-J. Biretta, C. Ritchie, S. Baggett, and J. Mackenty
03 Jan. 1995
Contents
- I. Overview.
- II. Cycle 3 Earth Flats.
- A. Application of these files
- B. Advantages over Cycle 0/1 flats
- C. Table of Cycle 3 Closure Flats
- III. Cycle 2/3 Delta Flats.
- A. Generation of files
- B. Application of these files
- C. Table of Cycle 2/3 Deltaflats
- IV. High-Fidelity Flats.
- A. Known problems/deficiencies in Earth flats and some Solutions
- B. Identification of High Fidelity Flats
- C. Calculation of High Fidelity Flats -- General Example
- D. Calculation of High Fidelity Flats -- Details
- Type 1 (F122M correction) - PC camera
- Type 2 (F8ND correction) - PC camera
- E. Table of High Fidelity WF/PC-1 Flat Fields
- V. Errata.
- VI. References.
I. Overview:
This memo describes three new flat field calibration products for calibration of WFPC1 images. Briefly stated, these products are:
(I) Cycle 3 Earth Flats. These averaged exposures of the bright sunlit earth. They are similar to, but have some advantages over previous Cycle 0/1 flats. These can be directly applied to unflattened data taken after 02 August 1993.
(II) Cycle 2/3 Deltaflats. These correct the new Cycle 3 flats for application to Cycle 2 data (i.e. data taken between 07 August 1992 and 02 August 1993).
(III) High-Fidelity Flats. These are flats where various artifacts in the Earth flats have been corrected, and should generally be accurate to a few percent. Their greatest advantage will be for calibration of broad band filters, where conventional Earth flats have 25 to 30% errors due to necessary use of neutral density filters. They come in different flavors which are optimized for application to Cycle 2 and 3 data, and for application to short and long exposures.
Below we describe each of these new products, and their advantages over previous calibration products.
II. Cycle 3 Earth Flats:
During Cycle 3 (August to December 1993) approximately 1400 exposures of the bright sun-lit Earth were observed. These have been processed and averaged, resulting in 84 new calibration flats.
A. Generation of these files:
These flats were produced in a manner similar to that used for previous Cycle 1 flats, using STREAKFLAT and NORMCLIP. Unlike previous flats, WFIXUP was also used to average across known bad pixels. Details regarding the input files and processing can be found in the history texts of the .r6h file headers. Unless otherwise noted, only Earth exposures meeting three criteria were used: (1) taken after the 02 August 1993 decontamination, (2) more than 400 DN, and (3) less than 1% saturated pixels in any group.
B. Application of these files:
For uncalibrated data, these files can be used directly with CALWFP. If the data are already calibrated but are unflattened, then these can be simply multiplied into the data images. For application to previously flattened data, the data should be multiplied by the Cycle 3 flat, and then divided by the old Cycle 0/1 flat (see header keyword FLATFILE).
For data taken during Cycle 2, unflattened images should be multiplied by both the Cycle 3 flat, and by the Cycle 2/3 Delta flat closest in wavelength (see Section II).
C. Advantages over Cycle 0/1 flats:
These new Cycle 3 flats will tend to give better removal of time-variable effects for data taken late in the WF/PC-1 mission. The previous Cycle 0/1 flats were, for the most part, observed before Feb. 1992, where as the Cycle 3 flats were observed after 02 August 1993.
For example, small changes in the detector QE occur at each decontamination, and there have been six decontaminations between the Cycle 0/1 flats and data taken near the end of the mission. These decontaminations are not an issue when the new flats are used on data taken after the last decontamination (after 02 August 1993). For data taken between 07 August 1992 and 02 August 1993, there is only a single decontamination event separating the those data from the Cycle 3 flats, and this can be corrected by using the new deltaflats (see Section II).
Also, the old Cycle 0/1 flats were generated before the appearance of the "permanent measle contamination" in Feb. 1992. Hence, they offer no correction for these artifacts. Deltaflats could be used to partially correct the measles, but these were available for only a small number of filters. The new Cycle 3 flats include the measles, and will make a partial correction for these artifacts. (We note that since the measle contaminants occur on out-of-focus surfaces in the cameras, and hence are subject to diffraction effects, flat fields can only make a partial correction for these effects.)
The new flats also cover many several filter/camera combinations which were not included in the previous flats. As we shall see (section III), these new combinations offer the possibility of removing various artifacts from the Earth flats.
D. Table of Cycle 3 Closure Flats:
Below we list the file names, and other information for the new Earth flats. These files are available in the HST data archive. A few flats have individual problems, and these are marked in the comments below; more detailed information can be found in the history text files (.r6h files) of the individual flats. Generic problems with Earth flats are described below (Section III), and elsewhere (Hester 1992; Biretta, et al. 1994).
The table columns are:
DATE = date installed into archive
CAM = camera (wf or pc)
FILTERNAMES = common names of filters
FLATNAME = rootname of file in the archive
S = data source (I = inflight)
EC = number of Earth exposures averaged to
make the flat
T = exposure time of the Earth flats
COMMENT = notes, etc.
