WFC3 SMOV Program 11433: IR Internal Flat Field Observations

Instrument Science Report WFC3 2009-42 WFC3 SMOV Program 11433: IR Internal Flat Field Observations B. Hilbert 27 October 2009 ABSTRACT We have analyzed the internal flat field behavior of the WFC3/IR channel using data obtained during the Servicing Mission Observatory Verification (SMOV) for Program 11433. The detector was illuminated to a high signal level through each filter using the Tungsten 2 lamp. Comparison of the resulting flat field images to those collected during ground testing reveals similar illumination patterns, and an increase in the Tungsten lamp flux of 6-10%. Signal rates show that the lamp s irradiance was constant to within 0.5% during these observations. These data show that the on-orbit internal flats can be used to update the calibration flat field images. Roughly 30% of the observations associated with this proposal were contaminated by a light leak while HST was pointed at the Earth. There is currently no strategy aimed at mitigating the effects of this light leak. Introduction Testing of the cooled WFC3/IR detector began in mid-june 2009, after installation of WFC3 in HST during the May 2009 servicing mission. Program 11433, IR Internal Flat Field Observations executed between July 12 and 21, 2009. All data were obtained with in full-frame mode with MEB2 as usual, using Science Mission Specifications (SMSs), i.e. no real time commanding. The temperature of the IR detector was -127.9 C (145.1 K) and constant to within 0.2 o C throughout the observations. All data were obtained at the 1

nominal 2.5 e - /ADU setting. Gain values used for subsequent data reduction were derived from SMOV data taken as part of the IR Functional Test (proposal 11420). Details are given in Hilbert and McCullough (2009b). Gain values for the four quadrants (following the usual convention of quadrant 1 in the upper left, with quad numbers increasing in the counter-clockwise direction) are 2.33, 2.28, 2.45, and 2.47 e - /ADU. Data Each of the 50 Visits in this proposal collected flat field data through one or more filters. The program as a whole acquired data through all 15 IR filters. Tungsten lamp #2 was used as the illumination source for all ramps. The exposure time for each filter was set to allow each pixel to collect a total of 60,000 65,000 e -. Table 4 in Appendix 1 lists all the files obtained for this program. All data were reduced using CALWF3 version 1.4.1. Basic data reduction steps included reference pixel subtraction, zeroth read subtraction, cosmic ray rejection, and non-linearity correction. We made one change to the default behavior of CALWF3 when processing these data: we set the UNITCORR keyword to PERFORM rather than the default value of OMIT, which is normally used to process dark current ramps.. This causes the output *flt.fits files to be in units of ADU/sec and have values that are normalized to the appropriate exposure time for each pixel. If UNITCORR is set to OMIT the *flt.fits files are output in units of ADU and each pixel contains the signal value of the final unrejected read. This leads to an image where each pixel s exposure time is different, which is not useful for analysis purposes. Tables 5 and 6 in Appendix 1 list the reference files used by CALWF3 for its processing steps. As with the IR dark current data (Hilbert and McCullough, 2009a), some of the images were affected by a light leak in the IR channel. The details of the light leak are discussed below. Affected files are identified in Table 4, and were not used for this analysis. Other files, not predicted to contain a light leak, had a non-nominal appearance suggestive of an extraneous light source. These files are also identified in Table 4 and were excluded from further analysis. SMOV internal flats appear qualitatively identical to ground flats, as can be seen in Figures 1 and 2. An uncontaminated on-orbit flat field image is shown in Figure 1, and comparable ground flat field image shown in Figure 2. The average brightness is higher in the on-orbit flat compared to that from the ground. This effect is discussed below. Figures 3 and 4 show the same flat field images with a histogram equalization stretch. Again, the flat fields appear qualitatively very similar. There is however a small change in the illumination pattern between the two. This change is most visible in the upper right quadrant, where the size of the black region in the corner was smaller during ground testing than in SMOV. This slight (~20-pixel) leftward shift in the illumination pattern is due to an update of the IR corrector position made after launch. Ratios of SMOV to TV3 2

flat field images, shown in the Analysis section below, highlight the changes in the illumination around the edges of the detector. Figure 1: F140W flat field. Figure is a linear stretch from 700 to 2100 e-/sec. Note the familiar features of the wagon wheel in the lower right corner, as well as the death star along the bottom edge. Figure 2: F140W flat field image taken during TV3 testing. The stretch is identical to that in the figure at left. Note that various features in the flat are identical between ground testing and SMOV, but that the flat taken during SMOV is brighter overall. This is due to an on-orbit brightening of the tungsten lamp, discussed below. Figure 3: Histogram equalization stretch of the SMOV flat field image shown above. Figure 4: Histogram equalization stretch of the TV3 flat field shown above. Note the slight shift of the illumination pattern to the left when compared to the image at left. The shift is most noticeable in the upper right corner, where the black region above is smaller than in the SMOV figure at left. 3

