NAME

hstpolima -- Reduce linear polarization data for HST instruments
USAGE hstpolima imalis errlis instpoltab xybin binrej errlim outroot

PARAMETERS

imalis = "" [list of file names]
The list of input sky-subtracted images to be reduced. Upto 64 images can be input. They must all be of the same dimensions and are assumed to be well aligned.
errlis = "" [list of file names]
The list of input error files corresponding to the sky-subtracted images. They must be of the same dimensions and the same as the input sky-subtracted images.
instpoltab = "" [table file name]
Name of STSDAS table file giving the parameters for polarization determination for the instrument corresponding to the input images. The table lists the polarizer filter name, the polarizer angle, the transmission for the components parallel and perpendicular to the polarizer axis, the transmission of the s and p rays (parallel and perpendicular to the pick-off mirror surface for the WFPC2 case) and the retardation of the s ray relative to the p ray, the instrumental polarization and its instrumental position angle.
xybin = "" [integer]
Binning factor in x and y direction (same for both axes) for output images.
binrej = "" [real]
Factor times rms on mean for binned data above which to reject values for inclusion in the mean value for that bin.
errlim = "" [real]
The upper limit on the polarization error (%) for data to be included in the output files. Pixels in output images with polarization error greater than errlim have the values set to zero in all the output images.
outroot = "" [name]
The root name for all the output files. Ten output files are produced distinguished by differing postfixes:
outroot + int - Stokes I parameter (total signal)

outroot + inter - error on Stokes I parameter 

outroot + q - Stokes Q parameter

outroot + qer - error on Stokes Q parameter 

outroot + u - Stokes U parameter

outroot + uer - error on Stokes U parameter 

outroot + pol - linear polarization (%)

outroot + poler - error on linear polarization (%)

outroot + pa - position angle of linear polarization vector 
(i.e. E-vector) in degrees

outroot + paer - error on position angle of linear polarization 
vector (degrees)

DESCRIPTION

This task is a comprehensive tool for producing reduced polarization images for the HST instruments with imaging polarizer facilities (currently FOC, WFPC2 and NICMOS). A non-HST instrument can also be handled - called 'SPECIAL'. The error frames are required to produce output error files and should contain all sources of error (photon noise, systematics (e.g. flat fielding errors) and read-out noise).
All instrument specific parameters are picked up from the header and the telescope rotation angle is read from the PA_V3 or ORIENTAT descriptor (see below for details and also examples). The polarizer and or colour filter name in the header is matched with values found in the instpoltab table file and these are used to set the instrument specific polarization parameters required to compute the Stokes parameter and linear polarization data. The correction for polarizer transmission and pick-off mirror induced polarization follows closely the treatment for WFPC2 by Biretta and McMaster (WFPC2 ISR 97-11). The instrumental polarization is included for effects that are not collected by these physically based parameters. If the values of the polarizer efficiencies for calculating each Stokes parameter are equal for all input images (such as by specifying the same polarizer orientation and instrument position angle), then no calculation of polarization is possible. The task exits with an error in this case.
All errors are computed by Gaussian error propagation. To correct for the bias in linear polarization values which occurs when the Stokes parameter errors are large and the determination of linear polarization is no longer Gaussian error dominated, the method advocated by Wardle & Kronberg (ApJ, 194, 249, 1974) is employed, using the solution of the Rice distribution (see also Simmons & Stewart, A&A, 142, 10, 1985 for a discussion). Tests with synthetic data show that this correction does an adequate job to pol/sigma(pol) of about 2. Below that value it is best to use Monte Carlo methods which are not directly dealt with here. The method of computing the errors on the polarization position angle depends on the polarization signal-to-noise (i.e. pol/sigma(pol)). If pol/sigma(pol) is greater than 8 then the simple method (sigma(theta) = 28.65*sigma(pol)/pol) is employed (Serkowski, Adv. in Astron. & Astrophys. 1, 289, 1962); if pol/sigma(pol) is less than 8 then the distribution function of the position angle errors is integrated to derive the 1 sigma error (see Naghizadeh-Khouei & Clarke, A&A, 274, 968, 1993).
The output images can be rebinned in the two axis directions by an integer number. This permits the error per bin to be reduced or the sampling to be matched to the resolution of the polarization maps. Individual pixel values in the input images can be rejected from inclusion in binned data by setting the sigma rejection parameter to a low value. Some care should be exercised here else valid points will be deleted from inclusion in the sum of a binned region leading to a bias in Stokes parameter determination. Largish values are useful for rejecting obvious artefacts such as hot pixels or cosmic rays. For unbinned data this parameter is ignored.
The bound to the maximum polarization error (%) in the output maps acts to supress points with low signal-to-noise which will produce wildly varying polarization values and position angles. Output points with polarization error greater than this bound have pixel values set to zero in all the output images.
On completion of the task the number of output pixels with small (less than 1.0E-10) or negative values which were set to zero are printed to the terminal. The number of output pixels whose polarization error exceeded the maximum polarization error bound is also printed to the terminal.
The following header keywords are required in order that the instrumental polarization parameters can be matched against the instrument table files:
for FOC - INSTRUME, FILTNAM1, FILTNAM2, ORIENTAT
for NICMOS - INSTRUME, FILTER, ORIENTAT
for WFPC2 - INSTRUME, FILTNAM1, FILTNAM2, DETECTOR, PA_V3
for SPECIAL - INSTRUME, FILTER, POLANG, ORIENTAT