CYCLE 3 WF/PC-1 FLATS
=====================
DATE CAM FILTERNAMES FLATNAME S EC T COMMENT
YYYY MM DD 1 2 (SEC)
-------------------------------------------------------------------------
1994 06 14 wf f336w e6e0944pw I 9 0.11 Cycle 3 flat field
1994 06 14 wf F8nd f336w e6e0944nw I 5 30 Cycle 3 flat field
1994 06 14 wf f368m f122m e6e0944tw I 6 80 Cycle 3 flat field
1994 06 14 wf f368m e6e0944rw I 12 0.20 Cycle 3 flat field
1994 06 14 wf f375m f122m e6e09452w I 7 200 Cycle 3 flat field
1994 06 14 wf f8nd f375m e6e09454w I 6 200 Cycle 3 flat field
1994 06 16 wf f375m e6g1340dw I 14 0.50 Cycle 3 flat field
1994 06 16 wf f413m f122m e6g1340gw I 7 40 Cycle 3 flat field
1994 06 16 wf f8nd f413m e6g1340iw I 3 40 Cycle 3 flat field
1994 06 16 wf f8nd f439w e6g1340kw I 4 20 Cycle 3 flat field
1994 06 16 wf f487n f122m e6g1340nw I 10 300 Cycle 3 flat field
1994 06 16 wf f8nd f487n e6g1340qw I 10 300 Cycle 3 flat field
1994 06 13 wf f487n e6d1028aw I 12 0.30 Cycle 3 flat field
1994 06 13 wf f502n f122m e6d1028dw I 11 400 Cycle 3 flat field
1994 06 13 wf f8nd f502n e6d1028gw I 9 400 Cycle 3 flat field
1994 06 13 wf f502n e6d1028jw I 10 0.30 Cycle 3 flat field
1994 06 13 wf f517n f122m e6d1028lw I 2 120 Cycle 3 flat field;
for tests only (note 1)
1994 06 13 wf f547m f122m e6d1028ow I 4 23 Cycle 3 flat field
1994 06 13 wf f8nd f547m e6d1028qw I 2 23 Cycle 3 flat field
1994 07 05 wf f555w f122m e751348dw I 2 6 Cycle 3 flat field
1994 06 13 wf f569w f122m e6d1028sw I 3 8 Cycle 3 flat field
1994 06 13 wf f8nd f569w e6d10291w I 2 8 Cycle 3 flat field
1994 06 13 wf f588w f122m e6d10293w I 10 400 Cycle 3 flat field
1994 06 13 wf f8nd f588n e6d10296w I 6 400 Cycle 3 flat field
1994 06 13 wf f588w e6d10298w I 14 0.18 Cycle 3 flat field;
(note 2)
1994 06 24 wf f588w e6o1002cw I 13 0.18 Cycle 3 flat field
1994 06 08 wf f8nd f606w e6810581w I 4 5 Cycle 3 flat field
1994 06 08 wf f622w f122m e6810583w I 4 8 Cycle 3 flat field
1994 06 08 wf f622w f8nd e6810588w I 4 8 Cycle 3 flat field
1994 06 08 wf f631n e681058aw I 3 0.35 Cycle 3 flat field;
for tests only (note 3)
1994 06 08 wf f648m f122m e681058dw I 5 50 Cycle 3 flat field
1994 06 08 wf f673n f122m e681058fw I 10 350 Cycle 3 flat field
1994 06 08 wf f673n e681058iw I 16 0.12 Cycle 3 flat field
1994 07 05 wf f8nd f702w e751348fw I 2 7 Cycle 3 flat field
1994 06 10 wf f8nd f785lp e6a1427rw I 2 6 Cycle 3 flat field;
for tests only (note 1)
1994 07 05 wf f8nd f785lp e751348hw I 2 6 Cycle 3 flat field;
improved version
1994 06 10 wf f791w f122m e6a1427tw I 4 10 Cycle 3 flat field
1994 06 10 wf f8nd f850lp e6a14282w I 2 6 Cycle 3 flat field;
for tests only (note 1)
1994 07 05 wf f8nd f850lp e751348jw I 2 6 Cycle 3 flat field;
improved version
1994 06 10 wf f889n f122m e6a14285w I 10 300 Cycle 3 flat field
1994 06 10 wf f8nd f889n e6a14288w I 7 300 Cycle 3 flat field
1994 06 10 wf f889n e6a1428aw I 6 0.20 Cycle 3 flat field
1994 06 10 wf f122m f1042m e6a1428dw I 5 80 Cycle 3 flat field
1994 06 10 wf f8nd f1042m e6a1428gw I 4 80 Cycle 3 flat field
1994 06 10 wf f1042m e6a1428iw I 3 0.11 Cycle 3 flat field
1994 06 20 pc f336w e6k1017kw I 4 0.35 Cycle 3 flat field
1994 06 20 pc f368m e6k1017nw I 12 0.60 Cycle 3 flat field
1994 06 20 pc f8nd f368m e6k0945ew I 8 300 Cycle 3 flat field
1994 06 20 pc f8nd f413m e6k0945hw I 15 200 Cycle 3 flat field
1994 06 20 pc f413m e6k0945kw I 6 0.23 Cycle 3 flat field
1994 06 20 pc f8nd f439w e6k0945nw I 3 80 Cycle 3 flat field
1994 06 20 pc f492m f122m e6k0945qw I 4 100 Cycle 3 flat field
1994 06 20 pc f8nd f492m e6k09460w I 3 100 Cycle 3 flat field
1994 06 20 pc f8nd f517n e6k09463w I 3 500 Cycle 3 flat field
1994 06 20 pc f517n e6k09466w I 11 0.35 Cycle 3 flat field
1994 06 17 pc f517n f122m e6h10066w I 11 500 Cycle 3 flat field
1994 06 17 pc f547m f122m e6h10069w I 7 80 Cycle 3 flat field
1994 06 20 pc f555w f122m e6k09468w I 6 26 Cycle 3 flat field
1994 06 20 pc f8nd f555w e6k0946cw I 2 26 Cycle 3 flat field;
for tests only (note 1)
1994 07 05 pc f8nd f555w e751348aw I 2 26 Cycle 3 flat field;
improved version
1994 06 13 pc f569w f122m e6d1559rw I 6 35 Cycle 3 flat field
1994 06 17 pc f606w f122m e6h1006hw I 4 20 Cycle 3 flat field
1994 06 17 pc f8nd f606w e6h1006kw I 5 20 Cycle 3 flat field
1994 06 15 pc f622w f122m e6f11159w I 5 35 Cycle 3 flat field
1994 06 15 pc f8nd f622w e6f1115cw I 4 35 Cycle 3 flat field
1994 06 15 pc f648m f122m e6f1115ew I 5 120 Cycle 3 flat field
1994 06 13 pc f664n e6d16003w I 14 0.26 Cycle 3 flat field
1994 06 13 pc f664n f122m e6d16000w I 16 500 Cycle 3 flat field
1994 06 15 pc f673n e6f1115hw I 7 0.60 Cycle 3 flat field
