Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.stsci.edu/hst/wfpc2/documents/isr/wfpc2_isr1003.pdf Дата изменения: Mon Aug 16 23:05:45 2010 Дата индексирования: Mon Sep 13 16:03:25 2010 Кодировка: Поисковые слова: earth's atmosphere

Operated for NASA by AURA

SP CE A TELESCOPE SCIENCE INSTITUTE

Instrument Science Report WFPC2 2010-03

WFPC2 CTE for Extended Sources: I. Photometric Correction
Grogin, N. A., Lucas, R. A., Golimowski, D., Biretta, J. August 13, 2010

A

BSTRACT

We measure the effects of charge transfer efficiency (CTE) losses on resolved sources in Wide Field and Planetary Camera 2 (WFPC2) images from 1995 to 2008. We compare medium and long exposures of the Hubble Deep Field ­ North taken with the F606W filter at several epochs against a "truth" mosaic of the field taken with the Advanced Camera for Surveys early in that instrument's operation. We adopt the Dolphin (2009) functional form for the CTE photometric correction and determine the optimal coefficients for extended sources on the Wide Field (WF) detectors as a function of observation date, pre-corrected source flux, background flux, and source position on the detector.

Intro duction
The charge transfer efficiency (CTE) of a charge-coupled device (CCD) can be nearly perfect upon fabrication, but it degrades steadily with exposure to energetic charged particles. This degradation of CTE is most notable in CCDs that operate for extended perio ds outside the Earth's atmosphere, such as those of the Hubble Space Telescope. The continuous cosmic-ray bombardment of the HST CCDs intro duces lasting defects in their lattice structure, which cause "charge traps" that prevent the transfer of accumulated charge from
Copyright c 2010 The Association of Universities for Research in Astronomy, Inc. All Rights Reserved.


pixel to pixel during read-out. These charge traps their electrons on varying timescales, that can extend to dozens or more parallel-clo ck cycles. There is evidence that a small fraction of WFPC2 trapped charge is released on much longer timescales, resulting in ghost images in later exposures (Biretta & Mutchler 1997). Imperfect CTE causes charge from an imaged source to be deferred into a tapering trail extending in the direction opposite the parallel read-out ("YCTE"). The charge trails also exist in the serial read-out direction ("XCTE"), but XCTE is typically a factor of 10 or more weaker than the YCTE. Sources far from the read-out amplifier, which undergo the most charge transfers during read-out, show more prominent deferred-charge trails. In extreme cases, the deferred charge can represent 30% or more of the original charge, particularly for faint sources with large Y-co ordinates in low-background images. Software remediation of the charge trailing is nontrivial in the presence of detector read-noise, source Poisson noise, and crowded fields (Massey et al. 2009; Anderson & Bedin 2010). As of 2010, the longest operational HST CCDs were those of the Wide Field and Planetary Camera 2 (WFPC2), which were in orbit from 1993 (Servicing Mission 1) until 2009 (Servicing Mission 4). Prior studies to characterize the WFPC2 CTE based their findings upon charge trails from hot-pixels (Biretta & Kozhurina-Platais 2005) or from unresolved external sources such as stars (e.g., Whitmore, Heyer, & Casertano 1999) or cosmic rays (Riess, Biretta, & Casertano 1999). A substantial portion of WFPC2 science is concerned with compact but well-resolved sources (typically distant galaxies), whose images might not be affected by CTE in the same way as unresolved sources. For example, extended sources may fill charge traps during read-out to a greater extent than unresolved sources (Riess 2000). This report contains the first quantitative assessment of the impact of evolving WFPC2 CTE on resolved-source photometry, and it provides a formula for correcting WFPC2 extendedsource photometry akin to earlier point-source corrections (e.g., Dolphin 2009). In the following sections of this report, we describe the observations and reductions, detail the photometry and the CTE-co efficient fitting, and summarize our findings along with recommendations for their usage.

