Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.stsci.edu/~golim/papers/golimowski2006.pdf Äàòà èçìåíåíèÿ: Thu Jun 1 19:39:20 2006 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 18:32:17 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï

The Astronomical Journal, 131:3109 ­ 3130, 2006 June
# 2006. The American Astronomical Society. All rights reserved. Printed in U.S.A.

A

HUBBLE SPACE TELESCOPE ACS MULTIBAND CORONAGRAPHIC IMAGING OF THE DEBRIS DISK AROUND  PICTORIS1
D. A. Golimowski,2 D. R. Ardila,3 J. E. Krist,4 M. Clampin,5 H. C. Ford,2 G. D. Illingworth,6 F. Bartko, ´i N. Ben I´ 8 J. P. Blakeslee,9 R. J. Bouwens,6 L. D. Bradley,2 T. J. Broadhurst,10 R. A. Brown,11 tez, C. J. Burrows,12 E. S. Cheng,13 N. J. G. Cross,14 R. Demarco,2 P. D. Feldman,2 M. Franx,15 T. Goto,16 C. Gronwall,17 G. F. Hartig,11 B. P. Holden,6 N. L. Homeier,2 L. Infante,18 M. J. Jee,2 R. A. Kimble,5 M. P. Lesser,19 A. R. Martel,2 S. Mei,2 F. Menanteau,2 G. R. Meurer,2 G. K. Miley,15 V. Motta,18 M. Postman,11 P. Rosati,20 M. Sirianni,11 W. B. Sparks,11 H. D. Tran,21 Z. I. Tsvetanov,2 R. L. White,11 W. Zheng,2 and A. W. Zirm2
Received 2005 December 22; accepted 2006 February 13
7

ABSTRACT We present F435W (B), F606W ( broad V ), and F814W ( broad I ) coronagraphic images of the debris disk around  Pictoris obtained with the Hubble Space Telescope's Advanced Camera for Surveys. These images provide the most photometrically accurate and morphologically detailed views of the disk between 30 and 300 AU from the star ever recorded in scattered light. We confirm that the previously reported warp in the inner disk is a distinct secondary disk inclined by $5 from the main disk. The projected spine of the secondary disk coincides with the isophotal inflections, or ``butterfly asymmetry,'' previously seen at large distances from the star. We also confirm that the opposing extensions of the main disk have different position angles, but we find that this ``wing-tilt asymmetry'' is centered on the star rather than offset from it, as previously reported. The main disk's northeast extension is linear from 80 to 250 AU, but the southwest extension is distinctly bowed with an amplitude of $1AU over the same region. Both extensions of the secondary disk appear linear, but not collinear, from 80 to 150 AU. Within $120 AU of the star, the main disk is $50% thinner than previously reported. The surface brightness profiles along the spine of the main disk are fitted with four distinct radial power laws between 40 and 250 AU, while those of the secondary disk between 80 and 150 AU are fitted with single power laws. These discrepancies suggest that the two disks have different grain compositions or size distributions. The F606W/ F435W and F814W/ F435W flux ratios of the composite disk are nonuniform and asymmetric about both projected axes of the disk. The disk's northwest region appears 20% ­ 30% redder than its southeast region, which is inconsistent with the notion that forward scattering from the nearer northwest side of the disk should diminish with increasing wavelength. Within $120 AU, the mF435W þ mF606W and mF435W þ mF814W colors along the spine of the main disk are $10% and $20% redder, respectively, than those of  Pic. These colors increasingly redden beyond $120 AU, becoming 25% and 40% redder, respectively, than the star at 250 AU. These measurements overrule previous determinations that the disk is composed of neutrally scattering grains. The change in color gradient at $120 AU nearly coincides with the prominent inflection in the surface brightness profile at $115 AU and the expected waterice sublimation boundary. We compare the observed red colors within $120 AU with the simulated colors of nonicy grains having a radial number density / rþ3 and different compositions, porosities, and minimum grain sizes. The observed colors are consistent with those of compact or moderately porous grains of astronomical silicate and /or graphite with sizes k0.15­0.20 m, but the colors are inconsistent with the blue colors expected from grains with porosities k90%. The increasingly red colors beyond the ice sublimation zone may indicate the condensation of icy mantles on the refractory grains, or they may reflect an increasing minimum grain size caused by the cessation of cometary activity. Key words: circumstellar matter -- planetary systems: formation -- planetary systems: protoplanetary disks -- stars: individual (  Pictoris) Online material: tar file