1994 06 24 pc f8nd f675w e6o10022w I 6 40 Cycle 3 flat field
1994 06 24 pc f675w f122m e6o10024w I 9 40 Cycle 3 flat field
1994 06 13 pc f8nd f702w e6d16006w I 3 26 Cycle 3 flat field
1994 06 24 pc f702w f122m e6o10027w I 5 26 Cycle 3 flat field
1994 06 20 pc f718m f122m e6k1017pw I 8 140 Cycle 3 flat field
1994 06 20 pc f725lp f122m e6k1017rw I 6 26 Cycle 3 flat field
1994 06 24 pc f785lp f122m e6o10029w I 3 30 Cycle 3 flat field
1994 06 13 pc f791w f122m e6d16008w I 3 40 Cycle 3 flat field
1994 06 13 pc f814w f122m e6d1600bw I 4 26 Cycle 3 flat field
1994 06 15 pc f875m e6f1115mw I 2 0.11 Cycle 3 flat field;
for tests only (note 1)
1994 06 15 pc f875m f122m e6f1115jw I 6 80 Cycle 3 flat field
1994 06 13 pc f8nd f875m e6d1600dw I 2 80 Cycle 3 flat field
1994 06 17 pc f8nd f889n e6h1006cw I 14 500 Cycle 3 flat field
1994 06 17 pc f889n f122m e6h1006fw I 10 500 Cycle 3 flat field
1994 06 20 pc f1042m e6k1017iw I 4 0.80 Cycle 3 flat field
Notes to table:
Note 1: These flats have ~5% artifacts in the form of narrow streaks. There are typically half a dozen or so 50-pixel-long streaks per CCD. These are useful for test purposes, and may be suitable for flat fielding if target falls away from streaks. This results from having only a small number of input earthcals which are at nearly the same streak angle.
Note 2: Contains image which pre-dates the August 1993 decontamination.
Note 3: These flats have ~5% artifacts in the form of broad streaks. These are useful for test purposes. This results from having only a small number of input earthcals which are at nearly the same streak angle.
III. Cycle 2/3 Delta Flats:
The above Cycle 3 flats can be applied directly to observations made after 02 August 1993. For observations taken during Cycle 2 (08 August 1992 to 02 August 1993), these flats should be multiplied by one of the delta flats listed below. If a deltaflat is not available for the exact filter used, then one nearby in wavelength should be chosen. The purpose of these delta flats is to correct QE changes which occurred during the 02 August 1993 decontamination, as well as changes in the measle pattern.
A. Generation of these files:
These deltaflats were produced in a manner roughly similar to that used for previous deltaflats (see Baggett and Mackenty, 1993). They are derived from exposures of faint lamps illuminating the backside of the shutter blade. For each filter typically a dozen such "internal flats" were selected to uniformly sample the time period from 08 August 1992 to 02 August 1993, and these were averaged with CRREJ in STSDAS. The same was done for the time period after 02 August 1993, though typically six frames were used for this period. Finally the deltaflat was computed as the ratio of (mean cycle 2 internal flat) / (mean cycle 3 internal flat). Since the two shutter blades have different reflection patterns, care was taken to use only data from one shutter blade. WFIXUP was used to average across bad or questionable pixels in the final deltaflats; these pixels are marked in the data quality files. See comments in the header text (.r8h files) for specific details regarding each deltaflat.
B. Application of these files:
These deltaflats should be multiplied into the Cycle 3 flats, when applying the Cycle 3 flats to Cycle 2 data.
The "edge droop effect" becomes significant at short wavelengths. This is caused by the continuous buildup of contaminants on the CCD window, which scatter short wavelength light. Since this effect increases continuously with time, we have generated F439W deltaflats which are made from data taken during three time intervals in Cycle 2; these intervals are specified by the "USEAFTER" dates in the table below. The choice of deltaflat for F439W depends on both the date the data were taken, and on the illumination pattern of the target. For science images which uniformly illuminate the CCD (planets, or faint targets where the sky dominates) the deltaflat for the appropriate Cycle 2 time interval should give optimum results. For targets which illuminate only a small portion of the CCD (stars, galaxies) the optimum result may be obtained using the earliest Cycle 2 deltaflat (USEAFTER=19920808), regardless of when the data were taken in Cycle 2. Longwards of 5000 Angstroms the edge droop effect was judged to be small (<3%), and so only a single deltaflat was generated.
C. Table of Cycle 2/3 Deltaflats:
The table columns are:
USEAFTER DATE = Earliest date on which the
deltaflat should be used.
This is 08 August 1992 for all
except the F439W deltaflats.