Analysis Light Leak Figures 5 and 6 show examples of the light leak in IR flat fields through the F105W and F139M filters. These images are both ratios of a single light leak-contaminated flat field to a median uncontaminated flat field, in order to accentuate the appearance of the extra light. Images such as these indicate that the light leak contributes an extra 1% - 4% above the nominal Tungsten lamp flux in the affected areas. The morphology of the light leak was filter-dependent. See Appendix 2 for a gallery of light leak images through all filters. Figure 5: Ratio image showing light leak through the F105W filter. Figure 6: Ratio image showing light leak through the F139M filter. The source of the light leak in these data appears to be similar to that identified in the dark current data from proposals 11447 and 11420 (Hilbert and McCullough, 2009a, 2009b). When HST is pointed at the day side of the Earth, a significant number of photons find their way to the IR detector. The light leak onto the dark current images was stopped by placing the channel select mechanism (CSM) in the UVIS position, thereby directing any extraneous light toward the UVIS detector. However, when IR internal flats are taken, the CSM is already close to the UVIS position. This is because the diffuser paddle, which directs the Tungsten lamp light onto the IR detector, is located on the same mechanism as the CSM and must be in the IR beam to direct tungsten lamp 4

illumination onto the IR detector. See Figure 7 for a drawing of the relative positions of the CSM, diffuser paddle, and IR detector. Figure 7: Schematic optical layout of WFC3 from its Instrument Handbook. By placing the channel select mechanism very close to the UVIS position, the diffuser paddle enters the IR channel light path and redirects the illumination from the Tungsten lamp onto the IR detector. As with the IR dark current data, we used our knowledge of HST s orientation at the beginning of each exposure to predict whether that observation was potentially contaminated by a light leak due to pointing at the day side of the Earth. See Figure 6 in Hilbert and McCullough (2009b) for an example of the bright Earth plot used to predict the HST s orientation over the course of an exposure. While light leaks were present in almost all of the images predicted by the bright Earth plots, we also found other images with non-nominal appearances. These images are identified in Table 4. According to the bright Earth plots, HST was not pointed at the day side of the Earth for these exposures. However, when compared to uncontaminated files, these images displayed abnormally bright or dark areas on the detector. The source of the light leak remains unidentified. Figures 8 and 9 show two examples. These contaminated images were also ignored in subsequent analyses. Figure 8 shows an F140W flat field with an unexpected appearance. The upper left and right corners of the detector appear darker than expected and the wagon wheel in the lower right corner shows an extended, darker area around a brighter area. The white streak on the left side of the detector is persistence from an earlier observation. Figure 9 shows an even stranger flat field image ratio. In this case, using the F127M filter, we see a dark area extending from the upper left corner across most of the top half 5

of the detector, while the bottom edge of the detector appears brighter than normal. The wagon wheel in this case appears as a bright area extending around a darker core, opposite to that in the F140W image. Also, the dark vertical streak on the right side of the detector is very prominent. For both images, HST was supposedly over twilight or the night side of the Earth and pointed away from the Earth, suggesting that contamination from a light leak should not occur. Light leak or non-nominal appearance suggestive of light leak affected roughly 30% of the ramps in this proposal. At the moment, there is no mitigation strategy for future observations, neither through scheduling nor a mechanical fix such as that used to fix the IR dark current light leak (Hilbert and McCullough 2009a). Figure 8: Ratio image of an F140W flat field to the mean F140W flat field. This image was taken while HST was above an area of the Earth in twilight and pointed away from Earth. The image is a histogram equalization stretch between 0.9 (black) and 1.1 (white). Note the slight darkening of the upper left and upper right. The white streak on the left is a persistence image from a previous observation. The wagon wheel also extends far out into quadrant 3. Figure 9: Ratio image of a single F127M flat field to the mean F127M flat field. This image was taken while HST was above the night side of the Earth, and pointed away from the Earth. The image is a histogram equalization stretch between 0.9 (black) and 1.1 (white). Note the opposite appearance of the wagon wheel (lower right corner) in this image compared to that in the figure at left, as well as the much deeper darkening of the upper half of the detector. Other features not usually visible in ratio images are also apparent. 6

