STScI Logo

Hubble Space Telescope
WFC3 STAN Issue 6, January 2011

WFC3 Space Telescope Analysis Newsletter - Issue 6, January 2011

For new information about WFC3 visit the "New in the Last 45 Days" and "Late Breaking News" sections of the WFC3 website at /hst/wfc3.

This and previous issues of the STAN can be found at /hst/wfc3/documents/newsletters.

This STAN is being released in time to support proposal development for Cycle 19.
The other items provide a mix of calibration, reference file and support updates.
1. Updated WFC3 IR Flat Fields
2. Plans for Updating UVIS Flat Fields
3. Charge Transfer Efficiency (CTE) in UVIS
4. Spatial Scanning Availability for Cycle 19
5. IR Persistence Update
6. UVIS High Resolution Imaging
7. IR High Resolution Imaging
8. WFC3 Grism Support
9. Data Processing Updates
10. When to Recalibrate
11. Spectroscopic ETC Issue
12. New Documentation

1. Updated WFC3 IR Flat Fields - Tomas Dahlen, Nor Pirzkal, Elena Sabbi, Jennifer Mack

During the spring of 2008, in the third and last thermal vacuum test (TV3), the Wide Field Camera 3 (WFC3) team carried out an intense ground-based campaign to, among other purposes, create flatfields for the reduction of all WFC3 on-orbit data. Ground-based flatfields were obtained using an optical stimulus (CASTLE) to simulate the sky illumination of the UVIS CCDs (Sabbi et al., ISR 2008-46) and the IR array (Bushouse, ISR 2008-28). These flat fields include both high frequency pixel-to-pixel (P-flat) and low frequency (L-flat) structures. After WFC3 was installed on Hubble in Servicing Mission 4, tests were performed during the Servicing Mission Observatory Verification (proposals CAL-11452 and CAL-11453) as well as in Cycle 17 calibration programs CAL-11911 and CAL-11928 to evaluate the quality of the ground-based flatfields.

The on-orbit testing showed that the ground-based IR flatfields have low frequency structures in them that are not representative of the true sky illumination pattern. This should mainly be caused by a mismatch between CASTLE and OTA. When ground-based flatfields are divided into science images they leave behind low-frequency residuals that affect photometry. To better determine the true low-frequency illumination pattern, sky images were constructed by combining a large number of science images after masking out objects. As a first step, sky images were created in different broad-band filters (F098M, F105W, F125W, F140W, and F160W) to investigate any wavelength dependence of the low frequency features. No significant difference between filters was found. The estimated uncertainty due to color is significantly less than 1%. Thereafter a gray sky image was constructed by combining over 2000 images, each with exposure time over 300s. The sky image was then smoothed, and as a final step, each of the existing ground-based flatfields were multiplied by the sky image and then normalized to unity to produce new flat-fields that correct for both pixel-to-pixel variations and low frequency structure.

Independent measurements of the star cluster Omega Centauri in different broad-band filters confirm that the low frequency amplitude and spatial features are corrected with the new flatfields, as well as the lack of a significant dependence on wavelength. Due to the relatively small number of bright cluster stars at infrared wavelengths, the low frequency flats computed from stellar photometry have lower signal-to-noise compared to the sky flats and were therefore used only to verify that the same correction is required for both extended and point sources.

The uncertainty in photometry due to the flatfield correction has an estimated rms of <0.7% over the whole detector, with a maximum peak-to-peak range of -2.0 to +1.9%. The uncertainties are typically larger at the edges of the detector. For the central part of the detector [129:896,129:896] (i.e., excluding a region around the detector edges with thickness 1/8 of the detector size), the uncertainty is <0.5% rms, with peak to peak range of -1.5 to +1.6%. For the edge region of the detector only (i.e., a frame with 128 pixel thickness), which includes the "wagon-wheel" feature, the rms is <0.8%, with peak-to-peak -2.0 to +1.9%. The introduced uncertainty in photometry due to flatfielding therefore typically has an rms of less than 0.01 mag, but at the edges of the detector the errors may reach 0.02 mag. In addition, the uncertainty in the wavelength dependence of the sky image correction should also be less than 0.01 mag.

The new flatfields (pfl) were ingested into the CDBS on December 7, 2010, 10.31pm. Data retrieved after this date are processed using the new reference files. WFC3/IR users interested in applying these new flat-fields to older data can either re-retrieve the data from the archive, or re-run the standard calibration pipeline CALWF3 on the IR raw data using the new flatfields. Alternatively, users who do not wish to reprocess their data with CALWF3 can apply the low frequency correction only to their calibrated data products. Detailed instructions on how to reprocess data can be found at the WFC3 webpage (/hst/wfc3/analysis/ir_flats).

