Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.stsci.edu/hst/acs/documents/isrs/isr0710.pdf Дата изменения: Mon Aug 20 21:38:21 2007 Дата индексирования: Sat Dec 22 23:55:30 2007 Кодировка: Поисковые слова: 47 tuc

Instrument Science Report ACS 2007-10

ACS Polarization Calibration ­ Data, Throughput, and Multidrizzle Weighting Schemes
M. Cracraft & W. B. Sparks August 20, 2007

A

BSTRACT

A subset of the polarized images from calibration proposals 9586, 9661, and 10055 were analyzed to help determine the polarization calibration accuracy level of the ACS camera. The polarization values found here are shown to be accurate to better than 1%. The absolute throughput values found while performing these calibration exercises did not match those currently listed in the synphot/ETC database, and we recommend these be updated in the system. Lastly, we examine the differences between two weighting schemes used in the multidrizzle software. The exposure time weighting, EXP, is found to more accurately preserve the flux levels than the error weighting scheme, ERR, for datasets with a small number of images being combined.

Introduction
The ACS camera has a versatile imaging polarization capability. Science programs using the polarizers have included observations ranging from Martian surface properties, to magnetic field structures of synchrotron jets, to observations of light echoes. The polarization levels in some of these objects can be a few percent or less, and a polarization calibration accuracy of ~ 1% is expected from the ACS polarizers (Pavlovsky et al 2006, ACS Data Handbook). This ISR looks at calibration data from three calibration proposals, discusses the throughput values of the polarizers, and describes the differences seen in output data when different weighting schemes are used in the multidrizzle software. The first section of this ISR describes how the polarization calibration data was processed and lists the results of photometry and the calculations performed on a set of

1


two polarized and two unpolarized calibration stars. A comparison of polarization degree and polarization angle is made to the published values obtained from ground observations for the polarized stars. This was done to test the methods described in the ACS Data Handbook and ISR ACS 2004-09. The second section discusses the throughput values of the ACS polarizers. Values found from on-orbit calibration data do not match those currently in the synphot/ETC database, and we recommend an update in order to enable the use of standard STScI calibration procedures. Section three is a comparison of what may happen when using two different weighting schemes in multidrizzle to combine a set of two flt images to create a weighted drizzled image. The error, ERR, method (recommended in the ACS Data Handbook) causes serious problems in polarization analysis (and presumably also photometry) when only a small number of images is used. Exposure time weighting, EXP, offers a robust and unbiased processing option, however.

I. Calibration Stars
The calibration data was derived from three different calibration proposals, 9586, 9661, and 10055. A variety of polarized and non-polarized sources were observed in these proposals. For this analyisis, the flat-fielded, distortion corrected, cosmic ray removed `drizzled' images ( `_drz.fits' files) were used as delivered from the MAST archive with standard OTFL processing. (The flats and distortion corrections appropriate for polarized data on the date of retrieval were applied by the pipeline.) The IRAF daophot task phot was used to perform aperture photometry on the specified stars. Count rates listed in this paper are from the drizzled images in units of electrons per second, and apertures of 10 pixels in radius were used for most of the data sets discussed here. A variety of annuli were used in the photometry software to find the sky estimates, based on whether there were any nearby stars in the images. The specific apertures and annuli used for each data set are detailed as they are discussed. Since the data for this ISR was gathered and analyzed over a period of several years, the date of retrieval from the archive will also be noted as each data set is mentioned. Using the equations from the ACS data handbook, which assume polarizers at 60 degree relative angles, Stokes parameter and polarization values were found for each image. In the equations below, r(POL) represents the flux in the image in the specified aperture, corrected for different transmissivities of the polarizers, as stated in ISR ACS 2004-09. This correction was applied to the POL60 and POL120 observations to obtain the flux relative to the POL0 observation before any further calculations were done. See Table 1 for these correction factors. (The March 2006 version 5.0 of the ACS Data Handbook lists a set of corrections in table 6.7 that can be applied to all three polarizers rather than only the POL60 and POL120 images. This table was also derived by J. Biretta [private communication] using the same data as in ISR ACS 2004-09. Since both