1 Based on guaranteed observing time awarded by the National Aeronautics and Space Administration ( NASA) to the ACS Investigation Definition Team (HST program 9987). 2 Department of Physics and Astronomy, The Johns Hopkins University, 3400 North Charles Street, Baltimore, MD 21218-2686. 3 Spitzer Science Center, Infrared Processing and Analysis Center, MS 220-6, California Institute of Technology, Pasadena, CA 91125. 4 Jet Propulsion Laboratory, 4800 Oak Grove Drive, MS 183-900, Pasadena, CA 91109. 5 NASA Goddard Space Flight Center, Code 681, Greenbelt, MD 20771. 6 Lick Observatory, University of California at Santa Cruz, 1156 High Street, Santa Cruz, CA 95064. 7 Bartko Science and Technology, 14520 Akron Street, Brighton, CO 80602. 8 ´ Instituto de Astrof´sica de Andalucia, CSIC, Camino Bajo de Huetor, 24, Granada 18008, Spain. i ´ 9 Department of Physics and Astronomy, Washington State University, Pullman, WA 99164. 10 School of Physics and Astronomy, Tel Aviv University, Tel Aviv 69978, Israel. 11 Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218. 12 Metajiva, 12320 Scenic Drive, Edmonds, WA 98026. 13 Conceptual Analytics, LLC, 8209 Woburn Abbey Road, Glenn Dale, MD 20769.

3109


3110 1. INTRODUCTION

GOLIMOWSKI ET AL.

Vol. 131

Since the initial discoveries of cool ($100 K ) dust around nearby stars by the Infrared Astronomical Satellite (Aumann 1985),  Pictoris has been the foremost example of a young, main-sequence star with a resolved circumstellar disk of dust. The disk likely comprises the debris from disintegrating bodies in a nascent planetary system rather than primordial dust from the dissipating protostellar nebula ( Backman & Paresce 1993; Artymowicz 1997; Lagrange et al. 2000; Zuckerman 2001). Spectroscopic evidence of multitudinous star-grazing comets ( Lagrange-Henri et al. 1988; Beust et al. 1990; Vidal-Madjar et al. 1994 and references therein) has motivated models of the disk as an admixture of gas and dust from colliding and evaporating comets located within a few tens of AU from  Pic ( Lecavelier des ´ Etangs et al. 1996; Beust & Morbidelli 1996; Thebault et al. 2003). The cometary origin of the dust is supported by the detection of broad, 10 m silicate emission such as that observed in the spectra of comets Halley, Kohoutek, and others ( Telesco & Knacke 1991; Knacke et al. 1993; Aitken et al. 1993; Weinberger et al. 2003; Okamoto et al. 2004). Ground-based, coronagraphic images of  Pic reveal an asymmetric, flared disk extending at least 1800 AU from the star and viewed nearly edge-on (Smith & Terrile 1984; Paresce & Burrows 1987; Golimowski et al. 1993; Kalas & Jewitt 1995; Mouillet et al. 1997a; Larwood & Kalas 2001). High-resolution Hubble Space Telescope (HST ) and adaptive optics images show that the inner part of the disk ($20 ­ 100 AU from  Pic) is warped in a manner consistent with the presence of a secondary disk that is inclined by $5 from the main disk and perhaps sustained by a massive planet in a similarly inclined eccentric orbit ( Burrows et al. 1995; Mouillet et al. 1997b; Heap et al. 2000; Augereau et al. 2001). HST and ground-based images also reveal concentrations of dust along the northeast extension of the disk about 500 ­ 800 AU from the star that have been interpreted as an asymmetric system of rings formed, along with other asymmetries in the disk, after a close encounter with a passing star ( Kalas et al. 2000, 2001; Larwood & Kalas 2001). Spatially resolved mid-infrared images show an asymmetric inner disk having depleted dust within 40 AU of  Pic ( Lagage & Pantin 1994; Pantin et al. 1997 ) and oblique clumps of emission 20 ­ 80 AU from the star ( Wahhaj et al. 2003; Weinberger et al. 2003; Telesco et al. 2005). These features suggest the presence of noncoplanar dust rings whose locations conform to the mean-motion resonances of a putative planetary system. Constraints on the sizes of the dust grains observed in scattered light have been based on multiband (BVRI ) imaging studies of the disk in both unpolarized ( Paresce & Burrows 1987; Lecavelier des Etangs et al. 1993) and polarized (Gledhill et al. 1991; Wolstencroft et al. 1995) light. The unpolarized images indicate that the disk is colorless (within uncertainties of 20% ­ 30%) at distances 100 ­ 300 AU from  Pic, although its B-band
Royal Observatory Edinburgh, Blackford Hill, Edinburgh EH9 3HJ, UK. Leiden Observatory, Postbus 9513, 2300 RA Leiden, Netherlands. 16 Institute of Space and Astronautical Science, Japan Aerospace Exploration Agency, 3-1-1 Yoshinodai, Sagamihara, Kanagawa 229-8510, Japan. 17 Department of Astronomy and Astrophysics, Pennsylvania State University, 525 Davey Lab, University Park, PA 16802. 18 Departamento de Astronom´a y Astrof ´sica, Pontificia Universidad Cai i ´ tolica de Chile, Casilla 306, Santiago 22, Chile. 19 Steward Observatory, University of Arizona, Tucson, AZ 85721. 20 European Southern Observatory, Karl-Schwarzschild-Strasse 2, D-85748 Garching, Germany. 21 W. M. Keck Observatory, 65-1120 Mamalahoa Highway, Kamuela, HI 96743.
15 14

