Wide Field Camera 3 Update

John MacKenty, mackenty@stsci.edu

Wide Field Camera 3 (WFC3) continues to function well. It has nearly completed its Cycle 17 science and calibration programs, and is now well into Cycle 18. This article highlights some of the accomplishments and future directions of the WFC3 calibration program, which has evolved through three stages.

Following installation, the initial calibration effort was designed to characterize the instrument and verify that its performance met the expectations from pre-flight calibration. The second stage, the Cycle 17 program, established the working calibrations and explored the long-term stability of the instrument. The third, the smaller Cycle 18 program is extending the calibrations, continuing to monitor stability, and opening up new parameter space.

Photometric Calibration

Following its May 2009 installation in Hubble, WFC3’s performance has met or exceeded our expectation. The system throughput is 5–15% greater than indicated by ground calibration. Careful monitoring of standard stars through a broad set of filters shows that the throughput is very stable (<0.3% variation in the ultraviolet-visible [UVIS] and <0.5% in the infrared [IR]). WFC3 is now well calibrated on the primary Hubble standard stars.

Astrometric Calibration

Given the wide use of MultiDrizzle software to combine exposures in the same filter—even taken with two or more different instruments—we have made a significant effort in Cycle 17 to obtain a stable astrometric calibration for WFC3. Astrometry with WFC3 was calibrated in ten UVIS and five IR filters to an accuracy of 0.1 pixel, and the solution for the geometric distortion was solid to better than 0.05 pixels. Nevertheless, combining images between visits (or instruments) still requires significant manual intervention, because the astrometry of the guide stars is considerably less precise. Also, the line-of-sight stability of WFC3 relative to a Fine Guidance Sensor is about 0.35 UVIS pixels or 0.31 IR pixels over a two-orbit visit.

Dark Frames

The WFC3 dark-frame calibration was straightforward in both the UVIS and IR channels. As expected, the monthly annealing of the CCD detectors successfully removes the majority of new warm pixels. Calibrations of the warm and hot pixel population are maintained on four-day centers, and are available from the Institute web site. The IR detector appears to be quite stable in the radiation environment of space.

We have added to the list of “bad” pixels some optical features on the mechanized mirror that selects the channel (and is closed during focus operations). We have aggressively masked poorly behaved pixels, particularly those with occasionally anomalous dark current. Users should consider which “bad” pixel features to exclude from their observations based upon both the number of dithered exposures available and their particular science goals.

Flat Frames

The low-spatial-frequency flat fields (L-Flats) have proven time-consuming to calibrate. During ground calibration, we obtained both internal and external flat fields. The internal ones provide reasonable high-spatial-frequency flats (pixel-to-pixel or P-Flats), but have significant limitations regarding illumination. Their primary utility is to verify the stability of the P-Flat obtained from the external, ground-illumination system. This calibration appears very stable in all filters in both channels. The mismatch in the large-scale illumination pattern, however, results in 4–5% peak-to-peak errors in the flat fields at low spatial frequencies.

The high level of residual L-Flat error was expected. We made a major effort in Cycle 17 to obtain good L-Flats via repeatedly stepped exposures of the globular cluster Omega Cen (which also supported the astrometric calibration). The analysis of these data is subject to significant systematic errors, driven in part by source crowding and variations in the point-spread function (PSF) due to telescope breathing.

For the IR channel, the combination of ~2000 long-exposure, broad-band images contains sufficient background flux to create a high-quality L-Flat. These data demonstrated both that the systematics in the cluster-based flats are understood and that the low-spatial-frequency errors in the ground calibrations are mostly independent of wavelength. Based on these results, we corrected all IR ground flats using the smoothed background-light L-Flats. This calibration provides the confidence needed to proceed with the L-flats derived for the star cluster for the UVIS channel, and we will release the results shortly.

We are exploring the potential value of flat fields obtained using the Moon-lit Earth as an illumination source, to validate and possibly improve the flat fields. Early experiments look promising. (The sunlit Earth is too bright for broad filters in either channel, and short exposures produce streaks.)