Bias Level Our first analysis task was to monitor the bias level of all ramps, in order to search for any patterns. We obtained a measure of the bias level of each ramp by calculating the sigma-clipped mean in the reference pixels of the zeroth read of each ramp. This was done prior to any CALWF3 processing (i.e. on raw ramps). Since we were working only with reference pixels, which are not light sensitive and should not be affected by light leak, we used all 215 ramps for this analysis. Figure 10 shows a plot of this mean bias level in quadrant 1 for all ramps. Visits are plotted in alternating blue and red points, for clarity. The mean bias level in quadrant 1 changes by roughly 25 electrons over the course of the observations. Figure 11 shows the mean bias level over four different Visits. Similar to what was seen in the dark current analysis (Hilbert and McCullough, 2009b), there is a repeating trend of bias level decreasing over the course of each Visit. We also see that the bias variation is repeatable. Visits 1 and 15 in Figure 11 (blue) were separated by 27 hours, and contain an identical sequence of six ramps. The exposure times for these six ramps, in order, are 43, 53, 903, 43 and 203 seconds. In Figure 11 we see the largest decrease in bias level after the third ramp, with its 903-second exposure time. Visits 8 and 22 (red), separated by 64 hours, are also identical to one another, and contain the same six ramps as Visits 1 and 15, but in a different order. Once again, the largest decrease in mean bias level comes after the third ramp, which also has an exposure time of 903 seconds. Figure 10: Mean bias level for all ramps. Visits are plotted in alternating colors for clarity. Figure 11: Mean signal in the zeroth read reference pixels of quadrant 1 for four Visits. In terms of the filters and sample sequences used Visits 1 and 15 were identical to one another. Visits 8 and 22 were also identical to one another. Note the similar pattern in bias level within the identical pairs of Visits. 7

Investigating this trend further, we produced the plot shown in Figure 12. This shows the mean bias level in each ramp, plotted against the exposure time of the immediately preceding ramp. There is the suggestion of a downward trend, shown by the red, best-fit line produced with the IDL Astronomy Library s robust_linefit.pro. This plot, along with those in Figures 10 and 11, suggests that the exposure time and sample sequence used for a particular ramp can affect the bias level in the following exposure. Generally, shorter exposure times will lead to higher bias levels. As the same number of detector readouts is packed into a shorter exposure time, the FPA or the detector electronics heat up, which causes an increase in the mean bias level of the detector. The reason that the bias level is high at the beginning of each Visit is because the IR detector is in auto-flush mode prior to the start of the first exposure. This continually removes accumulating charge from the detector, and is a similar operation to reading out the detector, except that the signal is not saved. This action heats the electronics, resulting in a higher bias level. Conversely, during a 903-second ramp, the detector electronics have time to cool, resulting in a lower bias level at the beginning of the next exposure. A similar effect has been observed in NICMOS data, to the point that bias levels in NICMOS data can be used to calculate the detector temperatures (Pirzkal, 2009). Further investigation may reveal that a similar analysis can be used to calculate the temperature of the WFC3-IR detector. Figure 12: Mean bias level versus exposure time of the immediately preceding ramp. The hint of a trend suggests that heating of the FPA or electronics associated with reading out the array increases the bias level of the detector. Unlike NICMOS, the WFC3 IR detector has reference pixels which track the bias level in each read of the detector. Using these reference pixels, CALWF3 corrects each 8

read of the WFC3 IR detector for the changing bias level. This implies that while the varying bias level presented above is an interesting detector effect, it should not affect science observations as long as reference pixel subtraction is performed, and will in fact be invisible to users unless the raw files are examined. Signal Rate After identifying the non-nominal flat field images noted in Table 4, our next task was to create a mean flat field for each filter. Using the processed uncontaminated data, we calculated the sigma-clipped mean value on a pixel-by-pixel basis in order to create the mean flat field images. Table 1 lists the sigma-clipped mean signal rate in each mean flat field. In order to compare these signal rates to those measured on the ground we calculated the ratio of the SMOV median flat field to that from TV3 for each filter. We then created a histogram of the ratio for all pixels and fit a Gaussian to the distribution in order to locate the peak ratio value. The peak values are shown in the right-most column of Table 1 and indicate 6% - 10% higher signal rates in SMOV than during TV3. This implies an on-orbit brightening of the Tungsten 2 lamp relative to the ground, in agreement with results from UVIS flat field data (Baggett et al, 2009). Filter Mean Signal Ratio to TV3 (e - /sec/pixel) F098M 314 1.103 F105W 725 1.083 F110W 1549 1.086 F125W 1149 1.070 F126N 54 1.075 F127M 285 1.080 F128N 63 1.075 F130N 66 1.072 F132N 65 1.069 F139M 252 1.069 F140W 1649 1.074 F153M 287 1.071 F160W 1146 1.078 F164N 70 1.080 F167N 73 1.061 Table 1: Robust mean signal levels from mean uncontaminated flat field image in each filter, along with a comparison of this mean level to that measured in TV3. 9

