WFC3 STAN Issue 24, December 2016
1. UVIS Single-Chip CTE Measurements
S. Baggett, V. Bajaj, M. Sosey, J. Anderson
The pixel-based CTE correction for subarrays has now been incorporated into the WFC3 calibration software calwf3. The new version (3.4) was installed in the pipeline on Oct 26,2016 and provides CTE-correction for UVIS subarrays that contain physical overscan regions (i.e., all nominal subarrays except apertures *M512*SUB and *M1K1*SUB). The CTE correction for full-frame data has been in the pipeline since Feb 2016.
As a result, by default calwf3 now produces two sets of products for the subarrays: the nominal calibrated files (*flt.fits, *drz.fits) as well as CTE-corrected calibrated versions (*flc.fits, *drc.fits). Although the correction was constructed based on full-frame data, it appears to work well on subarrays. A calibration program in 2017 (14880, PI Anderson) will provide data to update the model and explicitly evaluate its performance on subarrays.
We note that as part of calwf3 3.4, the subarrays also receive sink-pixel flagging. Sink pixels are a type of bad pixel that appear to contain a large number of CTE traps; the sinks register systematically low and can generate CTE trails (WFC3 2014-22). The sinks and their CTE trails are flagged in the data quality (DQ) extension with a bit value of 1024 in both the CTE- and non-CTE-corrected files as part of the DQICORR correction; the science pixels remain unchanged. Please see the calwf3 reference guide (ISR 2016-01) and the accompanying cookbook (ISR 2016-02) for more details.
Observers with UVIS subarray data processed before Oct 26, 2016 may re-request their images from the Mikulski Archive for Space Telescopes (MAST) to obtain the new products.
2. WFC3 UVIS Photometry
S. Deustua, J. Mack, V. Bajaj
On February 23, 2016, the WFC3 calibration pipeline, CALWF3, was modified to perform photometric calibration independently for each of the two UVIS CCDs. New keywords PHTFLAM1 and PHTFLAM2are populated in the image header with the inverse sensitivity (zero point) for UVIS1 and UVIS2, respectively, measured within an r = 10 pixel (0.4 arcsec) aperture. This is a departure from the approach implemented at the time WFC3 was installed in 2009. Based on feedback from the WFC3 user community and for consistency with other HST instruments, the photometric keyword values populated in the image header will revert to those for the infinite aperture in early 2017.
The previously existing keyword PHOTFLAM is populated with the value of PHTFLAM1 for backward compatibility with user software. When the new keyword FLUXCORR is set to PERFORM, the counts in the UVIS2 exposure are multiplied by the value of PHTRATIO = PHTFLAM2 / PHTFLAM1. After applying PHTRATIO, a point source should produce approximately the same number of electrons on UVIS1 and UVIS2 in the calibrated flt image, such that a single value of PHOTFLAM applies to both UVIS1 and UVIS2.
Empirical tests revealed that the PHTRATIO computed for the UV filters (F200LP, F218W, F225W, and F275W) did not equal the measured count ratio between chips after processing with CALWF3. On November 21, 2016, the team delivered a new IMPHTTAB, which scales PHTFLAM1 by a factor reflecting the empirical count ratio. The changes are shown in the table below. Observers who downloaded data before November 21 can either retrieve new data products from the archive or else multiply the UVIS2 science extension by (new PHTRATIO / old PHTRATIO) and update the values of PHTRATIO, PHTFLAM1, and PHOTFLAM in the image header.
The following is derived for a circular aperture with r=0.3962 arcsec (r=10 pixels) Old, New = Before, After 21 Nov 2016:
|Filter||Old PHTFLAM1||PHTFLAM2||Old PHTRATIO||New PHTRATIO||New PHTFLAM1|
|erg cm-2 Å-1 e-1||erg cm-2 Å-1 e-1||PHTFLAM2/ Old PHTFLAM1||PHOTFLAM2/New PHOTFLAM1||erg cm-2 Å-1 e-1|
3. Introducing PandExo: A Community Tool for Transiting Exoplanet Science with JWST & HST
Users can now simulate realistic uncertainties for their HST/WFC3 transmission and emission spectra using PandExo. The tool predicts uncertainties for any specified system, recommends an optimized observing strategy (best NSAMP and SAMP_SEQ values), and computes an observation start range in units of orbital phase (necessary for APT). The open-source, Python code for WFC3 uncertainties, as well as a Jupyter Notebook tutorial, are available here. An online version of PandExo will be available in early 2017, along with an arXiv preprint describing its functionality.