Data
Investigating the long-term variation of WFPC2 CTE for extended sources requires repeated monitoring of a high Galactic latitude field at considerable exposure depth during most of the operational life of WFPC2. The only such field meeting all these requirements is the Hubble Deep Field ­ North (HDF-N).

2


The HDF-N has been observed with WFPC2 through the F606W filter at three widelyspaced epo chs: during 1995 (Program 6337; P.I. Williams), during 2000 (Program 8389; P.I. Ibata), and during 2007­8 (Program 11032; P.I. Golimowski). The last epo ch includes observations with the telescope rotated by 180 with respect to the original HDF-N. This rotation allows sources in the WF3 field of view to be imaged at two lo cations symmetric about the center of the CCD. We thereby obtain additional leverage to constrain the variation of CTE with position on the detector. Mo derately deep observations of the HDF-N also have been taken with the ACS/WFC camera using the same filter (the "wide-V" F606W) employed in most of the WFPC2 monitoring epo chs. These observations come primarily fron the Great Observatories Origins Deep Survey (GOODS; Cycle 11 Program 9583), with follow-up visits in Cycles 12­13 for high-z supernova detection (Programs 9728, 10189, and 10339) and in Cycle 14 for grism pre-imaging (Program 10530). The HDF-N was additionally observed in a follow-up WFPC2 epoch in 1997 using the F814W filter (Program 6473; Gilliland, Nugent & Phillips 1999). These F814W observations are not as useful for constraining CTE evolution because they are relatively close in time to the original HDF-N, and because there exist no "truth image" ACS/WFC observations in F814W as is the case with F606W. For the present study, we restrict our analysis to the F606W observations. We also have opted to omit photometry from the WF4 CCD from this study, out of concern over the WF4 bias anomaly identified in 2002 (Biretta & Gonzaga 2005). We do not wish to alias low-level photometric errors caused by the WF4 anomaly into the CTE evolution term. Because the ACS/WFC observations were mostly taken early in that instrument's lifetime, the differential effect of CTE across those images is very small. We include a zero-point term in the CTE fitting to allow for a time-invariant offset between WFPC2 F606W and ACS F606W.

List of Observations Please consult the Appendix for the full lists of exposures, both ACS/WFC and WFPC2, used to create the image stacks analyzed in this study. The listed exposures have been visually inspected and found to be free from prominent cosmetic defects such as vignetted earth-shine or bright trails from satellite/debris glint. All F606W exposures were acquired at (for WFPC2) or near (for ACS/WFC) the coordinates of the HDF-N. The exposure times vary from 300­650 sec for ACS/WFC and 3


700­1400 sec for WFPC2. The position angles of the WFPC2 images are identical to the first HDF-N epo ch, with the exception of the 180-rotated Epo ch 2007a and Epo ch 2008. For these rotated WFPC2 epo chs, only the WF3 field is used (see Fig. 1). The position angles of the ACS/WFC images vary widely, as is evident from the stacked weight-map shown in Figure 2. The diverse pointing co ordinates and position angles mitigate any systematic celestial-co ordinate trend in low-level CTE effects across the ACS/WFC "truth" mosaic image.

Fig. 1.-- A portion of the ACS/WFC F606W "truth image" mosaic containing the WFPC2 fields investigated in this study (green boxes). The image is displayed in reverse color-map with a = 0.5 ("square ro ot") stretch to highlight the many faint galaxies. Green labels indicate the axis origin and orientation for the HDF-N WF2 and WF3; blue labels for the 180 -rotated epo chs (denoted WF3*).

4


Fig. 2.-- The ACS/WFC mosaic weight map corresponding to the science image of Figure 1. The ACS/WFC exposure depth within the WF2 and WF3 fo otprints (green boxes) varies between 8 ksec and 15 ksec, with a typical level of 11 ksec. The ACS/WFC CCD gap is imprinted on the weight map as the narrow darker bands, indicating lower exposure depth.