Given the change in lamp brightness between the time of ground testing and SMOV, we also searched for variations in the lamp brightness over shorter timescales. For each filter, we calculated the 3-sigma clipped mean signal rate in each of the individual flt files used to create the mean flat fields. Figures 13 and 14 show the mean measured signal rate in all flt files for F105W and F139M. The black dashed line shows the mean signal level. The scatter in these plots is typical of that for all filters. Figure 13: Variation in lamp brightness from the mean for all F105W observations. Figure 14: Variation in lamp brightness from the mean for all F139M observations. From the calculated mean signal in individual flats, we see no overall trend in lamp light with time. In all but one image of the 215-image dataset, the variation in measured signal rate is less than 1% of the mean rate. Table 2 lists the standard deviation of the signal variations in each filter, along with the maximum signal variation above and below the mean, in units of percentage of the mean signal. For example, from Figure 13, we see that the maximum deviations from the mean signal are about 0.4% above the mean, and 0.5% below the mean. The standard deviation of all signal levels shown in Figure 13 is 0.3%. As can be seen in Table 2, in general and especially in the wide band filters, the larger deviations from the mean signal levels were negative, i.e. less illumination than the mean. Sigma-clipped mean signal rates for each file were calculated using all science pixels on the detector. This implies a formal uncertainty on the mean values of ~0.001% from Poisson noise and read noise. Therefore, the variations shown in Table 2 should represent true changes in the tungsten lamp output. These values show that the lamp output flux was constant to within 1% for the duration of these observations. Photometric uncertainties in the IR channel will therefore be dominated by other sources. 10

Filter Sigma clipped stdev (%) Max Variation Below Mean(%) Max Variation Above Mean(%) F098M 0.2 0.2 0.2 F105W 0.3 0.5 0.4 F110W 0.1 1.1 0.2 F125W 0.2 0.6 0.2 F126N 0.1 0.1 0.1 F127M 0.2 0.7 0.3 F128N 0.3 0.3 0.5 F130N 0.1 0.04 0.3 F132N 0.2 0.6 0.2 F139M 0.3 0.4 0.3 F140W 0.4 0.9 0.2 F153M 0.1 0.1 0.2 F160W 0.2 0.8 0.2 F164N 0.3 0.4 0.3 F167N 0.4 0.5 0.4 Table 2: Variations in lamp brightness relative to the mean signal level for all filters. The second column shows the sigma-clipped standard deviation of the lamp brightness relative to the mean signal rate. The third and fourth columns list the most extreme variation in lamp brightness below and above the mean rate, respectively. These values are also presented relative to the mean signal level of the lamp for each filter s observations. Comparison to TV3 Internal Flats As seen in Table 1, the mean illumination level from the tungsten lamp increased between TV3 and SMOV. Beyond this change, we also wished to search for any other changes in illumination characteristics. Figures 15 and 16 show ratio images of mean SMOV to TV3 internal flat fields for F098M and F153M respectively. Visible features in these ratio images include dark upper and lower right corners, and bright upper and lower left corners. These are due to the change in the IR corrector position, as discussed in the Data section above. This change shifted the illumination pattern from the tungsten lamp to the left on the detector by ~20 pixels, as evidenced by the dark corners on the right side of the ratio image, and the bright corners on the left. The small collection of dark pixels in the center of the top edge of the ratio images contains unstable, low-qe and dead pixels. Small changes in the illumination level between TV3 and SMOV therefore result in large fractional changes in measured signal, causing these pixels to appear dark in the ratio images. 11

Figure 15: Ratio of F098M, Visit 11 image to mean TV3 F098M flat. Range is 1.033 to 1.125 (black to white). Figure 16: Ratio of F153M, Visit 15 image to mean TV3 F153M flat. Range is 1.001 to 1.095 (black to white). Ignoring the large brightness differences in the corners, the mottling in the image represents fluctuations in the ratio of between 1% and 2% (~1.06 to ~1.08). Figure 17: Histogram of the ratio image shown in Figure 15. Figure 18: Histogram of the ratio image shown in Figure 16. 12