For any questions about the WFC3 component of PandExo, please contact Kevin Stevenson at STScI.
4. Clarification on Changes to IR Spatial Scan Output Data
With the release of calwf3 v3.3, the WFC3 Science Team requested that HST Data Processing also change the way IR SCAN data was processed in the pipeline (see "Note!" under CRCORR in IR Pipeline). Specifically, that data processing set the value of CRCORR to OMIT. CALWF3 performs up-the-ramp fitting during the CRCORR step, which for SCAN data produces a minimally useful result. By setting CRCORR to OMIT we stop the ramp fit from happening and instead produce an FLT output image which contains the first-minus-last read result. To clarify, all IR data other than the SCAN data has CRCORR set to COMPLETE.
The output of the first-minus-last read result, by nature, is not a countrate image unless UNITCORR is set to PERFORM. When UNITCORR is COMPLETE, the pipeline will divide by the exposure time to produce a rate image. At the same time, it checks the value for the flatfielding step; if FLATCORR has been set to COMPLETE, meaning flatfielding has already been applied, then the output units of the final image will be in electrons, otherwise they are counts.
If you are unsure of the units of your IR data you can check the following keyword combinations:
- IF: UNITCORR == OMIT and FLATCORR == OMIT then your image is in units of counts.
- IF: UNITCORR == OMIT and FLATCORR == COMPLETE then your image is in units of electrons.
- IF: UNITCORR == COMPLETE and FLATCORR == OMIT then your image is in counts per second.
- IF: UNITCORR == COMPLETE and FLATCORR == COMPLETE then your image is in electrons per second.
5. UVIS Subarray Readout Times
F. Naqvi (GSFC), O. Lupie (GSFC), S. Baggett
When the entire WFC3/UVIS field of view is not required for an observing program, subarrays can be used to minimize readout time and data volume (WFC3 Instrument Handbook). The WFC3/UVIS subarray readout times have been summarized in a new Technical Instrument Report; the median readout times for the most popular subarrays are tabulated below.
For reference, a full-frame four-amp readout is 96 sec. Due to implementation details in the Flight Software, the subarray readout times do not scale linearly with subarray size and some subarrays, such as the quads, can take longer to read out than a full-frame image.
Copies of the Technical Report may be requested from email@example.com.
6. A Few Dither Patterns Place Targets on IR Blobs
P. McCullough & R. Ryan.
Specks of contaminants occasionally deposit on the front surface of the mirror of the channel select mechanism of WFC3 resulting in “blobs” imaged upon the IR detector (ISR WFC3-2010-06). On October 31, 2016, one such blob newly appeared near the center of the detector, close to, but not exactly where a target star typically would be positioned.This prompted us to study if any combination of apertures, dithers, and blob locations would cause a target star to be placed within a blob. By looping through all combinations of WFC3’s apertures, standard dither patterns, and all known blobs’ locations and radii, we find only a few cases of concern where a target would be placed on a blob. They are:
|Case||Aperture||Dither Pattern||Dither (x,y)||Blob #||R||r|
In this (thankfully) short table, the blob number and blob radius R (in pixels) are from Tables 1 and 4 of ISR WFC3-2014-21, and the separation (in pixels) between the target position and the center of the blob, r, is listed in the last column. We report above only combinations for which r < R, i.e. where the target will land within the given blob for that particular dither pattern and aperture. We recommend that users avoid these combinations if they are using staring-mode.
7. UVIS Single-Chip CTE Measurements
C. Gosmeyer & S. Baggett.
In WFC3 ISR 2016-17 we describe the first single-chip measurements of the charge transfer efficiency (CTE) of UVIS’s two-chip CCD. We find that Chip 1's CTE loss is greater than Chip 2's.
