Instrument Science Report ACS 2002-005 HRC Flats: Dust-Mote Patch J. Mack, R. C. Bohlin May 7, 2002 ABSTRACT During the flight level acoustic test in March 2001, one strong dust mote centered at (671,617) disappeared from the HRC detector window. As recommended in ACS ISR 2002-04, we have patched the external illuminated LP-flats using internal flats taken after the acoustic test. 1. Introduction In February and March 2001, flat field images were produced for the ACS HRC filters using the RAS/HOMS (Refractive Aberrated Simulator/Hubble Opto-Mechanical Simulator) with a continuum light source. These flats include both the low frequency L-flat and the high frequency pixel-to-pixel P-flat structure. The RAS/HOMS provides an external, OTA-like illumination above its refractive cutoff wavelength of ~3500 angstroms. Because the RAS/HOMS optics are opaque below 3500Å, the RAS/CAL system (Reflective Aberrated Simulator for Calibration) was used with a UV-bright deuterium lamp to simulate OTA sky illumination for the HRC UV flats. These observations were taken in August 2001. The UV-flats were then corrected by the deuterium illumination error as measured by the F435W ratio of the deuterium flat divided by the external RAS/HOMS tungsten baseline flat. For a detailed discussion on the creation of the flat field reference files, we refer the reader to ACS ISR 2001-11 (Paper I). During the flight level acoustic test on March 23, 2001, one strong dust mote disappeared from the HRC detector window. This feature is centered at pixel (671,617) has a full width of ~40px and a depth of ~10%. As recommended in ACS ISR 2002-04 (Paper III), the LP-flats can be patched by using internal flats which were taken after the acoustic Copyright© 1999 The Association of Universities for Research in Astronomy, Inc. All Rights Reserved. Instrument Science Report ACS 2002-005 test. The following report describes this correction and the steps required to match the level of the internal flats to the RAS/HOMS flats at the boundary of the patched region. 2. Data The internal flats are created using the ACS internal flat field calibration system and have a slightly different illumination than the external flats. Specifically, the internal lamps blur the shadows of the dust motes and cannot properly correct for new motes or for existing motes which may have shifted. Following the acoustic test, however, the dust mote in question actually disappeared. The ACS launch did not cause any further particulate migrations. Table 1 summarizes the images used for patching the HRC flats. Both the tungsten 3 and tungsten 4 (T3 and T4) lamps are used for filters redward of 3500Å. For the UV filters, the original RAS/CAL external flats are used, with no deuterium illumination correction. Since the RAS/CAL flats were taken after the disappearance of the (671,617) dust mote and the external RAS/HOMS tungsten flats were obtained before the acoustic test, this correction, which uses the F435W deuterium to RAS/HOMS tungsten ratio, would actually reintroduce the dust mote onto the flat. While the global L-flat illumination of the internal T3 and T4 flats and the external RAS/CAL flats each differ from the RAS/ HOMS flats, the pixel-to-pixel P-flat structure will be the same. ACS ISR 99-01 (Paper A) compares the P-flat statistics of internal and external flats for three broadband filters. In each case, the one-sigma rms scatter barely exceeds the counting statistics of the total electrons recorded, so that the external and internal flats are the same to better than 0.2% rms. Thus, with the appropriate normalization to match the external OTA illumination, the internal flats will adequately serve as patch data. To avoid any significant loss of signal-to-noise when applying the flat fields to the science data, the Poisson counting statistics of the external flats are in the 0.2 to 0.3% range, i.e. at least 110,000 electrons per pixel (see Paper I). The same signal-to-noise is required for the internal flats to be used as patches. For filters which did not have internal flats with high enough signal-to-noise, we used a filter at an adjacent wavelength. For the blemish free region at (590:690,434:534) on the HRC, Paper III demonstrated that the nearest neighboring flats (in wavelength) have the same pixel-to-pixel P-flat structure to within ~0.2%. 2 Instrument Science Report ACS 2002-005 In Table 1, the source of the patch data for each HRC filter is listed in column 3, the date of the patch observation in column 4, the ACS database entry number for each exposure in column 5, and the total signal (in units of 103 electrons per pixel) in column 6. For the F502N, F550M, F658N, and F660N filters, the internal flats with the adjacent filter given in column 2 are used for the patch. In Table 2, the statistics of these neighboring filters are calculated as in Paper III. The rms residual introduced by this using the adjacent filters is between 0.1 and 0.2%, as shown in Table 2. Table 1. Source of the dust mote patch for each HRC flat field. For four filters, adequate signal-to-noise exposures were not available for the internal flats, and a neighboring filter was used to patch the LP-flats. Adjacent Filter Patch Used Patch Source Date of Patch Observation Patch Database Entry Number Signal in Patch (103 electrons/pix) F220W RAS/CAL 2001 Aug 23 33999, 34000 101 F250W RAS/CAL 2001 Aug 23 33997, 33998 78 F330W RAS/CAL 2001 Aug 23 34002, 34003 75 RAS/CAL 2001 Aug 23 14 F435W T4 Lamp 2001 Nov 06 34009, 34010 34011, 34012 35095, 35096 148 F475W T3 Lamp 2001 Apr 04,Aug 02 28364, 32580 151 F555W T4 