2


tables were derived from the same data, they give equivalent results.) The polarization parameters were calculated with the results of the Stokes parameters, (I, Q and U) as follows in equations 1, 2 and 3. P is the value of the polarization degree, equation 4, and is the polarization electric vector position angle. D is the polarization angle in detector coordinates measured counter-clockwise from the x axis, equation 5. " S is the conversion of the polarization angle into position angle on the sky measured counterclockwise from North, equation 6 . The values of z are the orientations of the POL0 ! polarizer projected onto the detector while contains information about the camera geometry and are listed in the ACS data handbook in section 6.3.4. !or the HRC, the F values are z = 20.4 and = -69.6. For the WFC, the values are z = 51.8 and = -38.15. The values of used here came from ISRs ACS 2004-09 and 10, only slightly different than the values listed in the ACS Data Handbook. ( z ­ 90 since D is zero on the xaxis and " S is zero on the y-axis.)
"2% I = $ '[ r( POL0) + r( POL60) + r( POL120)] # 3& ! " 2% Q = $ '[2 r( POL0) ( r( POL60) ( r( POL120) # 3& "2% U = $ '[ r( POL60) ( r( POL120) # 3&

Eq. 1 !

]

Eq. 2

]

Eq. 3 Eq. 4 Eq. 5

!

Q2 + U P= I

2

!

$U ' 1 " D = tan#1& ) + z 2 % Q( $U ' 1 " S = tan#1& ) + PA _ V 3 + * 2 % Q(

Eq. 6

!

Table 1 lists the correction factors applied to the POL60 and POL120 fluxes, taken from Table 17 of ISR ACS 2004-09. The correction factors in this table were found using data for GD319 in proposal 9586, except for the F775W/POLV filter combination for the WFC, marked with an asterisk. The values listed for this combination in Table 17 of the above ISR do not match the flux values listed in Table 16 of that same ISR which lists the data from which the correction factors were derived. The values listed in Table 1 for this combination were determined from flux ratios using proposal 9661 data instead, j8mj91011_drz.fits, j8mj91021_drz.fits and j9mj9031_drz.fits.

3


Table 1 : Flux ratios used as correction factors to normalize the fluxes of the observations. POL60/POL0 POL120/POL0 Detector Band flux ratio Flux ratio F475W/POLV 0.972 1.001 F606W/POLV 0.979 1.014 W FC F775W/POLV* F435W/POLUV HR C F475W/POLV F606W/POLV 0.967 1.161 1.093 1.088 0.985 1.109 1.127 1.103

F775W/POLV 1.075 1.043 * The WFC F775W /POLV correction factors do not match those in Table 17 of ISR A CS 2004-09. Th ere was a d iscr epancy b etw een th e flux es listed in th at ISR, and the correction factors listed in the table. Th e values in Table 1 her e come from Proposal 9661, as stated ab ove.

Proposal 10055 Using data from Proposal 10055, and the procedures described above, the polarization values were found for the polarized stars Vela (No. 81), a double star, and BD+64deg106. The Vela data was taken with the WFC/F606W/POLV filters and the BD+64deg106 data was taken with the HRC/F606W/POLV filters. The Vela data was observed at three different roll angles. For a roll angle PA_V3 of 348.1 degrees, observations were taken at five different locations in the image, one centered and one in each quadrant of the image. The pixel positions of the star Vela (no. 81) are shown in Table 2 for the drz images whose dimensions are 2139 x 2088 pixels. The data for BD+64deg106 also covered three different roll angles, though the target star was roughly centered in each image.
Table 2 : Positions and PA_V3 v alu es for the Vela data in Pr op 10055 Pol 0 image X position Y position PA _ V 3 J8 U Q 1 0 0 1 1 J8 U Q 1 0 0 8 1 J8 U Q 1 0 0 9 1 J8 U Q 1 0 0 G 1 J8 U Q 1 0 0 H 1 J8 U Q 2 0 0 1 1 J8 U Q 3 0 0 1 1 1036 1449 1420 378 652 1043 1069 1047 1428 632 660 1463 1070 1074 3 3 3 3 3 4 4 4 4 4 8 8 8 8 8 .0 .0 .0 .0 .0 8 9 9 8 8