We measure CTE by taking star cluster observations (so that sources sample the whole detector) in pairs, where the pairs have the same exposure time, post-flash level, filter, and so on. In the nominal CTE monitor, the pairs differ only in the chip on which the target is observed. The second observation is taken after the telescope dithers up by a chip-height, so that sources that once fell near the amplifiers on the first chip are now imaged far from the amplifiers on the second chip, and vice versa. We measure CTE as the difference in the sources’ flux with distance to the amplifier. In this case, the CTE measurement is an average of the two chips.
In order to disentangle the CTE loss rates of Chip 1 from Chip 2, we performed a test in which the telescope was rotated by 180-degrees in two closely-spaced visits instead of being dithered in a single visit. The same target was then observed on each separate chip with stars near the amplifier in the first visit and far in the second visit. This strategy has not been used for the nominal monitor because it is more difficult to schedule.
The figure below illustrates that CTE loss is worse on Chip 1 (red solid) than on Chip 2 (blue solid) across all cases for which we had data: with and without the pixel-based CTE correction applied and for 60-second and 348-second exposures. CTE loss measurements for the nominal dataset from the same 2010-2011 monitoring program are plotted as dashed lines. They fall between the two solid lines as expected given that the nominal measurements are an average of the two chips. The difference in the CTE loss may be a result of the two chips originating from two different wafers; it is possible the two different lots contained different charge trap populations.
Users should still perform their analysis on the CTE-corrected versions of their observations, regardless of Chip 1’s higher CTE loss. The correction still yields better source recovery over not applying the CTE correction at all. In this dataset we found that faint source recovery for high background images is 20-30% and for low background almost 50%.
As consequence of the chip’s different levels of CTE loss, we will consider adding the 180-degree rotation test into the yearly CTE monitor and creating a chip-specific CTE correction software.
8. Cycle 23 Status of the IR Gain
C. Gosmeyer & S. Baggett
We summarize the status of the IR gain following the analysis of its yearly monitoring data from the Cycle 23 proposal 14376. We measure IR gain using the mean-variance method on pairs of SPARS50, 13-read internal flat fields. Four pairs are observed per epoch, roughly every six months.
The figure shows the gain values of the IR detector’s quadrants. The data range from Oct. 2010 to June 2016. Each point is the average of four measurements in one epoch. The black encompassing lines represent the standard deviation of each epoch’s average measurement. The table lists values calculated from the entire dataset starting in Oct. 2010. For information on the mean-variance method and for full stats on the previous cycles’ measurements see WFC3 ISR 2015-14.
Since the publication of the 2015 ISR, we found a bug in the plotting scripts, such that quadrants 4 and 3 were swapped. The quadrants are correctly plotted here:
(Table 1: Gain measurements for each quadrant in e-/ADU.)
We do not see large jumps in gain from the last cycle and do not anticipate adjusting the 2.5 e-/ADU commanded gain of the instrument. Since changes in the gain imply changes in the detector's overall health, we will continue this yearly monitor into the next cycle with proposal 14539.
9. ISR 2016-12 Anderson "Empirical Models for the WFC3/IR PSF"
It is difficult to model the severely undersampled WFC3/IR PSF, however without an accurate model of the PSF many projects will not achieve the accuracy that is possible. In an effort to remedy this, we have undertaken a detailed study of the WFC3/IR PSF for the most common wide-band filters: F105W, F110W, F125W, F140W, and F160W. The PSFs are purely empirical and are provided in their "effective" (i.e., pixel-integrated) form. They are constructed for use on the un-resampled "_flt" images, as those are the only images where the pixels pose direct constraints on the astronomical scene. Working with multiple dithered "_flt" images is complicated by HST's appreciable distortion, but we provide some tools for such analysis and will in the future provide more tools. This is part of an multi-pronged effort to make direct science on _flt images easier for the community.
More in-depth information can be found in the WFC3 ISR publication.
10. ISR 2016-14 Anderson "Supplemental Dither Patterns for WFC3/IR"
Dithering provides an important way to recover telescope resolution for undersampled detectors. The severely undersampled nature of the WF3/IR detector makes dithering particularly important. This document provides a framework for analyzing the effectiveness of particular dither patterns in terms of the sub-pixel sampling they achieve. It evaluates the patterns that are currently available in the instrument handbook for 2, 3, 4, 6, and 8 points and provides supplemental patterns for dithers of 3, 5, 7, 8, and 9 points. It is often the case that observations are sensibly broken up into any number of exposures to fit the orbit or prevent saturation while providing a desired S/N. With the new and old patterns, users can now dial in any dither for N=1 to N=9 pointings, which should add some flexibility and uniformity to observing programs. The patterns presented here are designed with optimal pixel-phase sampling in mind. As such, they are too tight to provide mitigation for blobs. At present, the new patterns cannot be directly specified in APT, but must be specified as a list of POS-TARGs.
