RATE, RATE_ERR and EXPOSURE images in CCIM.TABLE[1] for every SCD of whatever type, accompanied by a full set of auxiliary data describing the circumstances in which the data were taken. The CCIM images are raw in the sense that they include all the SCD's data irrespective of imperfections of any type, like glitches, dead pixels, badly illumination or defects of any other type. For each CCIM 3-row image, the CUFF reports a short summary such as:
LW(OBS) 1996-11-11(14:20:43) to 1996-11-11(14:21:40) CSCD361003010107
RA(J2000)=20 21 48.007 : DEC(J2000)=+39 57 00.67 : ROLL(J2000)=249.20
14.0-15.8 micron @ 6" per pixel
CCIM(22:24) 27 planes
<RATE>=9.6ADU/s <RATE_ERR>=0.1ADU/s <EXPOSURE>=56.7s
CGLL 192 glitches
CJAM maximum jitter [-21.5:+0.1,-8.2:+0.2] pixels
including details of entries made in some of the products discussed below.
CGLL
It is clear from the most cursory glance at CCIM images that glitches are an unwelcome but common feature of many image readouts as discussed above in Section 4.3. The `Multi-resolution Median Transform' method (see Appendix E) is applied to detect and flag these events on the entire observation's history of each pixel. A few thousand glitches are regularly detected in observations and reported in the CGLL, whose main purpose is to provide a cross-check against any unusual features that might be discovered in the data. Besides the time, detector coordinates and strength of the glitches, CGLL[1].INDEX(*) refers to the CCIM images in which they were detected. For those glitches detected during regular observations of the sky, the corresponding pixel celestial coordinates are reported in CGLL[1].(RA(*),DEC(*)). Zero rather than proper NULL values fill these columns for glitches detected in operational modes other than OBS. The CGLL was also originally designed to accommodate glitch model parameters but no such modelling was ever performed and the COUNTG(*), YG(*), ZG(*) and UTKG(*) columns contain only nominal or null values.
CJAM
Detailed use is made of the IRPH and IIPH pointing files for prime data or the CRPH and CIPH for parallel data. First, the CUFF reports any pointing condition that applies to the observation as a whole, such as:
No up-to-date QSS/STR misalignment measurement: IRPH12900907
after which the RPH file is used to provide the reference celestial
coordinates for each SCD using a combination of RPID and time-key.
These data
are later used to define the reference coordinates attached to the
images
and thus determine ISOCAM's astrometry.
It is obviously important to take account of data taken only when the
instrument was looking in the right direction. A crude assessment is
provided by the OTF which shows that the pointing
was within a threshold distance, usually 10
, of the target position
and data are indeed excluded if the OTF is low. However, the OTF
threshold is
usually equivalent to several pixels and more detailed information is
available in the Pointing History files which are
in any case, the ultimate source of the SPD OTF.
The instantaneous pointing data reported in the IIPH or CIPH, including the
RPID and OTF, have a time resolution of 0.5 s that is finer than SPD and
thus enable more complete checks. Experience has shown that during a
raster the OTF sometimes briefly remains high
immediately after a change in the RPID has signalled a move to a
new position. If AAC detects a condition
of this type in an SCD the high OTF is overruled as reported in the CUFF:
Overruling the first high QLA(OTF) at RPID=[2,1]
Similarly, if a move has apparently started during a single ISOCAM
integration but the CAM OTF is still
high, it is also overruled with the report:
Overruling OTF when final RPID <> [3,1]
The mean instantaneous direction during each integration is used to
calculate the offset from
the current reference direction otherwise known as the pointing
jitter. The jitter could have
been used to realign the individual exposures contributing to an image but
because of some confusion early in the mission about the reliability of the
pointing data the corrections were disabled and never reinstated. The
jitter
components in pixel units parallel to the detector axes are reported in
CJAM[1].DY and CJAM[1].DZ for each integration. Inspection shows that the
satellite's pointing system was evidently very stable with the jitter
amplitude usually only a fraction of a pixel. These are the
most useful CJAM data. The CJAM[1] MEMORY, UNSTABLE and GLITCHED columns
are obsolete.
CUFF
For some optical configurations, the field mirror did not entirely cover the detector causing the appearance of prominent linear features roughly parallel to the edges of the detector, whose positions can vary. Badly illuminated rows and columns of this type are masked out and reported in the CUFF with messages such as:
780902864<UTK<781068887 Badly illuminated columns [1] [2]
CUFF
Observers should also be aware that very bright sources can saturate the detector, rendering the photometry dangerously unreliable. In this unfortunate event, prominent messages in the CUFF should draw the observer's attention as in these disjointed fragments which both show the extra CCIM information and anticipate the CMAP and CPSL discussion below:
+--------+---------+---------+--------+--------+---------+---------+--------+
CWRN Saturation problems detected
******** ********* ********* ******** ******** ********* ********* ********
Of the data collected during this observation 193 pixel readout values
were marked as saturated, so that flux measurements should be treated
with caution. Shown below are
- the saturated pixel count for those CCIM images affected;
- the label `saturated [LW] image' for those CMAP images affected;
- the label `saturated' for those CPSL point sources affected.
******** ********* ********* ******** ******** ********* ********* ********
+--------+---------+---------+--------+--------+---------+---------+--------+
LW(OBS) 1997-06-14(11:12:16) to 1997-06-14(11:14:14) CSCD576007010150
RA(J2000)=17 53 36.572 : DEC(J2000)=+56 51 56.89 : ROLL(J2000)=5.18
11.6-17.0 micron @ 6" per pixel
CCIM(154:156) 105 planes
<RATE>=-2.2ADU/s <RATE_ERR>=0.1ADU/s <EXPOSURE>=117.6s
CGLL 532 glitches
88 saturated pixels
CJAM maximum jitter [-0.1:+3.5,-0.1:+0.1] pixels
CMAP(142:144) 1997-06-14(11:12:16) to 1997-06-14(11:14:14)
RA(J2000)=17 53 36.572 : DEC(J2000)=+56 51 56.89 : ROLL(J2000)=5.18
11.6-17.0 micron saturated LW RPID=[7,6] image of [32*32] [6"] pixels
<FLUX>=0.53mJy/arcsec^2 <FLUX_ERR>=0.12mJy/arcsec^2 <EXPOSURE>=113.7s
CCIM(OBS:DRK:FLT)=(154:0:0)
CPSL(58) 1997-06-14(11:12:16) to 1997-06-14(11:14:14)
11.6-17.0 micron
ISOC175332.0+565219 17 53 32.043 +56 52 19.3 8413.2 +/- 42.1 mJy saturated
One of the significant improvements introduced towards the end of OLP software development was the ability to account for the transient behaviour of the LW detector following a change in the illumination as discussed in Section 4.4. It takes a long time for the detector to settle down after a large change in incident flux. In earlier versions a very rough check on the stabilisation was provided by the QLA flag which was raised after a pre-programmed fixed number of exposures had elapsed. Now, by application of the Fouks-Schubert model, individual pixel data are calculated according to the model and stored in SCDs, providing estimates of what those pixel data would have measured in the absence of such non-linearities. The CUFF reports a few details of the calculations.
FS:Using transient corrector Fouks-Shubert v2.0
FS:++ Depth data cube : 3310 planes
FS:++ Pixels corrected: 3389440
FS:-- Using Taylor : 3170187
FS:-- Using Mueller : 218229
FS:$$ Mueller failed : 0
These models provide, therefore, a much better stabilised alternative to
the raw pixel values and the pipeline makes use of them in calculating
all of its highest-level scientific data products.