Controls the determination of multiple background signals PCA.COMFILL
"pca.pcathresh" ). Such missing values, whether caused by rejection of bad data or masking of
sources, are replaced by artificial data values representative of the surrounding data prior to the
calculation of the PCA model. If the pca.comfill parameter is false (i.e. zero) then such gaps in the data
are filled using linear interpolation between the neighbouring good values in the time-stream. if
pca.comfill is true (i.e. non-zero) then gaps are filled using the common-mode signal (i.e. the mean of
the remaining good values at each time-slice), together with corresponding bolometer gains. This
can in some case produce a more reliable PCA model. However, using this option is only
a good idea if there is a strong common mode present in the bolometer data. This will
usually only be the case if PCA is the first model specified in parameter "modelorder"
.
If pca.comfill is non-zero, gaps are filled using a COM and GAI model determined using a set of
parameters equivalent to those by the " normal" COM and GAI models, except that the name of each
parameter is changed from " com.xxx" to " pca.xxx" . Thus " pca.perarray" is used in place of the
normal parameter "com.perarray" . [0]