CAM = camera (wf or pc)
MODE = readout mode (full or area)
FILTERNAM = common name of filter
DATA_FILE = rootname of file in the archive
DESCRIP = average value of deltaflat in each CCD
CYCLE 2/3 WF/PC-1 DELTAFLATS
============================
USEAFTER CAM MODE FILTNAM DATA_FILE DESCRIP
DATE
19920808 pc full f439w e821116nw aves: pc1=0.988825 pc2=0.987382 pc3=0.991679 pc4=0.979673
19921130 pc full f439w e821125jw aves: pc1=0.994856 pc2=0.992372 pc3= 1.00386 pc4=0.994884
19930203 pc full f439w e8211320w aves: pc1=0.999157 pc2=0.997177 pc3=1.01398 pc4=1.00482
19920808 pc full f547m e8113302w aves: pc1=0.994264 pc2=0.991861 pc3=1.00484 pc4=0.999357
19920808 pc full f555w e8113305w aves: pc1=0.995486 pc2=0.993987 pc3=1.00647 pc4=1.00034
19920808 pc full f569w e8113307w aves: pc1=0.996229 pc2=0.993603 pc3=1.00617 pc4=1.00022
19920808 pc full f606w e8113309w aves: pc1=0.997309 pc2=0.996645 pc3=1.00701 pc4=1.00418
19920808 pc full f664n e811330cw aves: pc1=0.992573 pc2=0.995378 pc3=1.00724 pc4=1.00271
19920808 pc full f673n e811330ew aves: pc1=0.998685 pc2=0.999236 pc3=1.00194 pc4=1.00077
19920808 pc full f675w e811330gw aves: pc1=0.998288 pc2=1.00034 pc3=1.00026 pc4=1.00114
19920808 pc full f702w e811330iw aves: pc1=0.998633 pc2=1.00033 pc3=0.997967 pc4=0.996168
19920808 pc full f718m e8113311w aves: pc1=0.999649 pc2=1.00028 pc3=0.997892 pc4=0.0.997342
19920808 pc full f785lp e811330lw aves: pc1=1.00339 pc2=1.0029 pc3=1.00011 pc4=0.998251
19920808 pc full f791w e811330nw aves: pc1=1.00542 pc2=1.0057 pc3=1.00122 pc4=0.996179
19920808 pc full f875m e811330qw aves: pc1=1.00704 pc2=1.005 pc3=1.00271 pc4=0.997706
19920808 pc full f889n e811330sw aves: pc1=1.0061 pc2=1.0022 pc3=1.00056 pc4=0.994451
19920808 pc full f1042m e8113300w aves: pc1=1.00036 pc2=1.00028 pc3=0.992321 pc4=0.994532
19920808 wf full f439w e8q11180w aves: wf1=1.00753 wf2=1.00542 wf3=1.00872 wf4=1.003
19930110 wf full f439w e8q13281w aves: wf1=1.00867 wf2=1.00964 wf3=1.01497 wf4=1.00679
19930510 wf full f439w e8q13365w aves: wf1=1.00821 wf2=1.01001 wf3=1.01483 wf4=1.00724
19920808 wf full f547m e8q11184w aves: wf1=1.00196 wf2=1.00071 wf3=1.00659 wf4=0.999276
19920808 wf full f555w e8q11189w aves: wf1=1.00326 wf2=1.0043 wf3=1.00871 wf4=1.0031
19920808 wf full f569w e8q1118dw aves: wf1=1.00052 wf2=1.00141 wf3=1.00559 wf4=1.00048
19920808 wf full f606w e8q1118iw aves: wf1=1.00209 wf2=1.00321 wf3=1.00158 wf4=1.00274
19920808 wf full f622w e8q1118pw aves: wf1=1.00258 wf2=1.00247 wf3=1.00061 wf4=1.00318
19920808 wf full f631n e8q1118tw aves: wf1=1.00007 wf2=1.00308 wf3=1.00598 wf4=1.0031
19920808 wf full f673n e8q11193w aves: wf1=1.00241 wf2=1.00426 wf3=1.00326 wf4=1.00375
19920808 wf full f702w e8q11198w aves: wf1=1.00098 wf2=1.00211 wf3=0.997681 wf4=1.0023
19920808 wf full f785lp e8q1119dw aves: wf1=0.996289 wf2=0.99777 wf3=0.997166 wf4=0.99911
19920808 wf full f889n e8q1119iw aves: wf1=1.00174 wf2=1.00178 wf3=0.997629 wf4=1.00316
19920808 wf full f1042m e8q1117ow aves: wf1=0.998542 wf2=0.998563 wf3=0.998626 wf4=0.998465
IV. High-Fidelity Flats:
The Earth flats are known to have a number of deficiencies which limit the photometric accuracy of calibrated images. For example, many broadband flats are obtained through the F122M filter; this filter contains a 30% intensity gradient across the WF field, and imposes similar errors on the photometry. Also, many flats are short exposures (<1 sec.), and so contain ~5% artifacts related to the short exposure reciprocity failure.
To mitigate these problems we have created "High-Fidelity Flats" where errors are small or have been corrected. In some cases these are merely Cycle 3 Earth flats which have been selected for certain properties, where the errors should be small. In other cases, these are ratios of several flats where the errors are made to cancel out. Below we briefly review the most important problems with the flats, and various situations in which these problems do not occur. Further discussion of some points can be found in Biretta et al. 1994, and references there in.
A. Known problems / deficiencies in Earth flats and some solutions:
In general, Earth flats suffer from a number of defficiencies. The most important effects, briefly summarized, are:
(1.) Patterns attributable to neutral density filters. For the broad-band filters it is generally necessary to use a neutral density filter to prevent saturation of the CCD detector when observing the Earth flats. These neutral density filters introduce a number of artifacts.
In many cases the F122M filter is used to provide neutral density. This filter was designed as a short wavelength pass filter, but its red leak can be used to provide neutral density at long wavelengths. The filter consists of a very weak MgF lens, with a dielectric short pass filter deposited on the side closest to the pick-off mirror. The primary difficulty with this filter, is that the dielectric coating is non-uniform. There is a 25 to 30% brightness gradient running across the WF field, which is very weakly dependent on wavelength within the visual range. There are also several small pinholes in the F122M filter, which causes wavelength dependent "doughnuts" (images of camera relay optics pupils) to appear in the flats with strength 2% to 5%.
In other cases the F8ND filter is used to provide neutral density. This filter consists of a strongly curved, thin meniscus whose center of curvature lies at the pyramid apex. The surface is coated with a highly reflective metal film, which provides neutral density. Unlike the F122M filter, F8ND is highly uniform. Comparison of F8ND flats against the MDS sky flats show that any non-uniformities are less that 1% in intensity. The primary problem with F8ND, is that strong reflections can occur when it is crossed with certain other filters. The strongly curved surface of F8ND, together with its high reflectivity, causes light reflected from other filters to be strongly concentrated in a circular pattern centered near the pyramid apex. (These reflections are formed by light passing through F8ND, reflecting off other filter, reflecting from the curved surface of F8ND, and then passing through the other filter and onto the CCD.) The strength of the circular reflection depends on the filter composition, and varies from <1% to ~30%. These reflections caused the F8ND filter to be dismissed as unusable early in the mission, though more recent tests show it is capable of giving near-perfect flats in certain filters. For filters which consist only of anti-reflection coated colored glass (polarizers, F555W, F569W, F725LP, F785LP, and F850LP), the intensity of the filter reflection is less than 1%. Certain other filter constructions give very weak reflections. For example, in F336W the UG-11 colored glass component has a low throughput and lies between F8ND and the F336W thin film component; hence the filter reflection is reduced to 1-2% in intensity (depending on field position).