More in-depth information can be found in the WFC3 ISR publication.
11. An Update to the UVIS Photometric Stability and Contamination Monitor
C. Shanahan & C. Gosmeyer
Since SMOV in 2009, the photometric throughput of the UVIS detector on WFC3 has been monitored to evaluate its stability as a function of time, wavelength and chip (e.g. Gosmeyer et al., 2014). By imaging bright, isolated spectrophotometric standard stars, long-term trends in throughput can be measured. The program that accomplishes this also checks for the presence of contaminants on the CCD windows (none have been detected). Recently, several major changes to this monitor have been implemented and the new data have been analyzed; the results are discussed in an upcoming ISR (Shanahan et al. 2016, in prep).
Long-term trends in throughput:
To track the temporal stability of UVIS, we measure the flux of the white dwarf standard stars GRW70 and GD153 in several filters. We define a value, delta-flux, as the percent change in count-rate, with respect to the median count-rate of all images taken on the first visit. We perform a linear fit to this value with time to calculate a percent change in throughput.
The first plot below shows the long-term trends found for GRW70. The filters shown are a small subset that we define as the ‘key’ set of filters, where most of the observations are taken. We see a steady decline (0.2 - 0.3%/year) in throughput for all filters redward of F438W. Declines in throughput on UVIS1 (Quad. A) are steeper than for UVIS2 (Quad. C) in all filters to varying degrees.
The second plot below shows the same plot for GD153. Because GD153 was recently added to the calibration program and historical data is limited, we have far fewer observations in all filters. These low statistics may account for differences between GD153 and GRW70, so this report should be taken as preliminary. We see inconsistent trends in throughput in short wavelength filters, with significant declines found in most filters for both UVIS1 and UVIS2. Like GRW70, we see consistent declines in flux in filters redder than F438W, with declines in UVIS1 (Quad. A) greater than in UVIS2 (Quad. C). We will continue to gather data during cycle 24 and prioritize observations of GD153 in cycle 25 to gain a clearer picture of long term trends in throughput found with measurements of GD153.
(To clarify, errors on the slopes were calculated with Monte Carlo Simulations. We select a random subset of the data for each filter and chip combination and calculate the parameters of the best linear fit. The slope errors are the standard deviations of the distribution in slopes after 10000 iterations.)
Note that these results do not suggest that contaminants are present on the CCD windows, which would manifest as a sharper decline in shorter wavelength filters.
12. New Documentation
- ISR 2016-03 UVIS 2.0 Chip-dependent Inverse Sensitivity Values – S.E. Deustua, J. Mack, A.S. Bowers, S. Baggett, V. Bajaj, T. Dahlen, M. Durbin, C. Gosmeyer, H. Gunning, D. Hammer, G. Hartig, H. Khandrika J. MacKenty, R. Ryan, E. Sabbi, M. Sosey
- ISR 2016-11 The Effect of Repeated Exposures on Measured Fluxes in the WFC3/IR Detector – K. S. Long, S. M. Baggett, and V. Kozhurina-Platais
- ISR 2016-12 Empirical Models for the WFC3/IR PSF – J. Anderson
- ISR 2016-13 WFC3 Cycle 23 Proposal 14373: UVIS Gain – C. Martlin
- ISR 2016-14 Supplemental Dither Patterns for WFC3/IR – J. Anderson
- ISR 2016-15 Trace and Wavelength Calibrations of the WFC3 G102 and G141 IR Grisms – N. Pirzkal, R. Ryan, and G. Brammer
- ISR 2016-16 Reprocessing WFC3/IR Exposures Affected by Time-Variable Backgrounds – G. Brammer
The archive of all WFC3 Instrument Science Reports (ISRs) is here.
The latest version (4.0) of the WFC3 Data Handbook is here.
Need help? Start here.