We created the ACS/WFC truth image using the Multidrizzle software1 to stack the component images into a North-up, East-left mosaic with square 0. 05 pixels, very close to the native ACS/WFC pixel scale. To precisely align the component images prior to stacking, we determined delta shifts and rotation for each image by cross-correlating the drizzled single images with the GOODS-North "Version 2" mosaic. When performing this cross-correlation, we masked out the (few) stars in the GOODS mosaic. We found that the best compromise
1

http://stsdas.stsci.edu/multidrizzle

5


for jointly minimizing the mosaic's pixel-noise correlation, PSF FWHM, and weight-map smo othness, was to drizzle with a square kernel at pixfrac=0.8.

Data Characteristics CTE losses depend upon both the fluence of the source in electrons and also the background level of the image. While the former is directly proportional to exposure time, the latter is a more complicated function of exposure time and other time-variable factors such as telescope orientation with respect to the Earth limb. Fortunately the HDF-N monitoring epo chs have enough images of varying exposure time and sky background count-rate to construct multiple mosaics per epo ch. We created a total of 17 WF2 and WF3 mosaics (see Appendix), each of which approximately matches the depth of our ACS/WFC truth image. Figure 3 shows the sampling in epo ch (X-axis) and sky level (Y-axis). Due to the preponderance of long broadband exposures in our study, most HDF-N image stacks have large background levels: > 80e- . By contrast, point-source CTE studies (e.g., Dolphin 2009) generally investigate short exposures of crowded stellar fields (typically Cen) with low sky background. For example, the data-set used in the latest study by Dolphin (2009) has a median sky background of 1.16e- , and 84% of the sample has sky level < 9.2e- . Although we are not able to probe the most extreme CTE losses that o ccur at the lowest background levels, it is noteworthy that most of the WFPC2 science programs concerned with photometry of compact resolved sources have been taken with mo derate- to long-duration broadband exposures, whose backgrounds are much closer to 80e- than 1e- . After segregating the WFPC2 images into bins of comparable sky level and exposure time, we used the Multidrizzle software to combine individual exposures into a higher-S/N stacked image on the same world co ordinate system (WCS) as the ACS/WFC truth image. We corrected the WF images' WCS for both shift and rotation by cross-correlating singledrizzled images with the ACS/WFC truth image, after masking the latter's stars. This full-frame image registration is not perfect, because charge-trailing causes small but systematically varying displacements of the centroids of WFPC2 sources vis-`-vis their ACS/WFC a counterparts. However these CTE-induced astrometric errors are smaller for the brighter sources that dominate the cross-correlation. There is prior evidence that conclusions regarding extended-source CTE are largely insensitive to full-frame registration versus lo cal registration on a source-by-source basis (Riess 2000). Each WF stack comprises 10­30 dithered images, permitting a sub-sampled drizzle to 0. 05/pixel with pixfrac=0.6 while avoiding unacceptable fluctuations in the mosaic weight

6


Fig. 3.-- Sampling of the time domain (X-axis) and background level (Y-axis) for WFPC2 image stacks from the three HDF-N F606W monitoring campaigns used for CTE determination. The earliest observations are the original HDF-N in 1995; the last set of observations conclude in 2008. The GO-11032 observations include two epochs where the telescope was rotated 180 with respect to the original HDF-N, and thus only WF3 (crosses) overlapped the HDF-N. All three programs include at least one image stack of WF2 (triangles). map. The compact dither pattern in all the HDF-N epo chs insures that relative CTE losses among images of the same source in the dithered frames can be considered invariant. 7