(2.) Short exposure reciprocity effects. For exposures <<1 sec., the sensitivity of the CCD corners tends to increase. The effect is largest in the blue (~5%), and diminishes towards at red wavelengths. The effect is strongest for 0.11 sec exposures, and is nearly gone for exposure times >1 sec. Hence it is useful to optimize the exposure times of the flats to match the data being flat-fielded.
(3.) Persistent Measles. These have been discussed extensively elsewhere. (See Biretta et al. 1994, and references there in.) They are small (<10 pixel) features in the flats which are caused by particles on the CCD camera windows that appeared after the Feb. 1992 decontamination. Since these arise from particle contaminants on an out-of-focus surface, diffraction effects are important, and they are difficult to completely corrected with flats. In general, flats will be fairly successful in correcting uniformly illuminated targets (where the illumination most closely matches the flats). But for stellar targets, the flats will only partially correct the intensity error due to the measles. The effects are strongest for the F/30 beam of the PC camera, where 2 - 5% features appear in small regions of PC6, and much of PC5 and PC7, and most of PC8. The effects of the persistent measles are much weaker for the F/13 beam of the WF camera. The new Cycle 3 flats contain the measles, and will offer some correction for them. Since the correction is imperfect, observers are encouraged to examine the delta flats (both the Cycle 2/3 ones mentioned above, and the earlier ones) to check their location relative to targets in science images.
B. Identification of High Fidelity Flats:
From the foregoing discussion, it is apparent that some of the Cycle 3 flats listed in Section I should provide flat-fielding accurate to ~1%. These are flats where the F8ND neutral density filter is crossed with filters containing only anti-reflection colored glass (polarizers, F555W, F569W, F725LP, F785LP, and F850LP), as well as F336W where the construction damps the reflection. Since these involve long exposures (>1 sec.) they are optimum for long-exposure science images.
The narrow band flats taken without any neutral density should also provide excellent flat fielding, though these will generally be optimal for short exposures (<<1 sec.).
C. Calculation of High Fidelity Flats -- General Example:
Given the Earth flats listed in Section I, it is also possible to compute high fidelity flats for other filters. For example, the in the WF camera a high fidelity flat (accuracy better than 2%) can be computed for F555W using F569W together with a correction for wavelength dependent effects. The F122M filter Earth flats can be used to compute the wavelength correction, since the F122M pattern varies slowly with wavelength. From the list of available flats in Section I, a rough high fidelity long-exposure-optimized WF F555W flat can be generated using following equation:
(flat_wf555) = (flat_wf569+8nd) *
[ (flat_wf555+122) / (flat_wf569+122) ]
Where:
(flat_wf555) = desired WF F555W flat
(flat_wf569+8nd) = WF Earth flat taken through F569W + F8ND
(flat_wf569+122) = WF Earth flat taken through F569W + F122M
(flat_wf555+122) = WF Earth flat taken through F555W + F122M
In essence, the part in [ ] represents a correction for wavelength dependent effects between 555nm and 569nm, and is applied to the F569W+F8ND Earth flat, which is known to be near-perfect. In practice this might be computed in IRAF with a script like this:
imarith e751348dw.r6h[1] * e6d10291w.r6h[1] junk.r6h[1/4]
imarith e751348dw.r6h[2] * e6d10291w.r6h[2] junk.r6h[2]
imarith e751348dw.r6h[3] * e6d10291w.r6h[3] junk.r6h[3]
imarith e751348dw.r6h[4] * e6d10291w.r6h[4] junk.r6h[4]
imarith junk.r6h[1] / e6d1028sw.r6h[1] wf555.r6h[1/4]
imarith junk.r6h[2] / e6d1028sw.r6h[2] wf555.r6h[2]
imarith junk.r6h[3] / e6d1028sw.r6h[3] wf555.r6h[3]
imarith junk.r6h[4] / e6d1028sw.r6h[4] wf555.r6h[4]
!uncompress e751348dw.b6d
!uncompress e6d10291w.b6d
!uncompress e6d1028sw.b6d
imarith e751348dw.b6h[1] max e6d10291w.b6h[1]
junk2.b6h[1/4] pixtype="short"
imarith e751348dw.b6h[2] max e6d10291w.b6h[2]
junk2.b6h[2] pixtype="short"
imarith e751348dw.b6h[3] max e6d10291w.b6h[3]
junk2.b6h[3] pixtype="short"
imarith e751348dw.b6h[4] max e6d10291w.b6h[4]
junk2.b6h[4] pixtype="short"
imarith junk2.b6h[1] max e6d1028sw.b6h[1]
wf555.b6h[1/4] pixtype="short"
imarith junk2.b6h[2] max e6d1028sw.b6h[2]
wf555.b6h[2] pixtype="short"
imarith junk2.b6h[3] max e6d1028sw.b6h[3]
wf555.b6h[3] pixtype="short"
imarith junk2.b6h[4] max e6d1028sw.b6h[4]
wf555.b6h[4] pixtype="short"
!compress e751348dw.b6d
!compress e6d10291w.b6d
!compress e6d1028sw.b6d
imdel junk.*
imdel junk2.*
Since all of the input flats have exposures ~8 sec., the resulting flat is optimized for long exposures. A version of the same flat, which is optimized for short exposures may be computed from the available flats in Section I as::
(flat_wf555s) = (flat_wf555) *
[ { (flat_wf588) * (flat_wf569+122) } /
{ (flat_wf588+122) * (wflat_f569+8nd) } ]
Where:
(flat_wf555s) = desired WF F555W flat optimized for short exposures
(flat_wf555) = WF F555W flat derived above, which is optimized for
long exposures
(flat_wf569+8nd) = WF Earth flat taken through F569W + F8ND
(flat_wf569+122) = WF Earth flat taken through F569W + F122M
(flat_wf588) = WF Earth flat taken through F588N
(flat_wf588+122) = WF Earth flat taken through F588N + F122M
The portion in the [ ] effectively acts as a wavelength independent correction from the 400 sec. exposure time of (wf588+122) to the 0.18 sec. exposure time of (wf588).
