ISOCAM Interactive Analysis User's Manual Version 5.0 - ISO - ESA

276 CHAPTER 21. USING SLICE WITHIN CIA
21.4 Processing in SLICE
21.4.1 The SLICE syntax
Here again, SLICE differs from CIA. In SLICE, you do not manipulate directly the routines
that perform the actions, rather you describe with a set of structures the actions you want to
perform and then tell SLICE to do them. This philosophy is based on the necessity to perform
the data reduction steps in the correct order (and since there are many iterative processes
involved, this is very important) and on the desire to be able to “pipeline” the data processing
easily.
To describe the action (i.e. data reduction steps) you will perform, you have access to two
structures: the red param structure contains the parameters of the data reduction routines that
will be involved in the processing, and the act structure will contain the actions to be perform
(basically the act structure is a structure of boolean keywords, one for each data processing
step).
Once again, it is not our intention here to supersede SLICE’s manual. We highly recommend
that before you use SLICE, you read its user’s manual as well as M. A. Miville-Deschênes’ paper
to familiarize yourself with the concepts involved.
In this section we will only show an example of how to perform a perturbed flat-field correction
on your raster data to illustrate the main feature of SLICE’s syntax (sec. 21.5 provides a
more detailed example). The principle of this flat-field derivation is that to first order, the flatfield
is assumed to be constant over the whole raster and that temporal variations are treated as
small perturbations to this flat-field. It is also assumed that for a given readout, departures from
this single flat-field are dominated by high frequencies. To get them, a smoothed version of the
datacube is made which is subtracted to the original data. One then derives the perturbation
of the flat-field from a sliding mean on the modified cube.
The parameters for this flat-fielding method are:
flat smooth window
nplanes
flat thresh
The window size used to smooth each readout
the number of readouts used in the sliding mean
the percentage of data discarded from the flat field
computation, both in the top and bottom part of the
distribution
flat smooth window is generally a small number as it is applied to 32×32 images and the
window size is actually 2×flat smooth window+1, nplanes need not be larger than the number
of exposure per raster point as most of the signal has been removed, finally flat thresh larger
than 50% makes no sense.
Before starting the operation, one last information must be known, the TDT of your observation.
Although this is in principle not necessary when using SLICE in CIA, modifying this
would mean modifying SLICE, which we do not want to do. This information is quite easy to
get as it resides in your raster structure, in the field data.tdtosn. Note however that it is an
integer in the raster structure and that SLICE requires a string. Note also that for revolutions
smaller than 100, you should add the leading 0 to that string or SLICE will not find your data.
In our example, observation 45 taken on revolution 83, we would for instance have:
CIA> print,data.tdtosn
8301045
CIA> tdt = ‘08301045’