48.11 108.03

The photometry was done on Vela using a 10 pixel radius aperture and a sky annulus centered on the star with 35 to 38 pixel radii, and the data was retrieved on 1/19/2006. The BD+64deg106 data was processed with a 50 pixel aperture radius and sky annulus from 50 to 55 pixel radii. (This was one of the earliest datasets analyzed, retrieved on 7/6/2005, and was not processed with the standard radius of 10 pixels as were the other datasets.) Tables 3 and 4 list the published values of the polarization degree and polarization angle of the polarized stars Vela I (No. 81) and BD+64deg106 as shown in Whittet, et al.

4


(1992) and Schmidt, Elston, and Lupie (1992) respectively. These published values will help us to determine the accuracy of our method.
Table 3 : V ela I No. 81 Ground Po larimetry (p ) Band p (%) U B V R I 5.0 6.1 6.86 6.85 6.29 0.6 0.3 0.13 0.19 0.10 PA 1 1 1 1 179 ( PA ) 3 1 3 1 1

Tab le 4: BD +64deg106 Ground Polar imetry (p ) Band P(%) U B V R I 5 5 5 5 4 .1 .5 .6 .1 .6 1 0 8 5 9 0 6 7 0 6 0 0 0 0 0 .1 .0 .0 .0 .0 0 9 3 9 5 4 0 7 8 2

PA 9 9 9 9 9 7 7 6 6 6 .0 .1 .6 .7 .8 4 5 3 4 9

( PA ) 0 0 0 0 0 .5 .4 .1 .5 .3 8 7 8 4 2

Tables 5 and 6 contain the fluxes and polarization values obtained for stars BD+64DEG106 and Vela (No. 81) using proposal 10055 data. In these tables, the first column contains the image name of the POL0 pipeline drizzled image. Each data set has an associated POL60 and POL120 image as well. The fluxes obtained from each image are listed in columns two through four. The derived percent polarization (Eq. 4) is in column five with the polarization angle (Eq. 6) in column six. The final column lists the PA_V3 value used in equation six to determine the polarization angle.
Table 5 : BD+64D EG106 data from proposal 10055 BD+64D EG106 - polar ized star HRC with F606W/POLV filters Literature P(%) = 5.687 PA = 96 .63 deg Image set pol 0 imag e J8 U Q 4 0 0 1 1 J8 U Q 4 0 0 8 1 J8 U Q 4 0 0 J1 J8 U Q 5 0 0 1 1 J8UQ600B1 J8 U Q 6 0 U P Q PO L 0 Electrons/sec 647274 645337 649304 704965 666275 663241 PO L 6 0 Electrons/sec 760131 760477 769996 718297 726582 720569 PO L 1 2 0 Electrons/sec 721304 719661 730473 713385 790494 791938 Mean Standard Deviation P (%) 4.8 5.1 5.3 5.2 4.8 5.4 5.1 0.25 PA (deg) 98.07 98.16 100.73 100.73 93.34 94.55 97.59 3.08 PA _ V 3 284.24 284.25 284.23 343.96 43.71 43.73