brightness may be suppressed at 50 AU from the star.22 This neutral scattering by the disk has been customarily viewed as evidence that the dust grains are much larger than the wavelengths of the scattered light (31 m). However, Chini et al. (1991) noted that the B-, V-, and I-band scattering efficiencies of silicate spheres were similar for grains with radii of 0.2 ­ 0.3 m. Attempts to reconcile the neutral colors of the disk with the 10% ­ 25% polarization of scattered light from the disk have been problematic. Voshchinnikov & Krugel (1999) and Krivova et al. (2000) ¨ found that although the polarization alone is best fitted with a grain size distribution with a lower limit of a few microns, the observed neutral colors can only be replicated by adding submicron-sized grains and lowering either the refractive index of the grains or the proportion of the smallest grains. Given the adjustments needed to match the polarization models with the highly uncertain disk colors, a more precise multicolor imaging study of the  Pic disk is warranted. In this paper we present multiband coronagraphic images of  Pic's circumstellar disk obtained with HST 's Advanced Camera for Surveys (ACS; Ford et al. 2003; Gonzaga et al. 2005). These images reveal the disk between 30 and 300 AU from the star with unprecedented spatial resolution, scattered-light suppression, and photometric precision. These qualities permit the measurement of the disk's optical colors with 3 ­ 10 times better precision than previously reported from ground-based observations. By deconvolving the instrumental point-spread function ( PSF ) from each image, we accurately determine the brightnesses, morphologies, and asymmetries of the two disk components associated with the warp in the inner disk. Our fully processed images and results will likely serve as the empirical standards for subsequent scatteredlight models of the inner disk until the next generation of spacebased coronagraphic imagers is deployed. 2. OBSERVATIONS AND DATA PROCESSING 2.1. ACS Imaging Strategy and Reduction Multiband coronagraphic images of the A5 V star  Pic were recorded on UT 2003 October 1 using the High Resolution Channel ( HRC) of ACS ( Ford et al. 2003; Gonzaga et al. 2005). The HRC features a 1024 ; 1024 pixel CCD detector whose pixels subtend an area of 0B028 ; 0B025, providing a $29 00 ; 26 00 field of view ( FOV ). Beta Pic was acquired in the standard ``peak-up'' mode with the coronagraph assembly deployed in the focal plane of the aberrated beam. The star was then positioned behind the small (0B9 radius) occulting spot located approximately at the center of the FOV. HST was oriented so that the disk's midplane appeared approximately perpendicular to the 500 occulting finger and the large (1B5 radius) occulting spot that also lie in the FOV. Short, medium, and long exposures were recorded through the F435W (B), F606W ( broad V ), and F814W ( broad I ) filters over three consecutive HST orbits. All images were digitized using the default analog-to-digital conversion of 2 eþ DNþ1. This sequence of exposures was promptly repeated after rolling HST about the line of sight by $10 . This offset changed the orientation of the disk in the FOV by $10 and facilitated the discrimination of features associated with the disk from those intrinsic to the coronagraphic PSF. Immediately before the exposures of  Pic, coronagraphic images of the A7 IV star
22 Throughout this paper we compute the projected dimensions of the disk in AU using the trigonometric parallax of ¼ 0B05187 ô 0B00051 (or 1 / ¼ 19:28 ô 0:19 pc) reported for  Pic by Crifo et al. (1997 ) based on astrometric measurements conducted with the Hipparcos satellite. Consequently, the projected distances reported in this paper may differ from those appearing in papers published before 1997, which were based on an erroneous distance of 16.4 pc to the star.