Work is underway to generate such flats wherever possible, and this memo will be updated periodically to reflect those results.
D. Calculation of High Fidelity Flats -- Details:
(1.) Type 1 (F122M correction)- PC camera:
PC camera Type 1 Hi-Fi flats are computed, in detail, as follows. These are generated by removing the F122M filter artifacts from a flat observed as "filter"+F122M. These artifacts include: (1) a 14% gradient across all four CCDs, (2) 1-2% doughnuts 500 pixels in diameter caused by pinholes in F122M, (3) a 3% clover leaf pattern on P8 in broad band filters caused by the color contributed by F122M, and (4) 2 - 7% depression with in ~150 pixels of the corners of most CCDs in broad band filters due to color contributed by F122M.
A rough multiplicative correction for the F122M filter pattern is first derived by dividing an F555W+F8ND flat by an F555W+F122M flat:
C(F122M) = flat_F555W+F8ND / flat_F555W+F122M
The F122M correction image so derived contains the F122M density gradient and "doughnut" patterns due to pinholes, as desired. It also contains a number of defects which are undesirable: STREAKFLAT artifacts, patterns caused by the color imposed by F122M in combination with changes in the shape of the DQE curve with position on the CCD, and noise. These defects are removed as follows.
The STREAKFLAT artifacts are caused generally by hot pixels in the input Earth flats when there are few input images. These are simply removed by copying nearby pixel values into the corrupted pixels.
The patterns caused by the color imposed by F122M in combination with changes in the shape of the DQE curve with position on the CCD are more difficult to correct. These patterns are manifest as a 3% clover-leaf pattern on CCD P8, and as 2% low regions in the corners of most of the CCDs. Part of the cause can be seen by examining the DQE curves on page 28 of the WFPC1 Instrument Handbook (version 3.0). For example, near 6600 Angstroms all four CCD's have about the same QE, but by 5600 Angstroms the QE of CCD P8 has risen about 20% above that of the other CCD's. In greater detail, it appears that this change in P8 relative to the other CCDs is not uniform across P8, but occurs in a clover-leaf pattern. Since F555W is quite broad, and covers a wavelength range from 4600 to 6600 Anstroms, the blue color imposed by F122M shifts the effective response of the F555W filter in a manner which is spatially variant across P8. To correct this effect we use the F517N flats, which are sufficiently narrow that no color effect should be induced by F122M. A correction for the P8 pattern is derived as:
/ (flat_F555W+F8ND flat_F517N+F122M )
C'(F122M,P8) = C(F122M,P8) / smooth(---------------- * ----------------- )
/ (flat_F555W+F122M (flat_F517N+F8ND)')
where a slight sigma=1.3 pixel smooth is used to reduce noise. For the other CCDs we merely copy:
C'(F122M,P5) = C(F122M,P5)
etc. The image (flat_F517N+F8ND)' has the F8ND reflection pattern removed by averaging:
(flat_F555W+F8ND flat_F517N+F122M )
(---------------- * -----------------)
(flat_F555W+F122M flat_F517N+F8ND )
for P5, P6, and P7, and then dividing this average into flat_F517N+F8ND. (This works, since the F8ND reflection pattern is roughly symmetric about the pyramid apex. Recall that flat_F555W+F8ND has no F8ND pattern since F555W contains only colored glass.)
A correction for the color effect in the CCD corners is derived as follows, also using the F517N flats. First the F8ND reflection pattern is derived starting with flat_F517N+F122M / flat_F517N+F8ND, and then counts are subtracted from the pinhole patterns to level them with other regions of the image. A linear surface approximating the F122M gradient is then fit to the image (excluding the CCD corners and F8ND reflection pattern) and divided out, leaving an image which contains only the F8ND reflection pattern near the pyramid apex, and ~unity elsewhere. This is converted to an additive correction by subtracting unity, and then blanking regions outside the F8ND reflection pattern. The F8ND correction is then applied:
C''(F122M) = C'(F122M) /
(flat_F555W+F8ND flat_F517N+F122M )
smooth (---------------- * ----------------- - [F8ND correction])
(flat_F555W+F122M flat_F517N+F8ND )
where a sigma=20 pixel smooth is used to reduce noise and reduce image gradients contributed by remaining artifacts. Hence, the resulting F122M correction image C''(F122M) contains the F122M gradient and pinhole pattern only; the color effects, STREAKFLAT artifacts, and noise have been reduced or eliminated.
Finally, the desired flat is computed by applying the F122M correction C''(F122M) derived above:
flat_filter = flat_filter+F122M * C''(F122M)
For filters at wavelengths >7500 Angstroms, an additional adjustment is made for wavelength-dependent changes in the F122M pinhole pattern by slightly reducing intensities in the pinholes (~0.2%), and by smoothing the image to simulate the broadening of diffraction patterns at longer wavelengths:
flat_filter = flat_filter+F122M *
smooth ( C''(F122M) - [pinhole adjustment])
For F555W the C(F122M) correction file was used; the QE / color corrections are omitted since these errors are self correcting.
(2.) Type 2 (F8ND correction) - PC camera:
PC camera Type 2 Hi-Fi flats are derived by correcting the F8ND reflection in a flat observed as filter+f8nd.