Analysis
The most straightforward approach to the extended-source photometry would be source extractions upon each image stack independently, followed by catalog-matching according to source position. Because of the diversity in sky backgrounds and mean exposure times among our image stacks, an isophote-based source extraction such as that provided by SExtractor (Bertin & Arnouts 1996) would yield inconsistent source catalogs with systematic photometry variations as a function of signal-to-noise. Such variations would rival, if not overwhelm, the CTE effects that we hope to measure. For this reason, we instead ran SExtractor in dual-image mo de, using the ACS/WFC truth image as the detection image. In this manner we determined the photometry of each WFPC2 image stack consistently from the source positions on the single detection image. No catalog-matching was required to collate the photometry for a given source in the multiple image stacks. The chief drawback to this metho d is that by using a fixed aperture in all images, the photometry is susceptible to systematic biases from misalignment of the image stacks and from variations in the HST fo cus. We later discuss the mitigation of these potential biases when fitting the CTE functional co efficients. The input to our CTE co efficient fitting is a merged photometry catalog from all source extractions of the WF image stacks. We converted the SExtractor FLUX AUTO, an approximate measure of the total flux for an extended source, from the native counts/sec into both micro janskys and electrons (the latter from multiplying by the gain and by the mean exposure time in the given image stack). Because we seek to fit subtle changes in the CTE loss as a function of multiple variables, we exclude any source that do es not have at least S/N > 5 in al l image stacks. The resulting merged catalog includes 5105 entries for 638 distinct sources. The distribution of source fluences and fluxes (the former is the quantity of importance to the CTE) is shown in Figure 4. The distribution of source lo cations across the WF2 and WF3 detectors is shown in Figure 5, plotted versus their flux differential with respect to the ACS/WFC truth image. We adopt the functional form for CTE loss derived by Dolphin (2009) from WPFC2 images of standard star fields. We reprint the description from Dolphin's website2 , which is equivalent to the form given in Dolphin (2009), and which contains the co efficients that best-fit the unresolved sources: Given a star brightness CTS, background level BG (both in electrons), ob2

http://purcell.as.arizona.edu/wfpc2 calib/

8


Fig. 4.-- Left: Ensemble of source fluences used in this study, extracted from the HDFN WF2 (blue) and WF3 (red) fields, plotted against the signal-to-noise ratio as reported by SExtractor. The dotted line at S/N = 20 denotes our artificial S/N ceiling for the CTE-co efficient fitting: sources with S/N > 20 are treated as S/N = 20 in the fitting. Right: As left panel, but plotted versus source flux. Photometry of the same source at different exposure times and/or backgrounds will appear as a vertical lo cus on this plot. servation date DATE (in MJD), and star position on the image X and Y , the following sequence of calculations will provide the X C T E and Y C T E losses, both in magnitudes. lbg = 1 ln(BG2 + 1) - 1 2 DATE - 49461.9 yr = 365.25 X 800 (1 ) (2 ) (3 ) (4 ) (5 ) (6 ) (7 )

XCTE = 0.0077 в exp(-0.50 в lbg )(1 + 0.10 в y r ) в lct = ln(CTS) + 0.921 в XCTE - 7

c1 = max (1.0 - 0.201 в lbg + 0.039 в lbg в lct + 0.002 в lct, 0.15) c2 = 0.958 в (y r - 0.0255 в y r 2) в exp(-0.450 в lct) YCTE = 2.41 в ln{(1 + c2 ) в exp[0.02239 в c1 в (Y/800)] - c2 }

9


Fig. 5.-- Left: Ensemble of source X- and Y-co ordinates used in this study, extracted from the HDF-N WF2 (blue) and WF3 (red) fields, plotted versus the fractional flux difference between WFPC2 and ACS flux in F606W. Note that the offsets of 1 for lbg and 7 for lct were put in place for numerical stability, and do not affect the solution itself. For the resolved sources of our study, we determined the best-fit values of the thirteen free parameters of Dolphin's prescription, plus an additional parameter for a fixed zero-point offset between WFPC2 F606W and ACS/WFC F606W -- the source of our "truth image" photometry. Note that our extended-source brightnesses are determined not with Dolphin's "HSTPhot", but with SExtractor FLUX AUTO. We performed the non-linear least-squares optimization of the Dolphin co efficients by using an IDL implementation of the downhill simplex algorithm "amo eba" from Numerical Recipes (Press et al. 1992). To mitigate the possibility of the algorithm reporting a lo cal rather than global minimum of the co efficient optimization, we re-started the optimization multiple times while varying the starting positions of the optimization simplex. To prevent the optimization from being dominated by a few bright sources, which in theory have small CTE losses, we imposed a ceiling of S/N = 20 on the photometric catalog. This was accomplished by artificially increasing the SExtractor-computed error term FLUXERR AUTO for the S/N > 20 sources, such that the ratio FLUX AUTO/FLUXERR AUTO= 20. We also found that the optimization was more robust if we clipped 5% of the most extreme outliers when computing the sum-of-squares at each iteration of the amo eba. This 10