No. 6, 2006

MULTIBAND IMAGES OF DISK AROUND  PIC
TABLE 1 Log o f ACS / HRC Expo sure s Roll Anglea (deg) 87.2 Exposures (s) 1 2 2 1 2 2 1 2 2 1 2 6 1 2 3 1 2 3 1 2 6 1 2 3 1 2 3 ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; 40 100 225 15 50 200 40 100 200 10 100 338 10 250 653 10 250 645 10 100 338 10 250 653 10 250 645

3111

Star Pic ............

Orbit 1 1 1 1 1 1 1 1 1 2 2 2 3 3 3 4 4 4 5 5 5 6 6 6 7 7 7

Filter F435W F435W F435W F606W F606W F606W F814W F814W F814W F606W F606W F606W F435W F435W F435W F814W F814W F814W F606W F606W F606W F435W F435W F435W F814W F814W F814W

 Pic .............

110.0

 Pic .............

100.3

Note.-- Recorded over seven consecutive HST orbits on UT 2003 October 1. a Defined as the position angle of the y-axis of the raw HRC image, measured east of north.

Pic were recorded through the same filters to provide reference images of a star having colors similar to those of  Pic but no known circumstellar dust. A log of all HRC exposures is given in Table 1. The initial stages of image reduction (i.e., subtraction of bias and dark frames and division by a noncoronagraphic flat field) were performed by the ACS image calibration pipeline at the Space Telescope Science Institute (STScI; Pavlovsky et al. 2005). To correct the vignetting caused by the occulting spots, we divided the images by normalized ``spot flats'' that were appropriately registered to the approximate locations of the migratory occulting spots on the date of our observations ( Krist et al. 2004). We then averaged the constituent images of each set of exposures listed in Table 1 after interpolating over static bad pixels and eliminating transient bad pixels with a conventional 3 rejection algorithm. We then normalized the averaged images to unit exposure time and replaced saturated pixels in the long-exposure images with unsaturated pixels at corresponding locations in the shorter exposure images. Throughout this process, we tracked the uncertainties associated with each image pixel. In this manner we created cosmetically clean, high-contrast images and meaningful error maps for each combination of star, filter, and roll angle. Figure 1 shows 29 00 ; 10 00 sections of the reduced F606W images of  Pic and Pic obtained at each roll angle. 2.2. Subtraction of the Coronagraphic PSF To distinguish the brightness and morphology of the disk from the diffracted and scattered light of  Pic, the occulted star 's PSF must be removed from each image. By observing Pic and  Pic in consecutive HST orbits, we limited the differences between the

Fig. 1.-- 29 00 ; 10 00 sections of the F606W ( broad V ) images of  Pic (top, middle) and the reference star Pic (bottom) obtained with the ACS/ HRC coronagraph. The images are displayed with logarithmic scaling but without correction of geometric distortion. The dust disk around  Pic, which is viewed nearly edge-on, is evident without subtraction of the stellar PSF. The apparent position angles of the disk in the top and middle panels change in accordance with the HST roll offset of 9N7. The linear feature seen in all three panels is a component of the coronagraphic PSF, whose origin is currently unknown. It is collinear with the HRC's occulting finger, which lies beyond the FOV of each panel.