5


Table 6 : Vela (No. 81) data from proposal 10055 Vela calibration star (#81) - polarized star WFC w ith F606W/POLV filters Literature P( %) = 6.86 PA = 1 deg Image set pol 0 imag e J8 U Q 1 0 0 1 1 J8 U Q 1 0 0 8 1 J8 U Q 1 0 0 9 1 J8 U Q 1 0 0 G 1 J8 U Q 1 0 0 H 1 J8 U Q 2 0 0 1 1 J8 U Q 3 0 0 1 1 PO L 0 Electrons/sec 6 6 6 6 6 6 6 3 4 4 4 4 8 2 6 6 5 9 7 5 9 8 5 3 1 4 2 7 4 7 1 0 9 1 9 PO L 6 0 Electrons/sec 6 6 6 6 6 5 6 7 8 8 8 8 9 3 4 4 5 8 5 6 6 2 6 4 0 5 7 5 0 0 4 6 5 8 5 PO L 1 2 0 Electrons/sec 6 6 6 6 6 6 6 3 5 4 5 5 3 9 7 0 8 1 4 3 6 52 62 27 33 25 94 29 Mean P (%) 5.8 5.6 5.9 5.8 5.4 7.2 5.1 5.83 0.67 PA (deg) 6.35 7.73 7.38 7.0 8.88 4.33 179.5 6.02 2.81 PA _ V 3 348.08 348.09 348.09 348.08 348.08 48.11 108.03

Standard Deviation

As can be seen from Tables 5 and 6, the values of polarization degree and polarization angle found with ACS agree quite closely with the published values in tables 3 and 4, with some small potentially systematic variance of unknown origin. The rms of measured polarization degree for a variety of roll angles is " P = 0.25% (HRC) and " P = 0.67% (WFC). The rms of the derived position angle is "# = 3.08° (HRC) and "# = 2.81° (WFC).
! !

Proposals 9586 and 9661

! !

Data from proposals 9586 and 9661 were also analyzed in this same way to check more camera/filter combinations. Since the correction factors used in this paper were based on data from 9586, our analysis of this data was more a double-check, and allowed us to find the discrepancy with the F775W/POLV filter set in the WFC. The two polarized stars discussed in the previous section were examined, as well as two unpolarized stars, GD319 and G191B2B. Vela and GD319 are both double stars, but measurements were only taken of the brighter star for this analysis. All of the data from these two proposals used an aperture with a 10 pixel radius for the photometry. Various annuli were used in the photometry to calculate the sky values depending on the sizes and positions of the stars in the image. In proposal 9586, an inner sky annulus of 50 pixel radius with a width of 10 pixels was used for two data sets, j8gh04041 and j8gh040b1, while an inner sky annulus of 80 pixels with a width of 10 pixels was used for the rest of the data in this proposal. In proposal 9661, three datasets for star G191B2B, j8mj90pqq, j8mj91011, and j8mj950a1, used a 50 pixel radius inner annulus with a 10 pixel width. The rest of the data from this proposal used an 80 pixel radius inner annulus with a 10 pixel width. Two sets of images from these proposals, those for the HRC/F606W/POLV filters, were retrieved earlier than all the others. These data sets, j8mj40011, j8gh04011

6


and their associated POL60V and POL120V images, were retrieved on 9/15/2005. The rest of the proposal 9586 and the 9661 data discussed here were retrieved on 2/23/2006. Table 7 contains the fluxes and polarization parameters found for data from proposal 9586, while table 8 contains data from proposal 9661. Both of these contain the image name of the POL0 image, the filter set used, the fluxes for the star in each polarized image, and the derived polarization percent. For those stars which are polarized, the polarization angle (Eq. 6) is also shown. In the column titled P(%) ISR 04-09, the polarization found from fluxes listed in ACS ISR 2004-09 are shown. These fluxes were found with a 20 pixel radius aperture for HRC images and a 10 pixel radius aperture for WFC images as listed in the referenced ISR. Taking the fluxes listed, we applied the same correction factors and processed them through the same equations (Eq. 1-4) to provide a comparison to the values found with our fluxes. (ISR 04-09 did not list the polarization degrees or angles, only the countrates found.)
Table 7 : Proposal 9586 results GD 319 - unpolarized star Image set pol 0 imag e J8 G H 0 2 0 1 1 J8 G H 0 2 0 8 1 WF C PO L 0 Electrons/sec 91185 58654 PO L 6 0 Electrons/sec 88588 57286 PO L 1 2 0 Electrons/sec 91393 58172 P (%) P(%) I SR 0 4 -0 9 0.10 -