outlier-clipping has some a priori motivation, insofar as photometry differences between the WF stack and the ACS/WFC truth image can arise not only from CTE and Gaussiandistributed noise, but also from non-Gaussian-distributed systematics including misalignment, fo cus variation, imperfect cosmic-ray rejection, possible WFPC2 first-frame anomaly (McMaster & Biretta 2010), etc. Using the CTE co efficient values of Dolphin (2009) to correct our WF2 and WF3 photometry (with our FLUX AUTO as proxy for HSTPhot aperture counts), we measure the 2 per degree of freedom (hereafter 2 ) to be 1.71. If we use the Dolphin (2009) Ї values but optimize for a zero-point offset between WFPC2 and ACS/WFC photometry, we obtain a 2 = 1.66 with a best-fit offset of -0.0216 (in magnitudes, with the sense Ї Z PWFPC2 - Z PACS ). When optimizing al l the Dolphin co efficient values, as well as the zero-point term, we achieve 2 = 1.53 with a negligible zero-point term of -0.00021 mag. Ї By comparison, the uncertainty in WFPC2 zero-points bracketing F606W is 0.017 mag (Heyer et al. 2004). Although the differences in the above 2 values appear mo dest, they Ї are highly significant on account of our rich data-set with > 4830 degrees of freedom in the optimizations. Table 1 lists the best-fit CTE co efficient study) and unresolved sources (Dolphin 2009). equation of the Dolphin prescription in which positive-valued by construction, though some Dolphin equations above. values for the cases of resolved sources (this For convenience, we also note the particular each co efficient appears. All co efficients are appear with a leading negative sign in the

To visualize the dependence of CTE losses on the various observables, and to compare the point-source and extended-source CTE prescriptions, we consider a "worst-case" source that is faint (FLUX AUTO = 100e- ), situated far from the read-out amplifier (Y = 700), observed late in the WFPC2 era (3 February 2008), and toward the low end of our skybackground range (80e- ). Our predicted CTE loss for this source is 0.14 mag, compared with 0.18 mag using the Dolphin (2009) co efficient values. The four panels of Figure 6 show the variation of the CTE for this "worst case" source as each observable is varied while the others are held at their canonical values. In each panel, the dotted line denotes the source, the green curve is the best-fit CTE prediction, and the blue curve is the Dolphin (2009) prediction. As is evident in the top two panels showing the CTE dependence on source fluence and sky background, our data-set of long-exposure broadband photometry inhabits a region in the observables space that causes jagged CTE curves when using the Dolphin (2009) co efficient values (blue curves). For most of the observables, our computed CTE losses are below those of Dolphin across virtually the whole range of variation, sometimes by more than 0.1 mag. 11


Resolveda 4 0 2 2 0 1 0 0 4 0 0 1 0 9478.1 .4690 .20 .88 .0594 .080 .684 .022 .82в10 .0603 .673 .00в10 .1288

Unresolvedb Eqn. 49461.9 0.0077 0.50 0.10 0.201 0.039 0.002 0.15 0.958 0.0255 0.450 2.41 0.02239 2 3 3 3 5 5 5 5 6 6 6 7 7

-5

-4

Table 1: CTE Formula Co efficients for Resolved and Unresolved Sources

a b

This study. Dolphin (2009)

