December 14, 2011
Revised flat fields for the remaining 35 UVIS filters are now available from MAST. Observations in these filters retrieved after December 14, 2011 (03:02:58 AM) will be calibrated with these new reference files. Further details will be available in a forthcoming Instrument Science Report.
August 08, 2011
Revised flat fields for seven UVIS filters are now available from MAST. Observations in these filters retrieved after August 08, 2011 (9:00pm) will be calibrated with these new reference files.
Alternately, the new flat fields (listed below) may be obtained from the following page:
/hst/observatory/cdbs/SIfileInfo/WFC3/WFC3PFLbin1UVIS
F336W v8816165i_pfl.fits
F390W v8816166i_pfl.fits
F438W v8816167i_pfl.fits
F555W v8816168i_pfl.fits
F606W v8816169i_pfl.fits
F775W v881616ai_pfl.fits
F814W v881616bi_pfl.fits
The total change peak-to-peak with respect to the previous pipeline flats (obtained during ground calibration) ranges from 3.6% to 5.6%, increasing with wavelength from F336W to F814W. Calibration observations were also obtained for the filters F225W, F275W, and F850LP, and flats for these filters will be released in the near future. Flat fields for the remaining UVIS filters will be computed via wavelength interpolation of these 10 broadband filters.
The new flat-fields have been normalized to an “infinite” aperture. When using these flat-fields, users performing aperture photometry using apertures less than 10 pixels (0.4”) are urged to apply local aperture corrections to 0.4”.
Figure 1 shows the ratio of six of the new flats with respect to the ground-based flats. The most evident feature in each ratio is an extended blue/purple wedge that covers ~40% of the detector area and is caused by the removal of an extended ghost which was present in the ground flats. The wavelength dependence of the chip-dependent quantum efficiency is also evident.
Figure 1: Ratio of the new flats with respect to the ground-based files for the filters F336W, F390W, F438W (Upper Panel), F555W, F775W, and F814W (Lower Panel). The color bar ranges from 0.98 to 1.04. Because the ratio is similar to F555W, the F606W filter has been omitted from the figure.
Details on the preparation of the new flats are provided via the following links.
- 1: The UVIS Flare Subtraction
- 2: The Low-Frequency Corrections
- 3: Variations in Encircled Energy
- 4: Validation of the Solutions
1: The UVIS Flare Subtraction
During the spring of 2008, in the final thermal vacuum test (TV3), the Wide Field Camera 3 (WFC3) team carried out an intense ground-based campaign to, among other purposes, create flat-fields to be used for the reduction of all WFC3 on-orbit data. Ground-based flat-fields were obtained by simulating the sky illumination of the UVIS CCDs (Sabbi et al., ISR 2008-46) and the IR array (Bushouse, ISR 2008-28) using the optical stimulus (CASTLE). These flat-fields should include both the high frequency pixel-to-pixel (P-flat) and low frequency (L-flat) structures. Tests performed on orbit during the Servicing Mission Observatory Verification (SMOV) indicated that the ground-based flats did not fully remove low and medium frequency structures. These results were further confirmed during the calibration programs in Cycle 17 and Cycle 18.
A major source of error in the ground-based flats is due to an extended ghost that affects ~40% of the entire field of view. Because of the tilted UVIS focal plane, light is reflected multiple times between the detector and the two chamber windows. This ghost or ‘flare’ has imprinted a wedge-shaped feature in the ground-based flats (Figure 2). The strength of the flare changes as a function of wavelength and has its maximum in the upper left corner where it ranges from 2.0% (in the F336W filter) to 4.0% (in F814W filter). A simplified geometric model of the light reflections has been used to remove the flare from new flats, however the wavelength dependence of the reflectance, based on the refractive index of fused silica (SiO2), is still not well-constrained. Further calibrations are planned to better characterize this parameter.
Figure 2: Geometric model of the flare.
2: The Low-Frequency Corrections
Once the flare has been removed from the flats, residual low-frequency structures, which are caused by differences in the ground-based and in-flight optical paths, have been characterized using the same software and methodology developed for ACS (see ACS ISRs 2003-10 and 2002-08). Briefly, photometry of the globular cluster Omega Centauri, observed at various roll angles and with large dithered steps across the detector, has been used to quantify magnitude differences between measurements for the same stars at different positions on the detector.
Photometry of the Omega Centauri data was performed using an aperture radius of 0.2” (5 pixels) to minimize the uncertainties due to crowding. A spatially variable aperture correction to 0.4” (10 pixels) was then applied to remove any artifacts introduced by changes in the PSF, due to variations in encircled energy (EE) with detector position and changes of telescope focus (breathing) throughout the orbit. Because of this effect, users are urged to apply local aperture corrections to 0.4” when applying these new flats to photometry with aperture radii smaller than 0.4”.
