HST Data Handbook for WFPC2

4.2 Flatfield Correction Errors

---

This section summarizes the quality of the various flatfield versions that have been used in the pipeline and, where possible, provides suggestions for improvements. Note that images taken at different times will typically use different flats (i.e., the latest flatfields are not the best flatfields for data taken at all times). However, the flats tend to be stable and changes over time are 2% at most, excluding small-scale features such as dust spots. As always, observers can obtain the most up-to-date calibration by re-requesting their science images from the Archive. An alternative is to check whether the flatfield files used by calwp2 fall into one of the specific cases outlined below before submitting an Archive request.

Cursory information on the flatfielding performed can be found in the calibrated science image header keywords and HISTORY records. Flatfielding Linear Ramp Filter Images provides an example of a linear ramp filter image header: the flatfield correction switch was set to PERFORM in the raw header, but since a dummy flatfield was used in the pipeline during processing, the correction was effectively skipped and is noted as such in the calibrated header and HISTORY comments. More detailed information concerning the generation of the flatfields can be found in the flatfield reference file image header HISTORY comments.

Pre-1995 flatfields

WFPC2 data processed during the first few months of the mission were calibrated using flatfields created from pre-launch ground test data, which did not take into account the large-scale structure of the flatfield. These flats have rootnames¹ that begin with "d" or "e1" and if they have the keyword, PEDIGREE (many predated the existence of this keyword), it has been set to GROUND. Flatfields installed in CDBS prior to launch were obtained purely from data taken during the pre-launch Thermal Vacuum test, and thus do not reflect properly the illumination function typical of the HST OTA; in this case, zonal errors of several percent can be encountered.

In March 1994, the WFPC2 IDT (Instrument Definition Team) delivered an updated set of flatfields. These flats have rootnames beginning with "e3" and are a combination of pre-launch calibration data and early on-orbit EARTH-CAL images (observations of the bright Earth). These on-orbit images were flatfielded with the pre-launch data, then stacked and smoothed so that a map of the OTA illumination pattern could be developed. This map, which included the chip-to-chip normalization, was then applied to the ground flatfields and the resulting flats installed in CDBS. Though they are a combination of ground and on-orbit data, the PEDIGREE of these early flatfields is also set to GROUND. The HISTORY comments at the bottom of the flatfield reference files (.r4h) provide more details on their generation.

Reprocessing by OTFR will provide improved flatfielding for any images calibrated with the pre-1995 flats.

Flatfields from late 1995 / early 1996

A completely new set of flatfields were delivered between late 1995 (for the filters used in Hubble Deep Field observations) and early 1996. These flatfields are more accurate than the previous ones, although users may not necessarily need to recalibrate or re-retrieve data through OTFR as a result of this change. The new flats, with rootnames beginning with "g" and having a PEDIGREE of INFLIGHT, are effectively the preflight ground data flatfields that include an improved on-orbit illumination pattern correction applied to scales larger than 7 pixels. In the optical (400nm to 700nm), the new flats differ from the old by 1% or less over the vast majority of the chip, with differences growing to about 8% at all wavelengths in the outer 50 pixels of the chip. Longward of 850 nm, differences of up to 1.5% are seen across the main body of the chips; and shortward of 300 nm, the differences between the old and new flatfields are less than 3%. As always, the HISTORY comments at the bottom of the flatfield reference files (.r4h) provide more details on their creation.

Flatfields after 1996

The flatfields for WFPC2 are in the process of being updated (November 2001); these improvements are the result of on-going programs of WFPC2 ``EARTH-CAL'' observations (images of the bright Earth) that have made it possible for flatfields to be created for several different epochs. Depending upon the date of their final installation into CDBS, the new flatfields will have names beginning with the letters 'l' or 'm' (e.g., "l12" or "m1"). The availability of these new flatfields will be announced in the WFPC2 Space Telescope Analysis Newsletter (STAN), on the WFPC2 Advisory page, and via an Instrument Science Report.

The 1995 epoch flatfields, discussed in the previous section, will still be appropriate for data taken before or during 1995; however, the new flatfields will be appropriate for two new epochs. The first will cover September 1995 to November 1996, during which several strong dust spots appeared. The second covers November 1996 to May 2001, which was a relatively stable period. The new flats generally differ by less than 1-2% from the previous generation in terms of large-scale structure variations across the chips; however, the time-dependence is more pronounced for small-scale features and will allow these to be removed more effectively than with the old flats. In addition, the new flatfields offer improved pixel-to-pixel rms fluctuations of around 0.3% or less. Few users would need to recalibrate their data as a result of these changes; however, any images dating from September 1995 to May 2001 that are retrieved from the Archive after the installation of the new flatfields will automatically be calibrated with the new files.

Flatfields for special purpose filters

Flatfields for infrequently-used filters, such as the polarizers, have generally been available since late 1996 and have rootnames beginning with "g". Other filters, such as the Woods filter (F160BW) and some of the other UV filters, have flatfields with significant noise and other errors; in these cases, the application of a post-pipeline correction flat can provide significant improvement (see section 3.5.2). An alternative would be to use a new pipeline flatfield (it is possible that new flatfield reference files may be available for these filters in early 2002), as discussed in the previous section. Finally, note that by design, linear ramp filters do not have a flatfield applied during pipeline calibration; observers with linear ramp filter data should perform some post-pipeline processing, as recommended in Flatfielding Linear Ramp Filter Images.

As always, the WWW Reference File memo or the Archive's StarView can be used to peruse the available flatfields.

The effect of WFPC2 flatfields on point source photometry

WFPC2 flatfields are defined so that a source of uniform brightness produces the same count rate per pixel across the image. However, due to geometric distortion of the image by the optics, the area of WFPC2 pixels on the sky depends on the location on the chip; the total variation across the chip is a few percent, with the largest changes occurring deep in the CCD corners. Therefore, the photometry of point sources is slightly corrupted by the standard flattening procedure. This effect, and its correction, are discussed in Geometric Distortion. The photometry and astrometry of point sources are also adversely affected by the 34th row effect, see The 34-th Row Defect for more details.

¹ The rootname of a reference file is based on the year, date, and time that the file was delivered to the Calibration Data Base System (CDBS). The first character in the rootname denotes the year, in base 36, since 1980 (i.e., files installed in CDBS in 1993 start with "d", those from 1994 start with "e") while the second and third characters are the month and day, respectively, in base 36.