The f8nd reflection may be understood as follows: The beam from the OTA first passes through the f8nd filter, and then through the spectral filter (f555w, etc.). The reflection results from light striking the spectral filter bouncing back up to the f8nd filter, and then down through the spectral filter and into the CCD camera. The f8nd filter is a thin metal-coated meniscus with its center of curvature at the pyramid, so that the resulting reflection is confined to the center of the field (near the pyramid apex). The strength of the reflection depends on the composition of the spectral filter. For filters containing only anti-reflection coated colored glass, there is almost no reflection (<1%). But for filters containing multi-layer interference filters, the reflection intensity is determined by the detailed shape of the interference filter bandpass (i.e. fraction of light transmitted vs. reflected).
Typically the f8nd reflection is centered at the pyramid apex. It has an outer halo 400 pixels in radius with strength x (in 1-3 % range), and an inner core with radius 200 pixels where the strength reaches about 3x (3-10% range). Filter tilts can cause the reflection center to move up to about 100 pixels from the pyramid apex.
The f8nd pattern derived in the section above from ratios of flat_F555W+F8ND, flat_F555W+F122M, flat_F517N+F8ND, and flat_F517N+F122M is used for this correction. This additive pattern is scaled by a constant, added to unity, and then divided out of the filte+f8nd flat as follows:
flat_filter = flat_filter+f8nd / ( 1.0 + scale * [F8ND correction])
The scale factor is determined by trail-and-error, by examining ratios of the resulting flat divided by flats at similar wavelengths observed as filter+open. Values of the scale factor are typically in the range 0.6 to 1.4.
In some cases the filter is tilted, so that the f8nd reflection moves away from the usual position at the pyramid apex. In these cases some additional work is needed to smooth the f8nd pattern (to correct regions along pyramid edge) and then shift it so it is registered with the reflection pattern in flat_filter+f8nd.
E. Table of High Fidelity WF/PC-1 Flat Fields:
At present, the following flats in the archive can be regarded as high fidelity flats. These are thought to have systematic errors <2%.
The table columns are:
DATE = date installed into archive
CAM = camera (wf or pc)
FILTER = common name of filter
FLATNAME = rootname of file in the archive
T = effective exposure time
COMMENT = notes, etc.
HIGH FIDELITY WF/PC-1 FLATS
===========================
DATE CAM FILTER FLATNAME T COMMENT
YYYY MM DD (SEC)
--------------------------------------------------------------------
Optimized for Cycle 2 (08 Aug. 1992 to 01 Aug. 1993), long exposures (>1 sec.):
1994 12 27 pc f439w ecr1418hw 80 Hi-Fi Flat, Cycle 2,
for data taken 8 Aug 92 to 30 Nov 92
1994 12 27 pc f439w ecr1410iw 80 Hi-Fi Flat, Cycle 2,
for data taken 30 Nov 92 to 3 Feb 93
1994 12 27 pc f439w ecr1318ew 80 Hi-Fi Flat, Cycle 2,
for data taken 3 Feb 93 to 1 Aug 93
1994 12 27 pc f547m ecr1424gw 80 Hi-Fi Flat, Cycle 2
1994 12 27 pc f555w ecr1424kw 26 Hi-Fi Flat, Cycle 2
1994 12 27 pc f569w ecr1105lw 35 Hi-Fi Flat, Cycle 2
1994 12 27 pc f606w ecr1105ow 20 Hi-Fi Flat, Cycle 2
1994 12 27 pc f622w ecr1105rw 35 Hi-Fi Flat, Cycle 2
1994 12 28 pc f648m ecr1512nw 120 Hi-Fi Flat, Cycle 2
1994 12 27 pc f675w ecr1439pw 40 Hi-Fi Flat, Cycle 2
1994 12 27 pc f702w ecr1439sw 26 Hi-Fi Flat, Cycle 2
1994 12 28 pc f718m ecr1504aw 140 Hi-Fi Flat, Cycle 2
1994 12 28 pc f725lp ecr1512qw 26 Hi-Fi Flat, Cycle 2
1994 12 27 pc f785lp ecr14401w 30 Hi-Fi Flat, Cycle 2
1994 12 28 pc f791w ecr1504cw 40 Hi-Fi Flat, Cycle 2
1994 12 28 pc f814w ecr1512tw 26 Hi-Fi Flat, Cycle 2
1994 12 28 pc f875m ecr1504ew 80 Hi-Fi Flat, Cycle 2
Optimized for Cycle 3 (>3 August 1993), long exposures (>1 sec.):
1994 06 14 wf f336w e6e0944pw 30 Cycle 3 flat field
1994 06 13 wf f569w e6d10291w 8 Cycle 3 flat field
1994 07 05 wf f785lp e751348hw 6 Cycle 3 flat field
1994 07 05 wf f850lp e751348jw 6 Cycle 3 flat field
1994 12 ?? pc f336w pc336l* >200 Hi-Fi Flat, Cycle 3, Type 3
1994 12 15 pc f368m ecg1245iw 300 Hi-Fi Flat, Cycle 3, Type 2
1994 12 15 pc f413m ecg1245mw 200 Hi-Fi Flat, Cycle 3, Type 2
1994 12 27 pc f413m ecr1431iw 200 Hi-Fi Flat, Cycle 3, Type 2;
improved header, no cal impact
1994 12 15 pc f439w ecg1245rw 80 Hi-Fi Flat, Cycle 3, Type 2