The new WFC3 L-flat corrections for the 6 of the 7 filters (Top= F336W, F390W, F438W; Bottom= F555W, F775W, F814W) are shown in Figure 3 with a stretch of +/- 2.2%. The low-frequency correction increases as a function of wavelength from 2.7% to 4.5% from F336W to F814W.
Figure 3: Low-frequency corrections for the filters F336W, F390W, F438W (Upper Panel), F555W, F775W, and F814W (Lower Panel), as derived from stellar photometry. Because the correction is similar to F555W, the F606W filter has been omitted from the figure. Red indicates that the ground flats are overcorrecting the data, making the stars too faint. The solutions are presented as a 32x32 grid, such that one pixel in each tile corresponds to 128x128 pixels on the UVIS detector.
A previous alpha-release of the flat fields (which were available from the WFC3 webpage on March 25, 2011) include the same flare correction used for the new release, however the low-frequency corrections were derived from 5 pixel aperture photometry. Since that time, the spatial dependence of the PSF and it's relation to telescope breathing has been investigated in more detail to provide improved flats which can be used for any aperture. While the alpha-flats are only accurate for photometry performed with a 5 pixel aperture, the new flats now available from MAST can be applied to any dataset, as long as the photometry is corrected to an 0.4" aperture.
3: Variations in Encircled Energy
The low-frequency corrections were computed using aperture photometry in the core of the globular cluster Omega Centauri. A small radius (0.2”) aperture was chosen to minimize the effects of contamination from nearby stars. However a detailed analysis showed that for radii smaller that 0.36” (9 pixels) the UVIS PSF strongly depends on both detector position and the telescope focus at the time of observation. Beyond 0.4” the PSF is extremely stable.
The dependence of the low-frequency correction on aperture (and therefore encircled energy) is shown in Figure 4 for the F606W filter. Note that the solution becomes noisier when increasing the aperture radius because of crowding. Still the solution derived for a 9 pixel aperture is nearly identical to that computed from the 5 pixel photometry, once the correction to 0.4" aperture is applied, validating the robustness of the technique.
Figure 4: Low-frequency corrections derived for the F606W filter from photometry in an aperture of 4, 5, 6 and 9 pixels in radius (shown from left to right). The far-right panel shows the low-frequency correction computed from a 5 pixel aperture, after applying a local aperture correction to a 10 pixels (0.4”). The solutions are shown on a stretch of +/-2%.
Depending on which part of the orbit the observations were taken, the effects of telescope breathing can dominate the true low-frequency correction. Changes in the aperture correction from 5 to 10 pixels as a function of orbital position are shown in Figure 5 for five images in F336W acquired in sequence.
Figure 5: Upper Panel: Changes in the 5 to 10 pixel aperture correction as a function of position for 5 consecutive images acquired with the F336W filter. The total correction ranges from 0.073 mag (purple) to 0.041 mag (red). The telescope focus in microns is also plotted as a function of time, and the 5 consecutive observations are over-plotted in red (Lower Left panel). The median aperture correction and the standard deviation for the five images are shown in the lower-right panel.
4: Validation of the Solutions
The new flats have been independently validated with 3 separate methods. A first validation comes from the comparison between the low-frequency corrections, as derived from stellar photometry, and preliminary Earth flats obtained observing the dark Earth limb in the F606W and F814W filters.
The accuracy of the new F336W flat field has been verified by stepping the spectrophotometric standard GD153 across 46 positions over the detector. Figure 6 shows that once the new flat is applied, the average variation with respect to the mean is 1.000 +/- 0.003 with no systematic spatial correlation.
Figure 6: Variation relative to the mean magnitude of the spectrophotometric standard GD153 observed with the F336W filter using the new flat field. Magnitudes have been measured with a 10 pixel aperture (0.4”).
Figure 7 shows the difference in apparent magnitude for the stars found in in two separate images of Omega-Centauri acquired with a ~185 degree roll. The analysis was performed from photometry in a 5 pixel aperture using the ground-based flat (Left Panels), the March 2011 ‘Alpha-release’ flats (Middle Panels) and the new flats using a local correction to the 0.4” (Right Panel). While the ground flats show large spatial residuals (rms=0.025 mag), results obtained from the Alpha-Release, and the new flats are nearly identical, where the rms=0.010 mag. The similarity in the residuals is not surprising, since the Alpha-release flats were specifically designed to work for a 5 pixel aperture. As previously discussed, the new flats have been instead conceived to have a wider usage.
Figure 7: Difference in the apparent magnitude for stars in the magnitude range 16.0 < mF814W < 19.5 (with photometric errors less than ~0.1%) in two images of Omega-Centauri acquired with a 185 degree roll angle (~5 month apart). The photometry using a 5 pixel aperture is shown using the ground-based flats (Left Panels), the ‘Alpha-release’ flats (Middle Panels) and the new flats with a local correction to the 0.4” (Right Panel). The magnitude difference is plotted as a function of x-position for four separate regions selected by y-position. The red line represents zero difference and is shown only to guide the eye.
Created 08/08/2011 MJD
Modified 09/08/2011 MJD