EXAMPLES

1. To reduce a set of FOC polarimetry images:

cl> hstpolima imalis="@focim.lis" errlis="@focer.lis" instpoltab="foc-pol-filt-001.tab" xybin=1 binrej=10. errlim=2.0 outroot="foc001"

FOC polarization data usually consist of a set of three images taken at the three rotation angles of the polarizer of 0, 60 and 120 degrees at approximately the same orientation (ORIENTAT). See the HST Data Handbook, Section 8-7).

2. To reduce a set of NICMOS polarimetry images:

cl> hstpolima imalis="@nicim.lis" errlis="@nicer.lis" instpoltab="nicmos-pol-filt-001.tab" xybin=2 binrej=5. errlim=3.0 outroot="nic001"

NICMOS polarization data usually consist of a set of three images taken at the three rotation angles of the polarizer which are close to 0, 120 and 240 degrees (depending on the polarizer used short (S) or long wavelength (L)) at a similar ORIENTAT. See the HST Data Handbook Section 18-12 and Hines, D. C., 1998, NICMOS and the VLT Workshop proceedings, eds. Freduling, W. & Hook, R., p. 63.

3. To reduce a set of WFPC2 polarimetry images:

cl> hstpolima imalis="@wfpcim.lis" errlis="@wfpcer.lis" instpoltab="wfpc2-pol-filt-001.tab" xybin=2 binrej=20. errlim=5.0 outroot="wfpc001"

WFPC2 polarization data can consist of three or four images taken with three or four rotation angles of the polarizer quad filter at a given PA_V3, or three or more images taken on the same chip at differing values of PA_V3 and covering a range of rotation of at least 90 degrees. See the HST Data Handbook Section 28-16 and Biretta, J. & McMaster, M., WFPC2 Instrument Science Report 97-11. The header parameters FILTNAM1 and FILTNAM2 are translated to provide the polarizer position angle in conjunction with the chip number (for POLQ only).

4. To reduce a set of polarimetry images originating from a non-HST instrument:

cl> hstpolima imalis="@specim.lis" errlis="@specer.lis" instpoltab="special-pol-001.tab" xybin=1 binrej=5. errlim=0.5 outroot="spec001"

If the parameter INSTRUME is set to SPECIAL then non-HST imaging polarimetry data can be reduced. The filter name given by the header parameter FILTER must match the name in the instrument table file. The polarizer angle is indicated by the header parameter POLANG. The instrument rotation is indicated by the ORIENTAT parameter which may need to be translated into the instrumental system.

Format of the Instrument specific table file

The STSDAS table file is a nine column file with the following columns: Name of colour/polarizing filter Position angle of polarizer filter Normalised transmission parallel to polarizer axis Normalised transmission perpendicular to polarizer axis Reflectance of mirror E-vector parallel to surface (s wave) Reflectance of mirror E-vector normal to surface (p wave) Retardance (deg.) of s wave relative to p wave Instrumental polarization (%) Position angle (deg.) of instrumental polarization (in instrument frame)

with the following column names and format: FILTER CH*16 POLANG R F6.2 TRANSPAR R E10.4 TRANSPER R E10.4 REFLECTRS R F6.4 REFLECTRP R F6.4 RETARDPHI R F8.4 INSTPOL R F6.2 INSTPA R F6.2

Thus table files for different instruments can be produced for use with this routine.

TIMING

The computing time requirements depend on the data values. Data with uniformly high signal-to-noise runs through quickly. However if the values of pol/sigma(pol) are low, then the computing time is substantially increased, since the Rice distribution must be solved for the bias on the polarization values and, more specifically, the distribution on the position angle errors must be numerically integrated.

BUGS

Some care should be taken if ORIENTAT is much different between the input images and there is substantial instrumental polarization, since the position angle of instrumental polarization is (probably) fixed in the reference frame of the instrument. This situation is not correctly handled.