The exception is our best-fit variation of CTE losses with epo ch, shown in the lowerright panel of Figure 6. In this case, the Dolphin co efficient values pro duce the aesthetically superior CTE curve, with only a minor quadratic component. This quadratic term in the time evolution, intro duced subsequent to the Dolphin (2000) study, has been abused by our data-set to pro duce an unrealistic inflection of the CTE loss near MJD=52500. Continual exposure to cosmic rays should continually worsen the WFPC2 CTE. There have been no changes to the operating parameters of the WFPC2 CCDs (temperature, etc.) that would explain a lessening of CTE magnitude between our Epo ch 2000 data (MJD51900) and our Epo ch 2007­8 data (MJD54500). We suspect this prominent quadratic term is due to the sparse sampling of our data in the time domain -- effectively only three points (see Figure 3) to constrain a quadratic dependence. Because our latest HDF-N monitoring epo ch is near the end of the WFPC2 operational lifetime, and thus the time-extrapolation of the CTE curve will never be large, we do not

12


Fig. 6.-- Best-fit variation of extended-source CTE losses (Y-axis) versus source fluence (top left), background level (top right), source Y-co ordinate on the WF detector (lower left), and date of observation (lower right). The dotted line in each panel shows the CTE loss for a base-line source with FLUX AUTO = 100e- lo cated at Y=700 with an 80e- background, observed on MJD=54500 (3 February 2008). Green shows the full co efficient optimization; blue shows the Dolphin (2009) unresolved-source CTE co efficients with optimization only for the ACS-WFPC2 F606W zero-point offset. consider the time-domain CTE inflection of our best-fit co efficient values to present a crisis of confidence. 13


Summary and Discussion
We have analyzed multiple epo chs of HDF-N imaging with WFPC2 to derive CTE co efficients suitable for correcting the measured fluxes of localized extended sources against a mo derate to high background level. This represents the first quantification of extendedsource CTE losses for WFPC2 photometry, following upon the study by Riess (2000) of CTE-induced distortions to WFPC2 galaxy surface-brightness profiles. We recommend the use of our best-fit CTE co efficients (Table 1) to correct WFPC2 extended-source photometry, at least in the regime of mo derate- to high-background level, in preference to the Dolphin (2009) co efficient values derived for point-source CTE. We are currently completing a parallel study distortion in these HDF-N images, as quantified by parameters (Conselice 2003). The importance of particularly as regards weak-lensing studies, has imagery (Rho des et al. 2010). on WFPC2 CTE-induced morphological the galaxy concentration and asymmetry CTE correction to galaxy morphology, received recent attention in ACS/WFC

Although correcting for CTE photometry losses post hoc at the source level has been the fo cus of this study and most previous CTE studies, there has been recent ferment in developing software algorithms to back out the CTE charge-trailing at the pixel level, prior to photometric (or morphological) measurement (Massey et al. 2009; Anderson & Bedin 2010). If pixel de-trailing can be perfected, and without significantly worsening the noise properties of the de-trailed image, this approach should in principle be superior to a post hoc CTE correction derived from a statistical ensemble of sources with imperfect sampling of the dependent variables. While the current efforts are fo cused upon ACS/WFC correction, their success would bo de well for a pixel-based CTE correction to the WFPC2 archive.

Acknowledgments
We express our gratitude to the HST Telescope Time Review Board for granting additional HST orbits to the HDF-N monitoring program 11032, allowing us to repeat exposures corrupted by earth-shine. We thank R. Gilliland, W. Hack, and A. Ko ekemo er for helpful discussions regarding the Multidrizzling (and co-aligning) of our WFPC2 and ACS imagery.