Filter / polset

F475W/POLV F775W/POLV HR C

0.10 0.60

GD 319 - unpolar ized star Image set pol 0 imag e J8GH040L1 J8 G H 0 4 0 4 1 J8GH040B1 J8 G H 0 4 0 1 1

Filter / polset

PO L 0 Electrons/sec 3 4 2 5 4 8 5 6 6 3 6 4 6 1 0 9 9 6 4 6

PO L 6 0 Electrons/sec 4 5 2 6 0 2 7 1 2 6 5 6 3 2 5 4 6 0 0 5

PO L 1 2 0 Electrons/sec 3 5 2 6 8 4 6 2 4 3 6 4 3 9 2 8 2 5 3 1

P (%)

F435W/POLUV F475W/POLV F775W/POLV F606W/POLV

0.0 0.20 0.20 0.20

P(%) I SR 0 4 -0 9 0.10 0.10 0.10 0.20

BD+64D EG106 - polar ized star Image set pol 0 imag e J8GH030L1 J8 G H 0 3 0 1 1 J8 G H 0 3 0 G 1 Filter / polset

H RC P( %) = 5.687 PA = 96.63 PO L 0 Electrons/ sec 121038 250897 386647 PO L 6 0 Electrons/ sec 152089 299094 441107 PO L 1 2 0 Electrons/ sec 140803 298242 419269 P (%) 4.6 5.0 3.5 P(%) I SR 0 4 -0 9 4.6 5.1 3.7 PA (deg)

F435W/POLUV F475W/POLV F775W/POLV

97.3 97.3 98.7

Two sets of Vela data from proposal 9586 with filters F475W and F775W were also analyzed, but the star was saturated (a-to-d saturation) in at least one of the three polarized images, so the results are not included here.

7


Table 8 : Proposal 9661 results GD 319 ­ unpolar ized star H RC Image set Filter / polset pol 0 imag e J8MJ41041 J8MJ40011 F435W/POLUV F606W/POLV

PO L 0 Electrons/sec 34948 56392

PO L 6 0 Electrons/sec 40528 61938

PO L 1 2 0 Electrons/sec 38498 62608

P (%)

0.40 0.60

P(%) I SR 0 4 -0 9 0.40 0.50

G191B2B - unpolarized star W FC Image set Filter / polset pol 0 imag e J8MJ91Q8Q J8MJ91011 F475W/POLV F775W/POLV

PO L 0 Electrons/sec 217730 86460

PO L 6 0 Electrons/sec 211219 84274

PO L 1 2 0 Electrons/sec 218481 85814

P(%)

0.30 0.30

P(%) I SR 0 4 -0 9 0.10 0.30

G191B2B - unpolarized star H RC Image set pol 0 imag e J8MJ950A1 J8MJ95061 J8MJ95071 Filter / polset PO L 0 Electrons/sec 86856 113586 37399 PO L 6 0 Electrons/sec 100716 124254 40242 PO L 1 2 0 Electrons/sec 96238 128286 38938 P(%) P(%) I SR 0 4 -0 9 0.10 0.30 0.20

F435W/POLUV F475W/POLV F775W/POLV

0.10 0.10 0.20

BD+64D EG106 - po larized star Image set pol 0 imag e J8MJ03041 Filter / polset

HRC P(%) = 5.687 PA = 96.63 PO L 0 Electrons/ sec 129987 PO L 6 0 Electrons/ sec 151349 PO L 1 2 0 Electrons/ sec 133342 P (%) P(%) I SR 0 4 -0 9 5.2 PA (deg)

F435W/POLUV

5.2

96.7

Vela calibration star # 81 - polar ized star WFC P(%) = 6.86 PA = 1 Image set pol 0 imag e J8MJ21NDQ J8MJ20011 Filter / polset PO L 0 Electrons/ sec 72877 18718 PO L 6 0 Electrons/ sec 77308 20331 PO L 1 2 0 Electrons/ sec 79057 20687 P (%) P(%) I SR 0 4 -0 9 5.9 6.9 PA (deg)