Figures 17 and 18 show histograms of the ratio images presented in Figures 14 and 15, respectively. The peak values of the histograms correspond to the values shown in Table 1. The general shape of the histograms was the same for all filters. Table 3 lists the range of values observed in the ratio of SMOV flats to the mean internal flat fields collected in TV3 testing. Min and Max values are the ratio values where the histogram (such as that shown in Figure 17) falls to 1% of the maximum level. Filter Min Mean Max F098M 1.03 1.10 1.13 F105W 1.01 1.08 1.11 F110W 1.01 1.09 1.11 F125W 1.00 1.07 1.09 F126N 0.99 1.08 1.10 F127M 1.00 1.08 1.10 F128N 0.99 1.08 1.10 F130N 1.00 1.07 1.10 F132N 1.00 1.07 1.09 F139M 0.99 1.07 1.09 F140W 1.00 1.07 1.10 F153M 1.00 1.07 1.09 F160W 1.00 1.08 1.10 F164N 1.00 1.08 1.11 F167N 0.99 1.06 1.09 Table 3: Ratio values of SMOV internal flat fields versus the mean internal flat fields in TV3. Min and max values are the ratio values where the histogram falls to 1% of its maximum height. Bad Pixels With the detector well illuminated by the tungsten lamp, we looked for dead pixels. For this, we used the mean flat field images created for all filters. We searched all 15 mean flat fields, and marked as dead only those pixels with a signal rate less than or equal to 0 e-/sec in every flat field. From this search, we found 49 dead pixels on the IR detector. Repeating the search on a subset of internal flat fields from TV3 revealed the same 49 pixels as dead. Conclusions The IR Channel in WFC3 during SMOV performs essentially as it did during TV3. The internal flat field illumination pattern is very close to that observed in ground testing, with the exception that the tungsten lamp has increased in brightness by 6-10%, depending on the filter used in the observation. The dead pixel population also appears stable. Despite the observing efficiency decrease due to light leak, IR channel internal 13

flat field observations are suitable for generating delta flat field calibration files, should it be found that the calibration flat field images derived from ground testing are in need of modification. As was seen in SMOV dark current data, the bias level of the IR channel changes depending on observation history. However, the reference pixels successfully remove bias level variations. Unresolved Issues The IR channel does exhibit a light leak when taking flat field images and pointed at the day side of the Earth. During Cycle 17, the CSM will be placed in the UVIS position when taking darks, but currently, there is no mitigation strategy for light leaks contaminating flat fields. Future WFC3 calibration programs that collect internal flat field data should expect to have up to 30% of their flat fields contaminated by the light leak. References Baggett, S., E. Sabbi, P. McCullough, 2009. WFC3 SMOV Program 11422: UVIS SOFA and Lamp Checks. WFC3 ISR 2009-27. http://www.stsci.edu/hst/wfc3/documents/isrs/wfc3-2009-27.pdf Nov 2009. Hilbert, B. and P. McCullough, 2009a. WFC3 SMOV Program 11447: IR Channel Dark Current, Readnoise and Background Signal. WFC3 ISR 2009-21. http://www.stsci.edu/hst/wfc3/documents/isrs/wfc3-2009-21.pdf Nov 2009. Hilbert, B., and P. McCullough, 2009b WFC3 SMOV Program 11420: IR Channel Functional Tests. WFC3 ISR 2009-23. http://www.stsci.edu/hst/wfc3/documents/isrs/wfc3-2009-23.pdf Nov 2009. Pirzkal, N., 2009. NICMOS ISR 2009-007: CalTempFromBias: Implementation and Testing. NICMOS ISR 2009-07. http://www.stsci.edu/hst/nicmos/documents/isrs/isr_2009_007.pdf Nov 2009. 14

Appendix 1 Visit Filename Filter Sample Sequence / Exposure Light Leak # Reads Time (sec) 1 iaak01agq F110W RAPID / 16 43.98 1 iaak01ahq F125W SPARS10 / 7 52.94 1 iaak01aiq F130N SPARS100 / 11 902.94 Not Predicted 1 iaak01akq F110W RAPID / 16 43.98 1 iaak01alq F153M SPARS50 / 6 202.93 1 iaak01amq F153M SPARS50 / 6 202.93 2 iaak02giq F140W RAPID / 14 38.12 2 iaak02gjq F125W SPARS10 / 7 52.94 2 iaak02gkq F130N SPARS100 / 11 902.94 2 iaak02gmq F140W RAPID / 14 38.12 2 iaak02gnq F139M SPARS50 / 7 252.93 2 iaak02goq F139M SPARS50 / 7 252.93 3 iaak03idq F160W SPARS10 / 7 52.94 3 iaak03ieq F160W SPARS10 / 7 52.94 3 iaak03ifq F128N SPARS100 / 12 1002.94 3 iaak03ihq F127M SPARS50 / 7 252.93 3 iaak03iiq F127M SPARS50 / 7 252.93 4 iaak04i6q F105W SPARS10 / 10 82.94 4 iaak04i7q F105W SPARS10 / 10 82.94 4 iaak04i8q F128N SPARS100 / 12 1002.94 Predicted 4 iaak04iaq F098M SPARS50 / 6 202.93 4 iaak04ibq F098M SPARS50 / 6 202.93 Predicted 5 iaak05ikq F125W SPARS10 / 7 52.94 5 iaak05ilq F105W SPARS10 / 10 82.94 5 iaak05imq F160W SPARS10 / 7 52.94 5 iaak05inq F126N SPARS100 / 13 1102.94 5 iaak05iqq F140W RAPID / 14 38.12 5 iaak05irq F110W RAPID / 16 43.98 6 iaak06izq F139M SPARS50 / 7 252.93 6 iaak06j0q F132N SPARS100 / 12 1002.94 6 iaak06j1q F127M SPARS50 / 7 252.93 7 iaak07j9q F167N SPARS100 / 10 802.93 7 iaak07jaq F164N SPARS100 / 11 902.94 8 iaak08jeq F125W SPARS10 / 7 52.94 8 iaak08jfq F110W RAPID / 16 43.98 8 iaak08jgq F130N SPARS100 / 11 902.94 Not Predicted 8 iaak08jiq F153M SPARS50 / 6 202.93 8 iaak08jjq F110W RAPID / 16 43.98 Predicted 15