Note also that these new flatfields supersede the alpha release flatfields that were described in STAN issue 4, June 2010 and were available on the WFC3 webpage between May 26 and December 7, 2010.

2. Plans for Updating UVIS Flat Fields - Jennifer Mack, Tomas Dahlen, Elena Sabbi

While the CALWF3 pipeline still uses flats obtained during ground calibration for the UVIS channel, new flat fields are in preparation, and these will be available to users in the coming months. Stellar photometry of dithered exposures of Omega Centauri at various roll angles has been used to characterize the residual low frequency structures. By placing the same star over different portions of the detectors and measuring relative changes in brightness it is possible to identify local variations in the response of the detectors. These variations have been modeled using the same matrix solution algorithm developed for the Advanced Camera for Surveys (van der Marel, ACS ISR 2003-10). The algorithm divides the detector into a 32x32 grid and computes the flat field residual in each box. Preliminary results indicate that the required L-flat correction is ~0.8% rms, with a peak-to-peak of +/- 2.5% over the whole detector.

The photometric residuals show a wedge-shaped 'flare' in quadrant A extending diagonally into quadrant D. The strength of the flare is ~1% over the background, and it is the same amplitude (but in the inverse sense) as the flare in the ground flat. The flare is caused by light reflected from the detector, returning to the detector chamber window, and then back to the detector. A geometric model of ghost reflections on the UVIS flat predicts a wedge-shaped feature which is apparent in both the TV3 ground flats and in recent in-flight images of the dark side of the Earth acquired during periods of full moon illumination (CAL-11914).

The 32x32 grid matrix solution worked well for ACS, where the low-frequency residuals were smooth, but it does not adequately model the 'sharp' diagonal edges of the flare apparent in the WFC3 UVIS ground flats. For this reason, and because it is not a true variation in the detector QE, the flare should be removed prior to solving for the low-frequency residuals from stellar photometry. The flat fielding team is currently working on methods of separating the additive term of the ghost reflections from the multiplicative flat fields.

3. Charge Transfer Efficiency (CTE) in UVIS - Sylvia Baggett, Kai Noeske

As with other HST CCDs like WFPC2, STIS, and ACS, the on-orbit radiation environment is damaging the WFC3/UVIS detectors and causing a degradation in the WFC3/UVIS CTE over time. Specifically, moderately bright stars (500-2000 e- in 60 sec) in narrowband images with low background (0.1-1e-), situated far from the readout amplifiers, showed a CTE loss of ~2.5% and ~8% at around 5 and 17 months after installation, respectively. Moderately bright stars in broadband images with higher background (1-30e-), showed an average CTE loss of ~2% after 16 months in operation (Khozurina-Platais et al., 2011 in prep). The rate of the CTE degradation in WFC3 appears to be somewhat higher than expected, likely due to enhanced cosmic ray rates during the recent solar minimum.

If the CTE degradation continues at this pace, by March 2012 (midway through Cycle 19) moderately bright sources in low sky background images far from the readout amplifier will experience >16% losses while very faint sources may suffer losses of several 10's of percent. In this case, observers will want to consider whether their project would be adversely impacted by the CTE losses and if so, how best to mitigate the effect. One option which will be available to observers in Cycle 19 is charge injection (CI): an electronic insertion of charge into the chip which serves effectively as a pre or postflash but with a relatively low noise penalty (~17000 e- injected charge with ~15 e- noise in the CI rows, 4-5 e- noise away from the CI rows). For further discussion of the CTE levels measured in WFC3 along with the advantages and disadvantages of using CI, please see

