This section describes each of the FOC calibration steps, how they were determined, and how the pipeline task calfoc carries them out. Pipeline calibration of FOC imaging data:
- Dezooms zoom-mode images to produce square calibrated images.
- Determines the absolute sensitivity of the instrument configuration and sets photometry keywords allowing count rates to be converted to flux units.
- Corrects the geometric distortion of the image via interpolation of the data onto a rectified grid, creating a .c0h file.
- Applies a flatfield correction to the data that removes large-scale spatial non-uniformities, creating a .c1h file.
Figure 6.1: Flowchart of the Calibration Process for FOC Data
To determine the calibration steps applied to the data and the calibration reference files used to calibrate the data, look at the values of the calibration switches in the header of the raw (or calibrated) data. Before calibration, the calibration switches will have the value OMIT or PERFORM. The calibration process sets the switches for completed steps to COMPLETE in the header keywords of the calibrated data file.
|Switch||Processing Step||Reference File|
Remove instrument background by subtracting a dark count image; it is never done, for reasons given in the section titled "Background" on page 7-11.|
Multiply by format-dependent inverse flatfield;|
it is never done, as will be explained in "Format-Dependent Effects" on page 7-9.
Dezoom zoomed pixels by splitting each zoomed pixel in the sample direction into two pixels, each having half the flux of the original. Done only (but always) for data taken in zoom mode. Produces square pixels.|
Compute absolute sensitivity using throughput tables appropriate to observation mode (PHOTMODE). Names of actual throughout tables used are determined from graphtab and comptab tables. Names of throughput tables used are written to history section of calibrated data header. This step does not alter pixel values, it writes inverse sensitivity (PHOTFLAM), RMS bandwidth (PHOTBW), zero point magnitude (PHOTZPT), pivot wavelength (PHOTPLAM), and observation mode (PHOTMODE) to header of calibrated data.|
graphtab and throughput tables|
Perform geometric correction to rectify optical and detector distortion using geometric correction reference file.|
Correct for large scale detector non-uniformity by multiplying by the uniform detector efficiency file, which is reciprocal of a highly-smoothed flatfield. Done only for images.|
Flatfield and compute absolute sensitivity for spectrographic data. At the moment, this step is not done.|
6.2.1 Dezooming of Zoomed Images (PIXCORR)A somewhat unfamiliar aspect to using the FOC is that the pixel size can be doubled in the x direction, with a corresponding increase in the field of view and a decrease in the horizontal resolution. This process is known as zooming. If an image has been taken in zoom mode, the first processing step is to invert this zooming process by splitting the data values along the first image axis (the sample direction). The length of the first axis (NAXIS1) is doubled, and the length of the second axis (NAXIS2) remains unchanged. If the zoomed image contained n rectangular pixels (50 x 25 microns each) in the sample direction, the dezoomed image contains 2n square pixels (25 x 25 microns), each with half the flux of the original rectangular pixel. No attempt is made to do anything more sophisticated. The keyword PXLCORR is set to COMPLETE when this step is done, and to OMIT when the image is taken using normal pixels.
6.2.2 Absolute Sensitivity Correction (WAVCORR)This step does not modify the data itself, but instead computes a constant that can be used to convert the data values in the .c1d file to absolute fluxes. This constant is saved in the .c1h header file as the value of the PHOTFLAM keyword. The keywords that describe the absolute sensitivity (PHOTFLAM, PHOTMODE etc.) are derived using SYNPHOT (see page 3-16) applied to the photometric mode calculated using the instrument parameters. The photometric mode now includes the effect of format-dependent sensitivity (since May 18, 1994).
The sensitivity curve for the f/96 camera, often called the Detector Quantum Efficiency (DQE) curve, was derived from observations of a spectrophotometric standard through many of the medium and narrow band filters spanning the useful wavelength range of the detector. The DQE curve is combined with the filter transmission curves to derive the PHOTFLAM values or with SYNPHOT to convert measured counts into absolute flux values. The DQE derived for the f/96 relay is based on the flux that falls in a 1" radius aperture. This aperture size does not encompass all the flux from the star, especially in the UV. Note that this definition of the DQE treats all side diffracted or scattered light that falls outside the aperture as lost. If you wish to apply the DQE to different apertures or other photometric methods, you should normalize your results to that aperture size using an appropriate PSF (see "Point Spread Function" on page 8-2).
