AIPS HELP file for POLCO in 31DEC18
As of Mon Jun 18 1:25:58 2018
POLCO: Task to correct polarization maps for Ricean bias
INNAME Input image name (name)
INCLASS Input image name (class)
INSEQ 0.0 9999.0 Input image name (seq. #)
INDISK Input image disk drive #
OUTNAME Output image name (name)
OUTCLASS Output image name (class)
OUTSEQ 0.0 9999.0 Output image name (seq. #)
OUTDISK Output image disk drive #
PIXSTD 0.0 9999.0 Sigma of noise on Q or U map
PCUT 0.0 Blank below PCUT sigma.
<= 0 => no blanking
OPCODE If 'ZERO', blank with zeros,
else use magic blanking
BADDISK Disks to avoid for scratch
Use: POLCO is a task to correct total polarized intensity (P) images
for the bias introduced when combining the Stokes Q and U images
which are assumed to have a noise distribution obeying Gaussian
statistics with a standard deviation sigma. The P image obeys the
Rice distribution, and at high signal-to-noise, Gaussian
statistics are approached (see Vinokur, Ann. d'Ap., 1965, 28,
412). One needs to estimate the intrinsic polarized intensity
(Pint) given an observed polarized intensity (Pobs). If Pint=0,
then the bias in the observed image is Pobs(Pint=0) = sqrt(pi/2) *
sigma. This task uses the maximum likelihood (ml) estimator of
Pint from Pobs (see Killeen, Bicknell, and Ekers, 1986, ApJ, 302,
306). The ml estimator does not provide an analytic solution.
Therefore, this task uses an asymptotic expression together with
an empirically determined correction. There is no significant
loss of accuracy with this procedure. The magnitude of the
correction is approximately Pint/Pobs ~ 0.8 at Pobs/sigma=2 and
Pint/Pobs ~ 0.9 at Pobs/sigma=2.5. Pint/Pobs asymptotes to unity
with increasing Pobs/sigma. The user must input sigma (Jy).
The parameter PCUT is the ratio Pobs/sigma below which the output
image will be blanked. This may be used to avoid the (mis)use of
low signal-to-noise polarization data. Furthermore, it is likely
that a more conservative value of PCUT (e.g. 3-sigma) may be more
desirable. On the other hand, blanking will give a false sense of
the noise on the image (underestimating by about a factor of 2).
An empirical method for handling lower values of Pobs/sigma,
attributed to Wardle and Kronberg, is used for Pobs/sigma < 2.
This method gives negatives for the lowest values and represents
the noise much more correctly. Pobs <= 0 (which should not occur)
are still blanked.
The task COMB with OPCODE 'POLC' performs the same operation with
a somewhat different algorithm for the correction.
INNAME......First image name (name). blank=>any
INCLASS.....First image name (class). blank=>any
INSEQ.......First image name (seq. #). 0=>any
INDISK......Disk drive # for the first image. 0=>any
OUTNAME.....Output image name (name)
OUTCLASS....Output image name (class)
OUTSEQ......Output image name (seq. #)
OUTDISK.....Output image disk drive #
PIXSTD......Input sigma of noise distribution on Q or U image (Jy)
PCUT........Blank below (Polarized flux/sigma)=PCUT
OPCODE......Type of blanking, if 'ZERO' blank with 0.0.
Anything else gives magic blanking.
BADDISK.....Disks to avoid for scratch