8 iaak08jkq F153M SPARS50 / 6 202.93 Predicted 9 iaak09joq F125W SPARS10 / 7 52.94 9 iaak09jpq F140W RAPID / 14 38.12 9 iaak09jqq F130N SPARS100 / 11 902.94 9 iaak09jsq F139M SPARS50 / 7 252.93 9 iaak09jtq F140W RAPID / 14 38.12 9 iaak09juq F139M SPARS50 / 7 252.93 10 iaak10kgq F127M SPARS50 / 7 252.93 10 iaak10khq F160W SPARS10 / 7 52.94 10 iaak10kiq F128N SPARS100 / 12 1002.94 Predicted 10 iaak10kkq F127M SPARS50 / 7 252.93 10 iaak10klq F160W SPARS10 / 7 52.94 11 iaak11kvq F105W SPARS10 / 10 82.94 11 iaak11kwq F098M SPARS50 / 6 202.93 11 iaak11kxq F128N SPARS100 / 12 1002.94 11 iaak11kzq F105W SPARS10 / 10 82.94 11 iaak11l0q F098M SPARS50 / 6 202.93 12 iaak12lxq F105W SPARS10 / 10 82.94 Predicted 12 iaak12lyq F110W RAPID / 16 43.98 Predicted 12 iaak12lzq F126N SPARS100 / 13 1102.94 Predicted 12 iaak12m2q F140W RAPID / 14 38.12 12 iaak12m3q F160W SPARS10 / 7 52.94 12 iaak12m4q F125W SPARS10 / 7 52.94 13 iaak13jyq F127M SPARS50 / 7 252.93 13 iaak13jzq F132N SPARS100 / 12 1002.94 13 iaak13k0q F139M SPARS50 / 7 252.93 Predicted 14 iaak14lmq F164N SPARS100 / 11 902.94 14 iaak14lnq F167N SPARS100 / 10 802.93 15 iaak15mgq F110W RAPID / 16 43.98 15 iaak15mhq F125W SPARS10 / 7 52.94 15 iaak15miq F130N SPARS100 / 11 902.94 Predicted 15 iaak15mkq F110W RAPID / 16 43.98 Predicted 15 iaak15mlq F153M SPARS50 / 6 202.93 Predicted 15 iaak15mmq F153M SPARS50 / 6 202.93 Predicted 16 iaak16n1q F140W RAPID / 14 38.12 16 iaak16n2q F125W SPARS10 / 7 52.94 16 iaak16n3q F130N SPARS100 / 11 902.94 Predicted 16 iaak16n5q F140W RAPID / 14 38.12 Predicted 16 iaak16n6q F139M SPARS50 / 7 252.93 Predicted 16 iaak16n7q F139M SPARS50 / 7 252.93 Predicted 17 iaak17ngq F160W SPARS10 / 7 52.94 17 iaak17nhq F160W SPARS10 / 7 52.94 17 iaak17niq F128N SPARS100 / 12 1002.94 17 iaak17nkq F127M SPARS50 / 7 252.93 17 iaak17nlq F127M SPARS50 / 7 252.93 18 iaak18nrq F105W SPARS10 / 10 82.94 Predicted 18 iaak18nsq F105W SPARS10 / 10 82.94 Predicted 16