F775W/POLV F475W/POLV

6.2 6.9

5.0 0.85

In tables 7 and 8, the unpolarized stars show polarization degrees of less than one percent, and the polarized stars have polarization degrees near the published values listed in tables 3 and 4, though the polarization angle does vary within a few degrees, depending on the filter. Proposal 10378 Observations of the Boomerang Nebula, a highly polarized bi-polar reflection nebula, were also taken with the ACS/HRC camera, and the F606W/POLV filters for proposal 10378. The polarized images were combined using a script that creates images of each Stokes and polarization parameter based on methods described in Sparks & Axon

8


(1999 PASP). The same angles and correction factors are used in this method as described earlier. A great deal of structure can be seen in both I and P images of the nebula as shown in Figure 1. In the polarization image, the brighter regions indicate a higher level of polarization. The average levels of polarization for the northern and southern lobes of the nebula vary from each other; ~40% for the southern lobe and ~ 65 70% for the northern lobe, found in regions out to 11 and 7.5 arcseconds from the central scattering source respectively. These values are near those found by Taylor and Scarrott (1980 MNRAS), who listed the average polarization out to 30 arcseconds as 45% for the southern lobe and 60% for the northern lobe. Using these new images, a vector plot of this nebula was created that shows polarization angle vectors roughly at a tangent to a circle surrounding the central scattering source as expected in such a situation, see Figure 2. In order to determine to what extent the polarization angles differed from the expected tangent angles, an image made up of tangent angles to the central source was created and subtracted from the PA image. The resulting angle differences were plotted in the histogram shown in Figure 3. The black line is the histogram of empirical 's (i.e. the amount by which the electric vector differs from a 90° tangent vector), while the red line is a gaussian curve with sigma=3.5°. The empirical distribution is more skewed than the gaussian comparison, suggesting systematic uncertainties as well as random. It has a one-sigma width of 3.5° to 3.6° based on the location of percentiles of the distribution.
Figure 1: Stok es I (left) and Polarization (right) images of the Boomerang N ebula.

9


Figure 2 : V ector plo t of polarization angles of the Boomer ang Nebula.

Figure 3: Ang le d iffer ences between the d eriv ed polarization angle and the tangent angle.

From tables 5 and 6, we see that the rms of measured polarization degree for a variety of roll angles is " P = 0.25% (HRC) and " P = 0.67% (WFC). The rms of the derived position angle is "# = 3.08° (HRC) and "# = 2.81° (WFC). The Boomerang nebula analysis shows that the methods discussed here also work for extended objects and those with higher polarization, showing "# ~ 3.5° for the HRC. And though our ! ! derived polarization values are very close to the published values, an error estimate for ! !
!
10


the polarization percent cannot be determined from this test due to the large scale differences between the data sets. These tests show that the Advanced Camera for Surveys can be used to find polarization values for point sources accurate to better than 1% in polarization for a source that has an intrisic polarization of 6% and 3° position angle, confirming the uncertainties in the ACS Data Handbook.