References
Anderson, J. & Bedin, L. 2010, PASP, submitted Bertin, E. & Arnouts, S. 1996, A&AS, 117, 393 14


Biretta, J. & Gonzaga, S. 2005, Instrument Science Report WFPC2 2005-002 (Baltimore: STScI) Biretta, J. & Kozhurina-Platais, V. 2005, Instrument Science Report WFPC2 2005-001 (Baltimore: STScI) Biretta, J. & Mutchler, M. 1997, Instrument Science Report WFPC2 1997-005 (Baltimore: STScI) Conselice, C. J. 2003, ApJS, 147, 1 Dolphin, A. E. 2000, PASP, 112, 1397 Dolphin, A. E. 2009, PASP, 121, 655 Gilliland, R. L., Nugent, P. E., & Phillips, M. M. 1999, ApJ, 521, 30 Heyer, I., Richardson, M., Whitmore, B., & Lubin, L. 2004, Instrument Science Report WFPC2 2004-001 (Baltimore: STScI) McMaster, M. & Biretta, J. 2010, Instrument Science Report WFPC2 2010-001 (Baltimore: STScI) Massey, R., Stoughton, C., Leauthaud, A., Rho des, J., Ko ekemo er, A., Ellis, R., & Shaghoulian, E. 2010, MNRAS, 401, 371 Press W. H., Teukolsky S. A., Vetterling W. T., Flannery B. P. 1992, Numerical Recipes: The Art of Scientific Computing (2nd ed.; New York: Cambridge University Press) Rho des, J., Leauthaud, A., Stoughton, C., Massey, R., Dawson, K., Kolbe, W., & Ro e, N. 2010, PASP, 122, 439 Riess, A. 2000, Instrument Science Report WFPC2 2000-004 (Baltimore: STScI) Riess, A., Biretta, J., & Casertano, S. 1999, Instrument Science Report WFPC2 1999-004 (Baltimore: STScI) Whitmore, B., Heyer, I., & Casertano, S. 1999, PASP, 111, 1559

15


App endix
A. Specific Exposures Used in Image Stacks In this Appendix we tabulate all exposures used to create each of the epo ch and background combinations of WF2 and WF3 image stacks (see Fig. 3), as well as the exposures combined to make the the ACS "truth" image (Fig. 1). For WFPC2 these are the pipelinepro cessed " c0h.fits" files, while for ACS these are the pipeline-pro cessed " flt.fits" files. For WFPC2 epo chs containing more than one sky level, the exposure lists are ordered by increasing sky level. j8dnb1bpq j8dnb1bsq j8dnc8uxq j8dnc8v1q j8dnd2j5q j8dnd2j9q j8dne3z4q j8dne3z8q j8dnf9vwq j8dnf9vzq j8dng3cmq j8dng3d0q j8dnh4nsq j8dnh4nwq j8dnj2mrq j8dnj2muq j91wd8laq j91we4x7q j94sb1dsq j94sb1duq j9fag8xrq j9fag9y5q j9fah8l9q j9fah9urq Table 2: ACS "Truth" Exposures j8dnb2dcq j8dnc9f1q j8dnd3kdq j8dne7i3q j8dng0o jq j8dng4ejq j8dnh8weq j8n1r1dgq j91we5niq j94se1cuq j9fagfh8q j9fahfgwq j8dnb2dfq j8dnb7gyq j8dnb7h1q j8dnc9f4q j8dnd0g0q j8dnd0g3q j8dnd3kiq j8dne2ssq j8dne2t6q j8dne7i7q j8dne8j8q j8dne8jcq j8dng0omq j8dng1qzq j8dng1r3q j8dng4enq j8dnh3u3q j8dnh3u7q j8dnh8wiq j8dnh9v9q j8dnh9vdq j91wd2gsq j91wd3hiq j91wd7jmq j91we6u6q j91we8mgq j91we9tnq j94se1cwq j9fag1qeq j9fag2qlq j9fagghfq j9fah1dlq j9fah2dzq j9fahgh3q

u31p020at u31p020dt u31p020ft u31p030it u31p030pt u31p030qt u31p0713t u31p0715t u31p0803t u31p081ft u31p081ht u31p081tt u31p0827t u31p0829t

u31p020gt u31p021dt u31p021gt u31p0406t u31p0407t u31p0712t u31p080dt u31p080et u31p081dt u31p081vt u31p081zt u31p0824t