4. Driftscan Availability for Cycle 19 - Peter McCullough, John MacKenty

In the first half of 2011, we will be testing spatial scanning with HST during WFC3 IR exposures with HST programs 12325 and 12336. This capability will intentionally turn stars into well-defined streaks on the detector, or, for example, spread a stellar spectrum perpendicular to its dispersion. There are at least two motivations for implementing such a capability: 1) reducing the fraction of overhead in observations of very bright stars such as those suitable for spectral characterization of transiting planets, and 2) enabling observations of very bright primary calibrators that otherwise would saturate the IR detector. Program 12325's goal is to validate commanding for the spatial scanning, including synchronization of the spacecraft's motions and the detector readout. (Note that we will not read out the detector in a time-delay-integration (TDI) mode (sometimes called "drift scanning"); instead we are operating the detector nominally except for the detail of timing each exposure's beginning to capture the target at the intended initial position on the detector.) The first tests, in which a star is imaged with WFC3 IR under various parameterizations of HST's scanning speed and orientation, are expected at the end of January, 2011. The results of those initial observations will inform the remainder of program 12325 (scanning with a grism in place) and all of program 12336 (comparison of photometry in scanning mode to (normal) staring-mode photometry). Observers may propose in Cycle 19 to use spatial scanning for WFC3, but no other instrument (other instruments will have to wait for additional tests and a future proposal cycle). For Cycle 19, the scan speed must be less than 1 arcsecond per second, and only scans of a single line segment will be permitted per exposure, i.e., only a single starting point, rate, and direction will be permitted per exposure. We anticipate a minimum overhead of 36 seconds per exposure, allocated as follows: 26 s to prepare the instrument, primarily the detector, 5 s to ramp up the scan rate, and 5 s to ramp it back down after the exposure. Additional overheads such as buffer dumps will be the same as without spatial scanning. Actual overheads will be verified in the engineering tests. We advise observers that neither the CALWF3 pipeline nor generic analysis software are designed to calibrate spatially scanned WFC3 data and they should plan accordingly. The Cycle 19 TAC will be informed of the results of the engineering tests that have been completed prior to its meeting.

5. IR Persistence Update - Knox Long

When exposures of the IR detector on WFC3 reach near or beyond saturation, a residual image of these pixels can be seen in subsequent exposures. The amount of persistence is typically 0.65 e-/s, 1000 seconds after the pixel was saturated. The persistence arises from trapped charge being slowly released over time, and is not affected by the number of non-destructive readouts or resets of the detector after a pixel has been illuminated.

From the point of view of an observer, persistence can be characterized as "internal", induced by exposures the observer has made in a single visit, and "external", induced by exposures made by other observers that happened to be scheduled just prior to the observations in question. STScI does attempt to identify visits that are likely to cause a significant amount of persistence in subsequent observations using information from the Phase II proposals and in those cases, the mission planners suspend IR observations for several orbits. Nevertheless, in the period since WFC3 was installed, STScI, often prompted by observers inspecting their data, has identified about 10 cases where persistence was significant enough that it posed problems to an observer's ability to extract science from his/her data. Consequently, when observers receive data involving the IR channel, one of the items they should check for is persistence. We remind users that there are time limits associated with identifying observations that have "failed" from a scientific perspective, and therefore qualify for re-observation. Procedures for identifying persistence are described on


If, after investigation, an observer suspects that persistence could be a problem, the observer should contact being sure to provide sufficient information (preferably the dataset name or the name of the flt file) for STScI to identify the affected exposure or exposures. We will then work with you to determine whether we can provide corrections to remove the effects of persistence from your data, or to advise you that an HST Observation Problem Report (HOPR) be submitted.

In planning, observers can help to minimize the effects of external persistence by keeping exposure levels down to of order half saturation (35,000 e-), especially in the central portion of the image, and if this is not possible by noting in the Phase II observation description that the planned exposures could be a persistence problem.

The WFC3 group is working on software to aid in the evaluation and removal of persistence from IR data. At present, we are able to remove about 90% of the persistence in images based on a model that depends only on the amount of exposure a pixel received and the time since the earlier exposure. We are currently testing this software. When this is complete early in 2011, we expect to be able to provide to observers an image containing the estimated persistence in each of their flt files.

6. UVIS High Resolution Imaging - Ronald Gilliland

Images were obtained in six filters sufficiently deep to probe for companions to delta-magnitudes > 10 for the V = 12.0 calibration target P041C as part of CAL/WFC3-12354. The ability to detect faint companions of bright targets was assessed and contrasted to Adaptive Optics and Speckle imaging results from four ground-based systems. Applications of a survey mode where some 20 targets per night could be observed on an AO system, or the HST observations conducted as brief SNAPSHOTs are considered. WFC3-UVIS imaging is found to be superior to Keck, MMT and Palomar AO, and WIYN Speckle systems by a factor of greater than or about 5 in a metric of the fraction of stars within an area (to 2 arcsec radius) over delta-magnitude (to 10) which can be probed for faint companions. A detailed write-up is in Gilliland and Rajan 2011, ISR WFC3 2011-03.