coronagraphic PSFs of the two stars that would otherwise be caused by inconsistent redeployment of the coronagraph assembly, gradual migration of the occulting spot, or changes in HST 's thermally driven focus cycles ( Krist 2002). We measured the positions of the stars behind the occulting spot using the central peaks of the reduced coronagraphic PSFs ( Fig. 1) that result from the reimaging of incompletely occulted, spherically aberrated starlight by ACS's corrective optics ( Krist 2000). The positions of  Pic and Pic differed by $0.8 pixels ($0B02). This offset causes large differences between the coronagraphic PSFs in the immediate vicinity of the occulting spot, but the residual light at larger field angles (k500 from the star) after PSF subtraction is $103.5 times fainter than the disk's midplane at those field angles ( Krist 2000). Optimal subtraction of the coronagraphic PSF requires accurate normalization and registration of the filter images of the reference star Pic with the corresponding images of  Pic. Because direct images of the two stars were not obtained, we estimated the brightnesses of each star in each ACS bandpass using the HST synthetic photometry package Synphot, which has been developed and distributed by STScI ( Bushouse et al. 1998). In doing so we used the optical spectra of the A5 V stars 1 Serpentis and Praesepe 154 (Gunn & Stryker 1983) to approximate the spectrum of  Pic. Likewise, we approximated the spectrum of Pic with that of the A5 IV star HD 165475B. These proxies yielded synthetic Johnson-Cousins photometry that closely matches the Cousins BVRI measurements of Pic and  Pic reported by Bessel (1990). Assuming V magnitudes of 3.27 and 3.86 for Pic and  Pic, respectively, we computed synthetic flux ratios


3112

GOLIMOWSKI ET AL.

Vol. 131

Fig. 2.-- Multiband HRC images of the disk around  Pic after subtraction of the stellar PSF but before deconvolution of the off-spot PSF. The panels show 27B2 ; 8B3 sections of the F435W (B), F606W ( broad V ), and F814W ( broad I ) images. The image sections have been rotated so that the northeast extension of the disk appears to the left of each panel and the midplane of the outer disk (radius k100 AU ) is horizontal. Each color-coded panel shows the logarithm of the disk's surface brightness relative to the star 's brightness in that bandpass. The irregular, blackened regions near the center of each panel reflect imperfect PSF subtraction around the occulting spot, especially along the direction of the occulting finger, which is beyond the exhibited FOV.

Fig. 3.-- Same as Fig. 2, but with the vertical scale expanded by a factor of 4. The color-coded images represent the surface brightnesses of the disk relative to those measured along the spine of the disk. (See x 2.2 for details.) The expanded vertical scale exaggerates the warp in the inner region of the disk first reported by Burrows et al. (1995). The red dot seen to the left of the ``NE'' label in the bottom panel is a very red background source located 11B5 from  Pic at a position angle of 32N0.

F /F of 1.65, 1.75, and 1.90 for F435W, F606W, and F814W, respectively. We then divided the images of Pic by these ratios to bring the integrated brightnesses of the reference PSFs into conformity with those of  Pic. We aligned the normalized images of Pic with the corresponding images of  Pic using an interactive routine that permits orthogonal shifts of an image with subpixel resolution and cubic convolution interpolation. The shift intervals and normalization factors (i.e., F /F) were progressively refined throughout the iterative process. We assessed the quality of the normalization and registration by visually inspecting the difference image created after each shift or normalization adjustment. Convergence was reached when the subtraction residuals were visibly minimized and refinements of the shift interval or normalization factor had inconsequential effects. Based on these qualitative assessments, we estimate that the uncertainty of the registration along each axis is 0.125 pixels and the uncertainty of F /F in each bandpass is 2%. After subtracting the coronagraphic PSFs from each image, we transformed the images to correct the pronounced geometric distortion in the HRC image plane. In doing so we used the coefficients of the biquartic-polynomial distortion map provided by STScI ( Meurer et al. 2002) and cubic convolution interpolation to conserve the imaged flux. We then combined the images obtained at each HST roll angle by rotating the images of the second group clockwise by 9N7 ( Table 1), aligning the respective pairs of images according to the previously measured stellar centroids and averaging the image pairs after rejecting pixels that exceeded