1994 12 27 pc f439w ecr1431ow 80 Hi-Fi Flat, Cycle 3, Type 2;
improved header, no cal impact
1994 12 01 pc f492m ec115135w 100 Hi-Fi Flat, Cycle 3, Type 1
1994 12 27 pc f492m ecr10086w 100 Hi-Fi Flat, Cycle 3, Type 1;
improved version
1994 12 01 pc f517n ec115139w 500 Hi-Fi Flat, Cycle 3, Type 1
1994 12 27 pc f517n ecr1008dw 500 Hi-Fi Flat, Cycle 3, Type 1;
improved version
1994 12 01 pc f547m ec11513cw 80 Hi-Fi Flat, Cycle 3, Type 1
1994 12 28 pc f547m ecs11228w 80 Hi-Fi Flat, Cycle 3, Type 1;
header update, no cal impact
1994 07 05 pc f555w e751348aw 26 Cycle 3 flat field
1994 12 01 pc f555w ec11513iw 26 Hi-Fi Flat, Cycle 3, Type 1;
improved version
1994 12 28 pc f555w ecs1122kw 26 Hi-Fi Flat, Cycle 3, Type 1;
improved version; header update
1994 12 01 pc f569w ec11513mw 35 Hi-Fi Flat, Cycle 3, Type 1
1994 12 28 pc f569w ecs11230w 35 Hi-Fi Flat, Cycle 3, Type 1;
header update, no cal impact
1994 12 01 pc f606w ec115047w 20 Hi-Fi Flat, Cycle 3, Type 1
1994 12 28 pc f606w ecs1111pw 20 Hi-Fi Flat, Cycle 3, Type 1;
header update, no cal impact
1994 12 01 pc f622w ec11504dw 35 Hi-Fi Flat, Cycle 3, Type 1
1994 12 30 pc f622w ecs11124w 35 Hi-Fi Flat, Cycle 3, Type 1;
header update, no cal impact
1994 12 01 pc f648m ec11504iw 120 Hi-Fi Flat, Cycle 3, Type 1
1994 12 30 pc f648m ecs1112cw 120 Hi-Fi Flat, Cycle 3, Type 1;
header update, no cal impact
1994 12 01 pc f664n ec11504ow 500 Hi-Fi Flat, Cycle 3, Type 1
1994 12 28 pc f664n ecr1008hw 500 Hi-Fi Flat, Cycle 3, Type 1;
improved version
1994 12 ?? pc f673n pc673l* >200 Hi-Fi Flat, Cycle 3, Type 3
1994 12 01 pc f675w ec11504tw 40 Hi-Fi Flat, Cycle 3, Type 1
1994 12 30 pc f675w ecs1112kw 40 Hi-Fi Flat, Cycle 3, Type 1;
header update, no cal impact
1994 12 01 pc f702w ec11452ew 26 Hi-Fi Flat, Cycle 3, Type 1
1994 12 28 pc f702w ecs10502w 26 Hi-Fi Flat, Cycle 3, Type 1;
header update, no cal impact
1994 12 01 pc f718m ec11452iw 140 Hi-Fi Flat, Cycle 3, Type 1
1994 12 28 pc f718m ecs1050ew 140 Hi-Fi Flat, Cycle 3, Type 1;
header update, no cal impact
1994 12 01 pc f725lp ec11452mw 26 Hi-Fi Flat, Cycle 3, Type 1
1994 12 28 pc f725lp ecs1050lw 26 Hi-Fi Flat, Cycle 3, Type 1;
header update, no cal impact
1994 12 01 pc f785lp ec11452qw 30 Hi-Fi Flat, Cycle 3, Type 1
1994 12 28 pc f785lp ecs1050rw 30 Hi-Fi Flat, Cycle 3, Type 1;
header update, no cal impact
1994 12 01 pc f791w ec114530w 40 Hi-Fi Flat, Cycle 3, Type 1
1994 12 28 pc f791w ecs10513w 40 Hi-Fi Flat, Cycle 3, Type 1;
header update, no cal impact
1994 12 15 pc f814w ecg1245aw 26 Hi-Fi Flat, Cycle 3, Type 1
1994 12 15 pc f875m ecg1245ew 80 Hi-Fi Flat, Cycle 3, Type 1
Optimized for Cycle 3 (>3 August 1993), short exposures (<1 sec.):
1994 06 14 wf f336w e6e0944nw 0.11 Cycle 3 flat field
1994 06 14 wf f368m e6e0944rw 0.20 Cycle 3 flat field
1994 06 16 wf f375m e6g1340dw 0.50 Cycle 3 flat field
1994 06 13 wf f487n e6d1028aw 0.30 Cycle 3 flat field
1994 06 13 wf f502n e6d1028jw 0.30 Cycle 3 flat field
1994 06 24 wf f588w e6o1002cw 0.18 Cycle 3 flat field
1994 06 08 wf f673n e681058iw 0.12 Cycle 3 flat field
1994 06 10 wf f889n e6a1428aw 0.20 Cycle 3 flat field
1994 06 10 wf f1042m e6a1428iw 0.11 Cycle 3 flat field
1994 06 20 pc f336w e6k1017kw 0.35 Cycle 3 flat field
1994 06 20 pc f368m e6k1017nw 0.60 Cycle 3 flat field
1994 06 20 pc f413m e6k0945kw 0.23 Cycle 3 flat field
1994 06 20 pc f517n e6k09466w 0.35 Cycle 3 flat field
1994 06 13 pc f664n e6d16003w 0.26 Cycle 3 flat field
1994 06 15 pc f673n e6f1115hw 0.60 Cycle 3 flat field
1994 06 20 pc f1042m e6k1017iw 0.80 Cycle 3 flat field
Table Notes:
Contact J. Biretta for copies of these files; not yet archived as of 03 Jan. 1995.
Comments:
Type 1 = derived roughly as flat_filter = flat_filter+f122m *
[patch, smooth(flat_f555w+f8nd / flat_f555w+f122m)].
See above for detailed discussion.
Type 2 = derived by correcting f8nd reflection pattern.
Type 3 = derived by correcting short exposure reciprocity error.
V. Errata:
All known errors are corrected in the most recent versions of the High-Fidelity flats.
VI. References:
Baggett and MacKenty, 1993. "WFPC Deltaflat Corrections" STEIS Memo.
Biretta, et al., 1994. "WFPC Flat Field Calibration: Flats and DeltaFlats" in "Calibrating Hubble Space Telescope," eds. Blades and Osmer, p. 37.
Hester 1992, in "WFPC Final Orbital/Science Verification Report," ed. Faber.
MacKenty and Baggett, 1992. "Measles Contamination and Compensation with Delta Flat Fields," WF/PC Instrument Science Report 92-04.