II. Polarizer Throughputs
In the course of processing various polarimetric science data and the associated calibration data, it became apparent that the throughput curves being used by the synphot software for the polarizer throughputs to unpolarized light do not match the values found by examining an unpolarized star in both the polarized and unpolarized filters. We started by looking at calibration images of G191B2B, an unpolarized star, in proposal 9661. The specific set of observations examined were taken with the WFC/F606W/POLV filter set (j8mj90pfq_drz.fits, j8mj90piq_drz.fits, j8mj90plq_drz.fits and j8mj90pmq_drz.fits). The polarized fluxes in electrons per second, as listed in Table 18 of ISR ACS 2004-09, were divided by the unpolarized flux, and the average value obtained for the three polarizers was 0.386. In the current version of the synphot software, the value for this filter combination is listed as 0.243. (This is half of the sum of the Tpar and Tperp values, which are the throughputs for the parallel and perpendicular components of the polarizer. The values of Tpar and Tperp used here are from the POLV filter throughputs measured in laboratory tests by D. Leviton shown in Table 3 of the above referenced ISR and are the averages across all three polarizers. The POLUV filter transmission values are in Table 4 of the same ISR. ) In order to investigate this discrepancy, a set of calibration images from proposal 10055 of an outer region of 47 Tuc was also examined (j8uq80011_drz.fits ,j8uq80021_drz.fits , j8uq80031_drz.fits and j8uq80041_drz.fits). The photometry of each of the three polarized images was compared to the unpolarized image, and an average value for the throughput across the polarizers was found to be 0.388. Approximately 640 stars were used in this analysis. A set of V838 Monocerotis images from proposal 9694 were examined as well. Seventy-five stars in the field of view were examined, and the average ratio of the polarized to the unpolarized fluxes was 0.418, while the median value was 0.40. These values are slightly higher, possibly due to the fact that the nebulosity around V838 Mon is polarized. These stars might also be polarized since we are looking at low Galactic latitude and there is intervening ISM.
Tab le 9: Results of throughput an alysis Analysis method Absolute Proposal 9661 star G191B2 B 47 Tuc (640 stars) V838 Monocerotis (75 stars) Average 0.4 throughput value 0.386 0.388 18 Median 0.40

After testing the throughput for the WFC/F606W/POLV filter set with several data sets and better characterising the discrepancy between the on-orbit data and the laboratory tests, we decided to test other filter combinations. Table 18 in ISR ACS
11


2004-09 lists several data sets from proposal 9661 that have both polarized and unpolarized images. These data sets used a variety of camera and filter combinations, allowing the absolute throughput to be found for more than just one filter. The uncorrected flux of the target star from each polarized image was divided by the flux of the unpolarized image and averaged across the polarizers (POL0, POL60, and POL120) to derive the throughput. The values we obtained for each instrument from the proposal 9661 fluxes are listed in tables 10 and 11. Another way of looking at throughputs for a variety of filter combinations is to start with the correction factors used in the polarimetry calculations. The correction factors and throughput values are related as follows. Table 6.7 of the ACS Data Handbook gives correction factors (C) such that Stokes I is on the same absolute scale, same countrate, as an observation without polarizers. That is, Stokes I' should be on the same scale as an observation without polarizers, where Stokes I = C x (Stokes I), and Stokes I = 2 x (average rate with polarizers) = 2 x (rate without polarizers) x throughput using our definition of throughput T. But Stokes I is also the count rate without polarizers, hence it follows that T = 1/(2C). We include values of the equivalent throughput implied by Table 6.7 (DH), using the POL0 correction factors, in Tables 10 and 11. This shows that the empirical measurements differ from the synphot curve. There is a possibility that we have misinterpreted the Leviton lab results, which could explain the discrepancy between those values which are currently being used in synphot and the throughputs we find empirically. E.g. if the Leviton throughputs are in fact throughputs of two crossed polarizers oriented parallel or perpendicular to one another, then the throughput of a single polarizer to unpolarized light, neglecting the perpendicular component, would be 1 2 TLev = 1 2 2Tsyn = Tsyn 2 . This mapping comes closer to matching our results than the synphot values. Tables 10 and 11 display the results of the throughputs found using the various methods discussed above. ! he values we obtained for each instrument are listed in T column two as the on-orbit throughput. The values in column three are the throughputs implied by the correction factors in Table 6.7 in the ACS data handbook as described. The fourth column contains the current values being used in the synphot software as seen in tables acs_pol_v_004_syn.fits and acs_pol_uv_005_syn.fits in cracscomp. The values in the last column are a possible mapping of Tsyn if the Leviton ground data used to generate the synphot tables are misinterpreted.

!