18 iaak18ntq F128N SPARS100 / 12 1002.94 Predicted 18 iaak18nvq F098M SPARS50 / 6 202.93 Predicted 18 iaak18nwq F098M SPARS50 / 6 202.93 19 iaak19h0q F125W SPARS10 / 7 52.94 Not Predicted 19 iaak19h3q F126N SPARS100 / 13 1102.94 Predicted 19 iaak19h6q F140W RAPID / 14 38.12 19 iaak19h7q F110W RAPID / 16 43.98 Predicted 20 iaak20t9q F139M SPARS50 / 7 252.93 20 iaak20taq F132N SPARS100 / 12 1002.94 20 iaak20tbq F127M SPARS50 / 7 252.93 Predicted 21 iaak21dqq F167N SPARS100 / 10 802.93 Predicted 21 iaak21drq F164N SPARS100 / 11 902.94 Predicted 22 iaak22ehq F125W SPARS10 / 7 52.94 Not Predicted 22 iaak22eiq F110W RAPID / 16 43.98 22 iaak22ejq F130N SPARS100 / 11 902.94 22 iaak22elq F153M SPARS50 / 6 202.93 22 iaak22emq F110W RAPID / 16 43.98 22 iaak22enq F153M SPARS50 / 6 202.93 23 iaak23evq F125W SPARS10 / 7 52.94 Predicted 23 iaak23ewq F140W RAPID / 14 38.12 23 iaak23exq F130N SPARS100 / 11 902.94 23 iaak23ezq F139M SPARS50 / 7 252.93 23 iaak23f0q F140W RAPID / 14 38.12 23 iaak23f1q F139M SPARS50 / 7 252.93 24 iaak24g5q F127M SPARS50 / 7 252.93 24 iaak24g6q F160W SPARS10 / 7 52.94 24 iaak24g7q F128N SPARS100 / 12 1002.94 24 iaak24g9q F127M SPARS50 / 7 252.93 24 iaak24gaq F160W SPARS10 / 7 52.94 25 iaak25f4q F105W SPARS10 / 10 82.94 25 iaak25f5q F098M SPARS50 / 6 202.93 25 iaak25f6q F128N SPARS100 / 12 1002.94 25 iaak25f8q F105W SPARS10 / 10 82.94 25 iaak25f9q F098M SPARS50 / 6 202.93 26 iaak26ffq F105W SPARS10 / 10 82.94 26 iaak26fgq F110W RAPID / 16 43.98 26 iaak26fhq F126N SPARS100 / 13 1102.94 26 iaak26fkq F140W RAPID / 14 38.12 26 iaak26flq F160W SPARS10 / 7 52.94 26 iaak26fmq F125W SPARS10 / 7 52.94 27 iaak27fpq F127M SPARS50 / 7 252.93 Not Predicted 27 iaak27fqq F132N SPARS100 / 12 1002.94 27 iaak27frq F139M SPARS50 / 7 252.93 28 iaak28fyq F164N SPARS100 / 11 902.94 28 iaak28fzq F167N SPARS100 / 10 802.93 29 iaak29o4q F153M SPARS50 / 6 202.93 Not Predicted 29 iaak29oaq F110W RAPID / 16 43.98 17

29 iaak29obq F125W SPARS10 / 7 52.94 30 iaak30hgq F139M SPARS50 / 7 252.93 30 iaak30hhq F139M SPARS50 / 7 252.93 30 iaak30hjq F140W RAPID / 14 38.12 30 iaak30hkq F130N SPARS100 / 11 902.94 Predicted 30 iaak30hmq F125W SPARS10 / 7 52.94 30 iaak30hnq F140W RAPID / 14 38.12 31 iaak31omq F127M SPARS50 / 7 252.93 31 iaak31onq F127M SPARS50 / 7 252.93 31 iaak31ooq F128N SPARS100 / 12 1002.94 31 iaak31oqq F160W SPARS10 / 7 52.94 31 iaak31orq F160W SPARS10 / 7 52.94 32 iaak32otq F098M SPARS50 / 6 202.93 32 iaak32ouq F098M SPARS50 / 6 202.93 32 iaak32ovq F128N SPARS100 / 12 1002.94 Predicted 32 iaak32oxq F105W SPARS10 / 10 82.94 Predicted 32 iaak32oyq F105W SPARS10 / 10 82.94 Predicted 33 iaak33p0q F140W RAPID / 14 38.12 33 iaak33p1q F105W SPARS10 / 10 82.94 33 iaak33p2q F126N SPARS100 / 13 1102.94 33 iaak33p5q F160W SPARS10 / 7 52.94 33 iaak33p6q F110W RAPID / 16 43.98 33 iaak33p7q F125W SPARS10 / 7 52.94 34 iaak34niq F139M SPARS50 / 7 252.93 Not Predicted 34 iaak34njq F132N SPARS100 / 12 1002.94 34 iaak34nkq F127M SPARS50 / 7 252.93 35 iaak35pdq F167N SPARS100 / 10 802.93 35 iaak35peq F164N SPARS100 / 11 902.94 36 iaak36pgq F140W RAPID / 14 38.12 36 iaak36phq F105W SPARS10 / 10 82.94 36 iaak36piq F126N SPARS100 / 13 1102.94 Predicted 36 iaak36plq F160W SPARS10 / 7 52.94 Predicted 36 iaak36pmq F110W RAPID / 16 43.98 Predicted 36 iaak36pnq F125W SPARS10 / 7 52.94 Predicted 37 iaak37p9q F127M SPARS50 / 7 252.93 Not Predicted 37 iaak37paq F139M SPARS50 / 7 252.93 Predicted 37 iaak37pbq F132N SPARS100 / 12 1002.94 Predicted 38 iaak38ptq F167N SPARS100 / 10 802.93 38 iaak38puq F164N SPARS100 / 11 902.94 39 iaak39q0q F105W SPARS10 / 10 82.94 39 iaak39q1q F125W SPARS10 / 7 52.94 39 iaak39q2q F160W SPARS10 / 7 52.94 39 iaak39q3q F126N SPARS100 / 13 1102.94 39 iaak39q6q F140W RAPID / 14 38.12 39 iaak39q7q F110W RAPID / 16 43.98 40 iaak40qaq F127M SPARS50 / 7 252.93 Not Predicted 40 iaak40qbq F139M SPARS50 / 7 252.93 18