their local 3 values. Again, we tracked the uncertainties associated with each stage of image processing to maintain a meaningful map of random pixel errors. We combined in quadrature the final random-error maps with estimates of the systematic errors caused by uncertainties in the normalization and registration of the reference PSFs. Other systematic errors from cyclic changes of HST 's focus and differences between the field positions and broadband colors of Pic and  Pic are negligible compared with the surface brightness of  Pic's disk over most of the HRC's FOV ( Krist 2000). Our systematic-error maps represent the convolved differences between the optimal PSF-subtracted image of  Pic and three nonoptimal ones generated by purposefully misaligning (along each axis) or misscaling the images of Pic by amounts equal to our estimated uncertainties in PSF registration and F /F . The total systematic errors are 1 ­ 5 times larger than the random errors within $300 of  Pic, but they diminish to 10% ­ 25% of the random errors beyond $600 of the star. We refer to the combined maps of random and systematic errors as total-error maps. Figure 2 shows the reduced and PSF-subtracted images of the disk in each ACS bandpass. Each image has been rotated so that the northeast extension of the disk is displayed horizontally to the left of each panel. The images have been divided by the brightness of  Pic in each bandpass derived from Synphot.23
All calibrated surface brightnesses and colors presented in this paper are based on the following Vega-based apparent magnitudes for  Pic obtained from Synphot: mF435W ¼ 4:05, mF606W ¼ 3:81, and mF814W ¼ 3:68. The systematic zero-point errors are <2% (Sirianni et al. 2005), and the estimated errors from imperfectly matched reference spectra are $1% ­ 2%.
23


No. 6, 2006

MULTIBAND IMAGES OF DISK AROUND  PIC

3113

Fig. 4.-- Same images as in Fig. 2 after Lucy-Richardson deconvolution of the off-spot PSF. Pixels lying within a radius of 1B5($30 AU ) of the image center (i.e., the location of the occulted star) have been masked and excluded from the deconvolution. Because the Lucy-Richardson algorithm forces all pixels to have positive values, the deconvolved images exhibit no negative PSF-subtraction residuals and enhanced, correlated noise at faint signal levels. Consequently, photometry of the disk in these regions of the disk is less reliable than in regions having large S/ N ratios and small subtraction residuals.

Fig. 5.-- Same the off-spot PSF. deconvolved filter Side effects of the

images as in Fig. 3 after Lucy-Richardson deconvolution of The color-coded isophotes are more similar among these images than among the convolved images shown in Fig. 3. deconvolution process are described in Fig. 4.

The alternating light and dark bands near the occulting spot reflect imperfect PSF subtraction caused by the slightly mismatched colors and centroids of Pic and  Pic. The bands perpendicular to the disk have amplitudes that are $50% ­ 100% of the midplane surface brightnesses at similar distances from  Pic. These residuals preclude accurate photometry of the disk within 1B5 ($30 AU ) of the star and anywhere along the direction of the occulting finger. Along the midplane of the disk, the photometric uncertainties due to PSF subtraction are $5% ­ 10% at a radius of r ¼ 30 AU and less than 1% for r > 60 AU. Figure 3 shows alternate views of the disk in which the vertical scale is expanded by a factor of 4 over that presented in Figure 2 and the vertical dimension of the disk's surface brightness is normalized by the brightness measured along the ``spine'' of the disk. ( The spine comprises the vertical locations of the maximum disk brightness measured along the horizontal axis of each image, after smoothing with a 3 ; 3 pixel boxcar.) The expanded vertical scale exaggerates the warp in the inner disk first observed in images taken with HST 's Wide Field Planetary Camera 2 ( WFPC2) by Burrows et al. (1995). The multiband images shown in Figures 2 and 3 can be directly compared with the unfiltered optical image of the disk obtained with the Space Telescope Imaging Spectrograph (STIS) coronagraph (Grady et al. 2003) and shown in Figure 8 of Heap et al. (2000). 2.3. Deconvolution of the ``Off-Spot'' PSF Accurate assessment of the chromatic dependencies of the disk's color and morphology requires the deconvolution of the unocculted instrumental PSF from each HRC filter image. This