12


Table 10 : PO L_V absolute po larizer Throughputs to unpolar ized ligh t Filter Throughput onThroughput acs_pol_v_004_syn.fits orbit H RC / WFC implied by Table synphot throughput 6.7 (DH,PO L0) HRC/W FC F4 7 5 W 0.3424 / 0.3511 0.32 / 0.35 .2636 F6 0 6 W F6 2 5 W F7 7 5 W 0.3728 / 0.3858 0.3678 / n/a 0.4779 / 0.4953 0.35 / 0.38 0.48 / n/a 0.46 / 0.50 .2430 .2465 .3263

Tsyn 2

!

0.36 0.35 0.35 0.40

Table 11 : PO L_UV absolu te Polar izer Throughputs to unpolarized ligh t Filter Throughput onThroughput imp lied acs_pol_uv_005_syn.fits orbit H RC by Table 6 .7 synphot throughput (DH,POL0) F2 2 0 W .0634 .0851 F2 5 0 W .1827 .1469 ! F3 3 0 W .3058 0.29 .2625 F4 3 5 W .3317 0.30 .2745 F8 1 4 W .5569 .4309

Tsyn 2
0 0 0 0 0 .2 .2 .3 .3 .4 0 7 6 7 6

For reference, we convert our derived throughputs into the correction factor C convention of the Data Handbook in Table 12. Tweaking the correction factors to compensate for the different throughputs might be a way of avoiding making changes to synphot. The correction factors in column two are those taken from Table 6.7 in the ACS Data Handbook for the POL0 filters. The last column uses the formula T = 1/(2C) used above to translate our throughputs from tables 10 and 11, column two, into correction factors (C).
Table 12: Correction factors for the varying through puts for POL0 Filter / polset Correction factor fro m Correction factor Table 6.7(DH,POL0 ) found from on-orbit HRC / WFC data (HRC/ WFC) F475 W / POLV 1.5651 / 1.4303 1.460 / 1.424 F606 W / POLV 1.4324 / 1.3314 1.341 / 1.296 F625 W / POLV 1.0443 / n/a 1.36 / n/a F775 W / POLV 1.0867 / 0.9965 1.046 / 1.010 F220 W / POLUV 7.882 / n/a F250 W / POLUV 2.737 / n/a F330 W / POLUV 1.7302 / n/a 1.635 / n/a F435 W / POLUV 1.6378 / n/a 1.507 / n/a F814 W / POLUV 0.898 / n/a

In synphot, only a single polarization curve is used for each of the three polarizers, and that does not properly give the throughput to unpolarized light. An initial improvement will be to update the curve according to the values in tables 10 and 11. A refinement for the future could be to install different curves for each polarizer for each
13


camera (12 curves in all). This would require modification to the ACS graph table and a more throrough analysis of the relative throughputs of the polarizers (see ACS Data Handbook 6.3.2). What is the cause for the discrepancies between the ground measurements and our empirical on-orbit throughputs? There could be many factors contributing to this. It has been suggested (private correspondence with J. Biretta), that the fluxes used could be wrong, the ground data could be wrong, or more likely, misinterpreted, as discussed earlier. Another possibilty for the differences is that the other optics such as the mirrors and detectors could be affecting the results, as the polarization throughputs depend in a complicated manner on the optics and the polarization properties of the input light such as the degree of polarization and the polarization angle. However, approaching this topic as we have done simplifies the problem, focusing only on the fluxes seen in polarized and unpolarized images. This analysis offers a robust, empirical measurement of the amount by which the light of an unpolarized star is attenuated when the polarizer is inserted in beam.

III. Exposure Time Versus Error Weighting
When using multidrizzle to process a set of images, the user is given the option of choosing a weighting scheme to use in creating the final weighted images. One of these options, EXP, simply uses the exposure time, thus giving equal weight to each component image being combined, assuming each image has the same exposure time. Another method, ERR, uses the actual error images associated with the individual images being combined. Though the multidrizzle help file states that using the error weighting option, ERR, is `generally recommended to be the most accurate type of weighting for producing the final drizzled image' for ACS and STIS images, it should not be used unless the user is combining many images to make up the final drizzled image. In trouble shooting an apparent discrepancy in ERR weighted science data, a set of simulated polarization observations were generated, described below, and used to compare the tw