7. IR High Resolution Imaging - Abhijith Rajan

The WFC3 team is currently investigating the capability of the WFC3 IR detector to perform High Contrast Imaging. For this we have observed a bright standard star (V=6.86 mag) in the F128N filter in CAL/WFC3 program 12354. The exposures were carefully designed to maximize the number of orients, to investigate angular differential imaging (Marois et al. 2005), and to saturate the inner half arcsec to get sufficient flux in the wings of the PSF.

To measure the detection limit of the camera we injected fake planets in the images (at 0.5", 0.8", 1.1", and 1.7" from the star) with delta magnitude of approximately 10 with respect to the star (as measured by the ETC). Our tests show that with every additional orient we were able to improve the detection limit of the fake planets. These results indicate that the WFC3/IR detector can be used for high contrast imaging (Rajan and Gilliland (in prep)).

8. WFC3 Grism Support - Howard Bushouse

The calibration, user support, and software support for the WFC3 grism modes has, until recently, been provided by a team from the Space Telescope European Coordinating Facility (ST-ECF), as they had done previously for the NICMOS and ACS grism/prism modes. As of January 1, 2011, the ST-ECF has closed and will therefore no longer be providing this support. All responsibilities have now been taken over by the WFC3 group at STScI. The aXe software package, which is used to reduce and calibrate WFC3 and ACS grism data, will continue to be available within the STSDAS software package. Information about aXe and access to the aXeSIMweb tool will now be available at the web site Information about the WFC3 grism modes and the calibration files that are used with aXe to reduce WFC3 grism data are now available within the STScI WFC3 web site at /hst/wfc3/analysis/grism_obs.

Inquiries about the WFC3 grism modes, calibration files, and aXe software should now be directed to the STScI help desk at, where they will be routed to the appropriate individuals.

We thank the ST-ECF team, consisting of Harald Kuntschner, Martin Kuemmel, and Jeremy Walsh, for their years of support for the WFC3 and other HST grism modes and wish them well in their future endeavors.

9. Data Processing Updates - Howard Bushouse

Calwf3 v2.1 and MultiDrizzle v3.3.7 have been in use in the STScI OPUS pipeline since August 25, 2010. A few updates have been made to calwf3 since that time, which are included in calwf3 v2.2. This version of calwf3 is expected to go into operations in the OPUS pipeline by mid-January 2011 and will also be released to the public in STSDAS v3.13 at about that same time. Information about the changes included in calwf3 v2.2 can be found in the Pipeline area of the WFC3 web site, at /hst/wfc3/pipeline/CALWF3ReleaseNotes/notes/CALWF32p2.html.

Briefly, there were four changes made to calwf3. The most important change for users, in terms of the scientific validity of the calibrated data, is the first one described below.

1) The IR ZSIGCORR routine in calwf3, which estimates the amount of signal contained in the zeroth read for each pixel of an IR exposure, had been skipping the calculation for pixels that had a non-zero Data Quality (DQ) flag value in the associated DQ array. This could then lead to slightly incorrect linearity corrections in the subsequent NLINCORR step, due to the lack of compensation for the zero read signal. This was most noticeable in the regions of IR images affected by the so-called "blobs", which are flagged with DQ=512, if the pixels within a blob area contained a strong signal. The ZSIGCORR step has been modified to perform the zero read calculation for all pixels, regardless of their DQ values.

2) The routines that check for the existence of some of the calibration reference tables used during calwf3 processing have been upgraded to return informative error messages when the tables can't be found.

3) The logic for processing IR images has been upgraded to allow for re-entrant processing. This can be useful when there is a desire to process an image only part way and then use the partially processed image as input for additional processing.

4) The logic that is used to execute the WF3REJ task from within calwf3, which is used to combine multiple images from CR-SPLIT and REPEAT-OBS associations, has been upgraded to skip this step in the event that only one of the expected images actually exists.

10. When to Recalibrate - Knox Long, Howard Bushouse

Most observers retrieve and inspect data from MAST shortly after their observations are taken. Many of us then take some time before beginning "serious data analysis", the data analysis that is necessary to produce a refereed publication. Generally speaking, if there is a very significant delay between the time an observer has last retrieved data and the time at which serious data analysis begins, we recommend that you re-retrieve the data from the archive. The reason for this is that either the calibration software (CALWF3) may have been updated or the calibration files used by the software may have changed.