deconvolution of the ``off-spot'' PSF is especially important for the F814W images, because very red photons (k k 0:7 m) passing through the HRC's CCD detector are scattered diffusely from the CCD substrate into a large halo that contributes significantly to the wide-angle component of the PSF (Sirianni et al. 2005). Unfortunately, no collection of empirical off-spot reference PSFs exists yet for the HRC coronagraph. Consequently, we can deconvolve the off-spot PSFs only approximately by using synthetic PSFs generated by the Tiny Tim software package distributed by STScI ( Krist & Hook 2004). Tiny Tim employs a simplistic model of the red halo that does not consider its known asymmetries ( Krist et al. 2005b), but this model is sufficient for assessing the general impact of the red halo on our images of  Pic's disk. We generated model off-spot PSFs using the optical prescriptions, filter transmission curves, and sample A5 V source spectrum incorporated in Tiny Tim. For simplicity we approximated the weakly field-dependent PSF in each bandpass with single model PSFs characteristic of the center of the FOV. The model PSFs extended to an angular radius of 1000 . We corrected the geometrically distorted model PSFs in the manner described in x 2.2, and then deconvolved them from the PSF-subtracted images of  Pic obtained at each roll angle. In doing so we applied the Lucy-Richardson deconvolution algorithm ( Richardson 1972; Lucy 1974) to each image outside a circular region of radius 1B5 centered on the subtracted star. ( The amplitudes and spatial frequencies of the PSF-subtraction residuals within this region were too large to yield credible deconvolved data.) The imaged FOV lacked any bright point sources by which we could judge convergence of the deconvolution, so we terminated the computation after 50 iterations. Examination of intermediate stages of the process showed no perceptible change in the deconvolved images after $45 iterations.


3114

GOLIMOWSKI ET AL.

Vol. 131

Figures 4 and 5 are the deconvolved counterparts of Figures 2 and 3, respectively. The compensating effect of the deconvolution is especially evident when comparing the morphologies of the disk in Figures 3 and 5. The color-coded isophotes of the disk in each bandpass are much more similar in the deconvolved images than in the convolved images. The amplified, correlated noise in the deconvolved images is characteristic of the Lucy-Richardson algorithm when applied to faint, extended sources. It is a consequence of the algorithm's requirements of a low background signal, nonzero pixel values, and flux conservation on both local and global scales. These requirements also account for the disappearance of the negative PSF-subtraction residuals near the occulting spot. The requirement of local flux conservation ensures reliable photometry in regions where the ratio of signal to noise (S/ N ) is large and where PSF-subtraction residuals are small, but it makes photometric measurements elsewhere less accurate and their uncertainties nonanalytic. 3. IMAGE ANALYSIS The processed ACS/ HRC images shown in Figures 2 and 3 are the finest multiband, scattered-light images of  Pic's inner disk obtained to date.24 Earlier ground-based, scattered-light images show the usual effects of coarse spatial resolution and PSF instability caused by variable atmospheric and local conditions (Smith & Terrile 1984; Paresce & Burrows 1987; Golimowski et al. 1993; Lecavelier des Etangs et al. 1993; Kalas & Jewitt 1995; Mouillet et al. 1997a, 1997b). The unpublished BVRI WFPC2 images of the disk described by Burrows et al. (1995) have comparatively low S/ N ratios because WFPC2 lacks a coronagraphic mode, and directly imaged starlight scatters irregularly along the surface of its CCD detectors. These conditions forced short exposure times to avoid excessive detector saturation and created irreproducible artifacts in the PSF-subtracted images. Moreover, the construction of a WFPC2 reference PSF from images of  Pic obtained at several roll angles allowed possible contamination of the reference PSF by the innermost region of the disk. The unfiltered STIS images of the disk ( Heap et al. 2000) compare favorably with our HRC images, notwithstanding their lack of chromatic information and partial pupil apodization (Grady et al. 2003). Both sets of images show the same region of the disk, although STIS's narrow occulting wedge permitted imaging of the disk's midplane about 0B4 (8 AU ) closer to the star. The HRC's circular occulting spot and Lyot stop exposed the regions around the projected minor axis of the disk that were obscured in the STIS images by diffraction spikes and the shadow of the occulting bar. The HRC images have twice the spatial resolution of the STIS images, and they exhibit better S/ N ratios in the regions of the disk between 150 and 250 AU from the star. 3.1. Disk Morphology Figure 6 shows isophotal maps of our F606W images of the disk before and after deconvolution of the off-spot PSF. These maps are qualitatively similar to those gen