[Top] [Prev] [Next] [Bottom]

32.13 Polarimetry

In FOS polarimetry mode the incoming beam was split into orthogonal polarization components by a Wollaston prism behind the entrance aperture. Although SINGLE apertures were always used, the two dispersed beams ultimately illuminated different portions of the photocathode located above and below the position illuminated by the standard unpolarized SINGLE aperture beam. The two photocathode positions accordingly required separate flatfields, wavelength calibrations, and Y-base deflections. The two spectra were alternately deflected onto the diode array and recorded in 10-second intervals throughout an exposure. Polarimetry exposure sequences (POLSCANs) consisted of 4, 8, or 16 individual exposures at unique polarizer rotations. These sequences had to fit within one orbital visibility. In the limit of minimal jitter or GIM motion, the same portions of the photocathode were nominally illuminated and recorded in each exposure.

As with non-polarimetric observations, no concurrent WAVECAL exposure was routinely taken with polarimetry data, so the limiting accuracies of the default pipeline polarimetry wavelength calibration are the same as for the non-polarimetry case described earlier (See "Wavelength Calibration" on page 32-49.). Different dispersion relations are applied to the two pass directions. When the beams are combined, the lower pass direction is simply shifted onto the wavelength grid of the upper pass direction with no interpolation or resampling. The error in this arbitrary shift can be determined by simply comparing the output wavelengths for the two pass directions on a pixel-by-pixel basis.

All FOS polarimetry calibration observations were taken with POLSCAN=16.

Polarimetry flatfields were always produced via the often subjective, non-superflat, continuum-fitting technique. The same comments made earlier in the flatfield section (See "Flatfield Calibration" on page 32-38.) about this method of flatfielding apply here, as well, but are of much less importance as polarimetry data are often binned heavily in analysis.

Polarimetry flux calibration is performed with the FLX_CORR method and the WD absolute flux system. This has little effect on the derived values of Q and U as the influence of the sensitivity function divides out in the calculation of those quantities. Recall that no correction is made in the FLX_CORR calibration for the influence of telescope focus or time-dependent photocathode sensitivity variation. These factors can cause an additional variable flux calibration error of 0-15% for pre-COSTAR observations, but should not impact post-COSTAR calibration.


Since the pre-COSTAR PSF overfilled all apertures, spacecraft jitter could impact some exposures in a POLSCAN sequence more than others, thereby introducing photometric effects that limited polarization measurement accuracy. For similar reasons, prior to the implementation of the onboard GIM correction, FOS polarimetry was not feasible.

Following implementation of the onboard GIM correction, pre-COSTAR polarimetry accuracies were limited by the effects of residual GIM motion, FGW positioning, and jitter on the fraction of the s-curve of the large PSF that was recorded by the diode array. Visual inspection of pre-COSTAR calibrated polarimetry observations indicates that variations in these quantities produced scatter in total polarization of the order of 0.5% and occasionally was somewhat worse. The uncertainty in the retardation calibration also contributed a systematic instrumental polarization equal to 2% of the linear polarization (see FOS ISR 078). The impact of photon statistical uncertainties always must be considered, but it can be minimized by appropriate binning of the data.


The COSTAR mirrors added two low angle-of-incidence (7 degrees) reflections ahead of the FOS optics. These additional reflections acted to convert some of the linearly polarized component into the circular component and vice versa. A correction algorithm for the linear polarization has been devised and will be implemented in calfos in early 1998. Until that time no post-COSTAR polarimetry data are fully calibrated by STScI algorithms available for general release. In the meantime, if you have a need for recalibration of post-COSTAR polarimetry, please contact help@stsci.edu.

The wavelength-dependent post-COSTAR instrumental polarization in Q varies from 0-3% over the FOS/BL 1600-3300Å spectral range and COSTAR-induced U varies from 0-0.5% over the same range. Panel a) of Figure 32.29 shows the magnitude of the COSTAR-induced instrumental polarization in high S/N observations of unpolarized spectrophotometric standard BD+28D4211 that have been combined into 64 bins. The discontinuity in U around 2300 Å. is due to the break between the G190H and G270H gratings. Panel b) of Figure 32.29 shows a residual of ~0.08% in Q for the same binned spectrum after correction for the instrumental polarization with the algorithm to be included in calfos. Note that in the 1800-2100Å region where the waveplate retardation goes through a 180 degree rotation, the limiting residual is ~0.2%. For bright objects, the polarization angles are known to within about +/-5 degrees. Koratkar et al. (1998, Ap.J. submitted) find that the post-COSTAR combined effects of residual GIM and spacecraft jitter do not produce polarization greater than 1%. Additional sources of error in polarization are due to the photon statistics of the observation and the error in the retardation calibration (2% of the linear polarization-see FOS ISR 078). Post-COSTAR polarimetry observations made with only four polarizer rotations (POLSCAN=4) contain an additional 0.4% uncertainty in Q. Note that polarimetry data for even very bright calibration sources must be binned in post-calibration data reduction in order to reach the levels of precision stated here.

Typically, photon statistics will be the dominant source of uncertainty in your observations.

The forthcoming FOS ISR 150 will be the definitive discussion of the technical and calibration details of FOS polarimetry. This document, which will be announced on the FOS WWW page is planned for release in spring, 1998.

Figure 32.29: Unpolarized Standard Star BD+28D4211: a) Prior to Correction and b) After Correction for Post-COSTAR Instrumental Polarization

figure courtesy A. Koratkar, 1998, Astrophys. J. submitted.

[Top] [Prev] [Next] [Bottom]

Copyright © 1997, Association of Universities for Research in Astronomy. All rights reserved. Last updated: 01/14/98 14:55:10