However, there are intermediate cases where it is less obvious what you should do. StarView provides a mechanism to see the currently recommended calibration files, which can then be compared to the calibration files that were used for the data an observer has on his/her machine.

To use the StarView web tool, go to:

and open the WFC3 tab, and then the page entitled "WFC3 Best Reference Files". This will allow you to query the Calibration Data Base System (CDBS) for the most current calibration files on the basis of, for example, proposal ID or dataset name. The results can then be compared to the files used when you retrieved your data.

To aide in these comparisons, members of the WFC3 team have written a simple python script, which for historical reasons is called After exporting the results of a StarView query to a comma-separated (CSV) file, the script can be used to produce a simple text file that identifies the calibration files that have been changed. One is then in a much better position to decide whether any updated versions of the data are needed.

You then have the choice of either re-retrieving the datasets from MAST, which will apply the latest calibration files during processing, or you can recalibrate the data yourself at your home institution. In order to do the latter, the calibration file keywords in the raw image headers of your datasets must be updated before reprocessing. The upref python script produces a simple IRAF script that can be used to perform these keyword updates.

More information about the script and how to use StarView to get the currently recommended calibration files can be found at: /hst/wfc3/analysis/should_I_recalibrate

11. Spectroscopic ETC Issue - Howard Bushouse

The ETC v19.1, which was released in December 2010, continues to have a shortcoming in spectroscopic mode calculations that also existed in previous versions of the WFC3 ETCs. Specifically, the WFC3 spectroscopic ETCs were built upon the ACS spectroscopic ETC, which does not convolve the input spectrum with a point spread function. In contrast, the NICMOS spectroscopic ETC does employ this convolution. There are two pitfalls here. First, and most importantly, a user can obtain an overly optimistic estimate of the signal-to-noise ratio in an emission line if an extremely narrow line width is specified in the input spectrum. Second, comparison of the results from the ETCs for these instruments will not necessarily be realistic.

Emission lines can be simulated via a template, a user-supplied spectrum, or the emission lines section of the ETC input page. It is possible via these methods for a user to specify a line that is narrower than a pixel. For the ACS and WFC3 ETCs, the result will be a simulation with most or all of the flux confined to a single pixel, whereas in reality the emission line would span multiple pixels.

The WFC3 and ACS spectroscopic ETCs will incorporate PSF convolution in a future release, but for now this problem can be mitigated by specifying an appropriately broad emission line. The instrumental resolution for each of the grisms is about 1.5 pixels (35 A) for the IR/G102, 1.5 pixels (70 A) for the IR/G141, and 1.9 pixels (27 A) for the UVIS/G280. These are the widths that should be specified for an infinitesimally narrow line. For lines with a significant intrinsic width, the width specified in the ETC should be the quadrature sum of the intrinsic and instrumental widths. For example, if a line has an intrinsic width of 20 A and the observer is using IR/G102, the line width should be specified as sqrt(400+1225)=40 A. Using the unconvolved line width in the ETC can lead to significant errors in the estimated S/N and/or exposure time. For example, a WFC3/IR G102 simulation using an emission line width of only 20 A, as opposed to a more realistic width of 40 A, leads to a difference in the exposure time necessary to reach a given S/N of 70-100%, depending on the line strength.

12. New Documentation

These new ISRs have been published since the last STAN (October 2010):
ISR 2011-03 WFC3 UVIS High-resolution Imaging Performance -- R. Gilliland and A. Rajan
ISR 2011-02 WFC3/UVIS Charge Injection Behavior: Results of an Initial Test -- H. Bushouse et al.
ISR 2011-01 Master Sky Images for the WFC3 G102 and G141 Grisms -- M. Kuemmel et al.
ISR 2010-17 WFC3/IR Persistence as Measured in Cycle 17 using Tungsten Lamp Exposures -- K. Long et al.
ISR 2010-16 IR Channel Subarray Dark Current Behavior -- B. Hilbert
ISR 2010-15 Boosting Count-rates with Earth Limb Light and the WFC3/IR Count-rate Non-linearity -- Riess and Petro
ISR 2010-14 The Photometric Performance of WFC3/UVIS: Temporal Stability Through Year 1 -- J. Kalirai et al.

The complete WFC3 ISR archive is at:/hst/wfc3/documents/ISRs/

Need help? /hst/wfc3/help.html

To subscribe or unsubscribe to the STAN, send a message to with a blank subject line and the following in the body: [un]subscribe wfc3_news
The Space Telescope Science Institute is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS5-26555.