40 iaak40qcq F132N SPARS100 / 12 1002.94 41 iaak41qjq F164N SPARS100 / 11 902.94 Predicted 41 iaak41qkq F167N SPARS100 / 10 802.93 Predicted 42 iaak42ucq F140W RAPID / 14 38.12 Not Predicted 42 iaak42udq F105W SPARS10 / 10 82.94 42 iaak42ueq F126N SPARS100 / 13 1102.94 Predicted 42 iaak42uhq F160W SPARS10 / 7 52.94 Predicted 42 iaak42uiq F110W RAPID / 16 43.98 Predicted 42 iaak42ujq F125W SPARS10 / 7 52.94 Predicted 43 iaak43qmq F127M SPARS50 / 7 252.93 43 iaak43qnq F139M SPARS50 / 7 252.93 43 iaak43qoq F132N SPARS100 / 12 1002.94 44 iaak44v6q F167N SPARS100 / 10 802.93 44 iaak44v7q F164N SPARS100 / 11 902.94 45 iaak45veq F105W SPARS10 / 10 82.94 45 iaak45vfq F125W SPARS10 / 7 52.94 45 iaak45vgq F160W SPARS10 / 7 52.94 45 iaak45vhq F126N SPARS100 / 13 1102.94 45 iaak45vkq F140W RAPID / 14 38.12 45 iaak45vlq F110W RAPID / 16 43.98 Predicted 46 iaak46tsq F127M SPARS50 / 7 252.93 Not Predicted 46 iaak46ttq F139M SPARS50 / 7 252.93 46 iaak46tuq F132N SPARS100 / 12 1002.94 47 iaak47vnq F164N SPARS100 / 11 902.94 47 iaak47voq F167N SPARS100 / 10 802.93 48 iaak48vuq F140W RAPID / 14 38.12 48 iaak48vvq F105W SPARS10 / 10 82.94 48 iaak48vwq F126N SPARS100 / 13 1102.94 48 iaak48vzq F160W SPARS10 / 7 52.94 48 iaak48w0q F110W RAPID / 16 43.98 48 iaak48w1q F125W SPARS10 / 7 52.94 49 iaak49hvq F127M SPARS50 / 7 252.93 Not Predicted 49 iaak49hwq F139M SPARS50 / 7 252.93 49 iaak49hxq F132N SPARS100 / 12 1002.94 Not Predicted 50 iaak50l5q F167N SPARS100 / 10 802.93 Predicted 50 iaak50l6q F164N SPARS100 / 11 902.94 Predicted Table 4: Data table for 11433 flat fields. Files labeled as Predicted in the right-most column were contaminated by a light leak that was predicted based on HST s orientation. Files labeled with Not Predicted were also contaminated by light leak despite information on HST s orientation suggesting that they would be uncontaminated. Both Predicted and Not Predicted light leak files were excluded from data analyses. 19

Tables 5 and 6 list the reference files used by CALWF3 to process the data collected in proposal number 11433. Table 5: Reference files used for all ramps collected in this proposal. Darkfile names are listed in the table below, as we have a different dark current calibration file for each sample sequence. Files listed in both of these tables can be downloaded from the HST archive. File Filename Purpose BPIXTAB t291659ni_bpx.fits Bad pixel table CCDTAB t2c16200i_ccd.fits Detector calibration parameters OSCNTAB q911321mi_osc.fits Detector overscan table CRREJTAB t3j1659ki_crr.fits Cosmic ray rejection parameters DARKFILE See Table 6 Dark current correction ramp NLINFILE sbi18555i_lin.fits Non-linearity correction coefficients GRAPHTAB t2605492m_tmg.fits HST graph table COMPTAB t6i1714pm_tmc.fits HST components table Sample Sequence RAPID SPARS10 SPARS50 SPARS100 Dark File for CALWF3 t6119322i_drk.fits t611932di_drk.fits t611932ji_drk.fits t611932bi_drk.fits Table 6: Dark current correction ramps used by CALWF3 for processing the internal flat field ramps. 20

Appendix 2 Light leak morphologies through each IR filter. 21

22