Table 3: WF2; Epo ch 1995 (GO-6337); Sky Level 1

16


u31p030kt u31p0402t u31p0506t u31p0604t u31p070ft u31p070it u31p070lt u31p070yt u31p071bt u31p071et u31p071it u31p071lt u31p0809t u31p080jt u31p080kt u31p080mt Table 4: WF2; Epo ch 1995 (GO-6337); Sky Level 2 u31p0404t u31p070bt u31p070ct u31p070ht u31p070pt u31p0710t u31p071ft u31p071mt u31p071vt u31p080nt u31p080ut u31p080xt u31p0813t u31p0815t u31p0818t Table 5: WF2; Epo ch 1995 (GO-6337); Sky Level 3 u5 u5 u5 u5 ei0 ei0 ei0 ei0 1 5 6 7 0 0 0 0 8 a a a

u31p0606t u31p0701t u31p0718t u31p0719t u31p071pt u31p071st u31p080tt

u31p070mt u31p070ot u31p071qt u31p071tt u31p080zt u31p0810t

r u5ei0502r u5ei0504r u5ei0506r u5ei0508r u5ei0509r r u5ei0602r u5ei0604r u5ei0606r u5ei0608r u5ei0609r r u5ei0702r u5ei0704r u5ei0706r u5ei0708r u5ei0709r r

Table 6: WF2; Epo ch 2000 (GO-8389); Sky Level 1 u5ei0106r u5ei0306r u5ei0409r u5ei0607m u5 u5 u5 u5 ei0 ei0 ei0 ei0 1 3 4 7 0 0 0 0 a 9 a 1 r r r r u5 u5 u5 u5 ei0 ei0 ei0 ei0 2 4 5 7 0 0 0 0 4 2 5 3 r r r r u5 u5 u5 u5 ei0 ei0 ei0 ei0 2 4 5 7 0 0 0 0 6 4 7 5 r r r r u5 u5 u5 u5 ei0 ei0 ei0 ei0 2 4 6 7 0 0 0 0 9 7 3 7 r u5ei0304m r u5ei0408r r u5ei0605r r

Table 7: WF2; Epo ch 2000 (GO-8389); Sky Level 2 u9y90901m u9y90902m u9y90907m u9y90908m u9y91003m u9y91004m u9y90903m u9y90904m u9y90905m u9y90906m u9y90909m u9y9090am u9y91001m u9y91002m u9y91005m u9y91006m u9y91007m u9y91008m

Table 8: WF2; Epo ch 2007b (GO-11032) 17


u31p020at u31p020dt u31p020ft u31p030it u31p030pt u31p030qt u31p0713t u31p0715t u31p0803t u31p081ft u31p081ht u31p081tt u31p0827t u31p0829t

u31p020gt u31p021dt u31p021gt u31p0406t u31p0407t u31p0712t u31p080dt u31p080et u31p081dt u31p081vt u31p081zt u31p0824t

Table 9: WF3; Epo ch 1995 (GO-6337); Sky Level 1

u31p030kt u31p0606t u31p0701t u31p0708t u31p070st u31p0718t u31p0719t u31p071bt u31p071et u31p071it u31p0809t u31p080jt u31p080kt Table 10: WF3; Epo ch 1995 (GO-6337); Sky Level 2

u31p070yt u31p071lt

u31p0402t u31p070ht u31p071ft u31p080zt

u31p0506t u31p0604t u31p0709t u31p070it u31p070lt u31p070mt u31p071pt u31p071st u31p071vt

u31p070et u31p070ft u31p070ot u31p070pt u31p080mt u31p080tt

Table 11: WF3; Epo ch 1995 (GO-6337); Sky Level 3

u31p0404t u31p070bt u31p070ct u31p071tt u31p071wt u31p080nt u31p0812t u31p0813t u31p0815t

u31p0710t u31p071jt u31p071qt u31p080ut u31p080xt u31p081