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

21.5. A WORKED EXAMPLE 279
determining both simultaneously is not yet possible, we are going to use an iterative process. In
most cases, very few iterations are necessary. This is the outline of the process:
• We experiment with the flat-field methods to define the best choice of parameters per
method.
• With that choice of parameters/method, we estimate the long-term transient component
and subtract it from the cube.
• In that new cube, we make a more refined correction of the flat-field.
In general, this is enough to get a significant improvement of the data quality. However, if
needed one can loop on the long-term transient, variable flat-field correction. In fact, as you
will see later (see sec. 21.5.3), the first step of this process is really a way to determine which
flat-field method to use in the second step.
For the rest of this section, I will assume that the LW3 and LW2 data are stored in CIA
raster structures (PDS) called lw3 and lw2. Remember that to work with SLICE, it needs to
be initialized (once) with the command:
CIA> @cia slice init
and that to transfer a raster structure to SLICE, youhavetotype
CIA> raster2slice,lw3
and respectively lw2 for the LW2 data structure. SLICE cannot handle more than one
structure so you will need to perform all the processing raster per raster, although for simplicity’s
sake, we are describing them in parallel in the following sections.
Once all the processing is done, transfer your data back from SLICE to CIA with:
CIA> new lw3 = raster2slice()
if you were working on the LW3 data, modify accordingly for the LW2 data.
Important warning: It may happen that even though you have corrected for short-term
transient, there is still a transient artifact at the start of the raster. In that case, it is better
to completely mask the stabilization frames at the start of the raster. If this is not done, then
the flat-field and long-term transient determination may fail (mostly because the sliding means
cannot work on the very first and last frames by construction). This is what happens here for
the LW2 case and thus we have masked the first 14 frames of the LW2 raster (see Table 21.2).
21.5.2 Choice of flat-field methods/parameters
For this first flat-field determination, we are working with data that are potentially still heavily
affected by a long-term transient, we must thus use a method which is not too affected by that.
SLICE has 6 different methods to build the flat-field, the first three of which are common with
CIA. The first one computes a single flat for all the data, and is equivalent to the “auto” method
of CIA. If you need to apply it, you can do it with:
CIA> red param=set red param(tdt=’65801627’)
The second one corresponds to the “calg” method of CIA. However, it is worth stating how
it is used since this is not described in SLICE’s original manual. Simply set:
CIA> red param=set red param(tdt=’65801627’,/library flat)