Lamp 2001 Apr 03 28120, 28121 200 F606W T3 Lamp 2001 Aug 02 32576 124 F625W T4 Lamp 2001 Jul 13 31283, 31284 200 T3 Lamp 2001 Aug 02 32567 125 T4 Lamp 2001 Apr 03 270 F850LP T3 Lamp 2001 Apr 04, Aug 02 28295, 28296 28297 28361, 32597 F892N T3+T4 Lamps 2001 Apr 04, Aug 02 32575, 28312 114 Filter F344N F502N F475W F550M F555W F658N F625W F660N F625W F775W F814W 3 177 Instrument Science Report ACS 2002-005 Table 2. Comparison of HRC flat field statistics and the pixel-to-pixel fine structure for neighboring filters. F502N F550M F658N F660N Poisson(%) 0.22 0.22 0.21 0.22 Actual sigma(%) 0.74 0.74 0.65 0.59 Sigma Flat(%) 0.71 0.71 0.62 0.55 Minimum 0.97 0.96 0.96 0.96 Maximum 1.10 1.10 1.11 1.11 F475W F555W F625W F625W Poisson(%) 0.23 0.22 0.22 0.22 Actual sigma(%) 0.85 0.78 0.64 0.64 Sigma Flat(%) 0.81 0.75 0.60 0.60 Minimum 0.96 0.96 0.96 0.96 Maximum 1.10 1.10 1.11 1.11 Poisson(%) 0.32 0.31 0.30 0.31 Actual sigma(%) 0.36 0.33 0.32 0.32 Resid. sigma(%) 0.17 0.12 0.09 0.08 NUMERATOR DENOMINATOR RATIO 3. Normalization The small rectangular region (652:690,596:638) suggested in Paper III is used to replace the dust mote feature in the LP-flats. For F606W and longer wavelength filters, a slightly larger region (650:692,595:642) is required to mask the dark ring surrounding the dust mote. In Figure 1, the top-left panel shows the original F475W RAS/HOMS LP-flat, centered on the dust mote at (671, 617). In the top-right panel, the tungsten 3 internal flat has been cut and pasted directly into the desired region of the external flat. While the pixel-topixel structure is the same, a normalization is required to match the flats at the boundary of the patched region. This requires a fit, both in the x and y directions, to eliminate any discontinuity across the flat. Any offset would cause a discontinuity in ACS images which have been calibrated using the patched flat fields. 4 Instrument Science Report ACS 2002-005 Figure 1: F475W Dust Patch. The top-left panel shows the RAS/HOMS LP-flat, centered on the (671, 617) dust mote. The top-right panel shows the internal flat, spliced into the region (652:690,596:638). The internal flat is then normalized to match RAS/HOMS flat in the y-direction (lower-left) and in the x-direction (lower-right). The image stretch is from 0.98 (white) to 1.01 (black). The flat field values of the original dust mote are all white at this stretch, but are as low as 0.90 at the center of the ring. Figure 2 illustrates the fitting which has been applied to match the external and internal flats at the location of the patch. The top panel is the average of columns 652:690 and shows the shape of the illumination in the y-direction of the flat. The solid line corresponds to the upper-right panel in Figure 1. The discontinuity in the RAS/HOMS curve occurs where internal lamp flat has been spliced into the patch region. To appropriately scale the internal flat, the ratio of the external to internal flats is determined at the left and right edges of the patch rectangle. Then, a linear fit is applied as a function of row number to align the two flats. The result is the dashed line which corresponds to the lower-left image in Figure 1. While the flats now are continuous in the y-direction, a clear discontinuity is still apparent in the x-direction. 5 Instrument Science Report ACS 2002-005 Figure 2: The average of columns 652:690 (top panel) and the average of rows 596:638 (bottom panel) in the RAS/HOMS flat, where the internal lamp flat has been spliced into patch region (solid line). A linear fit has been applied to align the flats in the y-direction (dashed line). A fit in both the y- and the x-directions is required to align the external and internal flats at the patch boundary (dotted line). 6 Instrument Science Report ACS 2002-005 The bottom panel in Figure 2 is the average of rows 596:638 and illustrates the shape of the illumination pattern in the x-direction of the flat field. Again, the dashed line shows the correction resulting from a fit in the y-direction only. Although it is much smaller, a discontinuity still exists at the boundary. A second linear fit, this time in the x-direction, is then applied as a function of column number to align the two flats. The result is plotted with a dotted line and corresponds to the lower-right panel of Figure 1. The values in the associated error array of the internal flat will not match the RAS/ HOMS error in the region of the patch. Because the statistical uncertainty is determined from the total signal in electrons in the source image, the error will implicitly be different for each flat. 4. Future Patches Subsequent reports will describe the patches required for the HRC filters in combination with the polarizers and with the coronograph. Recent data indicates, however, that the coronographic spots, which were originally aligned with the Fastie occulting finger, have shifted to the right by ~50 pixels and down by ~10 pixels following gravity release. The HRC coronograph flats cannot be patched simply using the internal lamp flats but must utilize exposures of the bright earth or other suitable astronomical source. References Bohlin, R. C., Hartig, G., Lindler, D. J. Meurer, G., & Cox, C. 1999, Instrument Science Report, ACS 99-01, Paper A, (Baltimore:STScI). Bohlin, R. C., Hartig, G., & Martel, A. 2001, Instrument Science Report, ACS 01-11, Paper I, (Baltimore:STScI). Bohlin, R. C. & Hartig, G. 2002, Instrument Science Report, ACS 02-01, Paper II, (Baltimore:STScI). Bohlin, R. C. & Hartig, G. 2002, Instrument Science Report, ACS 02-04, Paper III, (Baltimore:STScI). 7