The f/48 detector, for various reasons, has not been adequately calibrated in orbit. Only the pre-launch DQE curve has been used for the f/48 camera. On-orbit measurements taken in December 1993 showed that the measured fluxes from a spectrophotometric standard were about 60% of those expected. The f/48 curve was not updated with this information.
Multiplying the data values in the calibrated image by the value of PHOTFLAM and dividing the result by the actual exposure time (EXPTIME) converts the values to flux density Fλ in units of ergs cm-2 s-1 Å-1. The current CDBS filter transmission, mirror reflectivities, and detector quantum efficiency curves are used to compute the conversion factor (PHOTFLAM) between detected count rate and a source flux Fλ averaged over the bandpass. The pivot wavelength, rms bandwidth, and zero-point magnitude are also saved in the header as the values of PHOTPLAM, PHOTBW, PHOTZPT, respectively. Finally, the observation mode (PHOTMODE) is written to the header, and this mode is used by SYNPHOT to determine the inverse sensitivity.
Status of Sensitivity FilesAll COSTAR-corrected data taken before October 22, 1994, used the predicted DQE curve for the FOC+COSTAR sensitivity, with the measured DQE curve being applied to images taken after October 22, 1994. Therefore, the PHOTFLAM keyword will be incorrect for images taken prior to that date and should be recalculated if needed for data analysis. The actual files used to calculate these keywords are recorded in the HISTORY records at the bottom of the .c0h header file. The DQE file appropriate for COSTAR-corrected f/96 observations is foc_96_dqe_004.tab.
The DQE file appropriate for pre-COSTAR f/96 observations is foc_96_dqe_003.tab.
6.2.3 Geometric Correction (GEOCORR)This correction removes both the optical distortion that arises in the telescope and the detector distortion produced by the electronic imaging system of the FOC. Optical distortion occurs upstream from the detector itself and arises primarily from the off-axis position of the FOC. Detector distortion occurs within the FOC's electronic imaging system, which consists of a three-stage image intensifier optically coupled to an Electron Bombarded Silicon (EBS) target TV tube.
Both the image intensifiers and the TV-camera section of the image system contribute to detector distortion. Intensifiers rely on an electric field for accelerating, and a magnetic field for focusing photoelectrons, and any irregularities in the uniformity of either results in distortion. The source of distortion within the target or TV camera section arises from the scanning of the target. This scanning distortion is due primarily to variations in the speed of the scanning beam at the ends of the sweep (where it must change direction), and the fact that the beam carries out an angular sweep across a plane target, with imperfections in the scanning electronics adding secondary effects. For these reasons, each video format has its own peculiar distortion characteristics, so the distortion measured for one format cannot be used directly to correct an exposure taken in a different format.
To facilitate correction of geometric distortion, reference points called Reseau marks were etched onto the first of the bi-alkali photocathodes in the intensifier tube. These Reseau marks form an orthogonal grid of 17 rows and 17 columns with a separation of 1.5 mm (60 pixels), each Reseau being 75 microns (3 x 3 pixels) square. The detector distortion was originally determined by illuminating the photocathode with an internal light source, i.e., an internal flatfield. The observed positions of the Reseau marks, when compared to the expected positions, provided a map of the detector distortion across the field. The optical component of the distortion was determined independently from ray-tracing models of the HST and FOC optics and was applied to the reference Reseau grid to give the expected positions.
Unfortunately, the detector distortion for the FOC clearly showed variations on spatial scales smaller than the spacing of these Reseau marks, particularly near the scan line beginning, and therefore models based only on the Reseau marks inadequately represent the true distortion. A new method of determining distortion based on overlapping observations of crowded starfields was developed to determine the net distortion (the optical distortion is naturally folded into this new method). These observations yielded a two-dimensional spline distortion model from which the new geometric correction files were generated. The new scheme removes distortion by transforming each pixel in an undistorted image to a quadrilateral virtual pixel in the distorted image using the derived distortion model. The flux within the distorted pixel is then calculated as the sum of the contributions from each pixel the distorted pixel covers, where the weightings are simply the areas common to distorted and undistorted pixels. This procedure is illustrated in Figure 6.2. Because the transformed pixels fit together with no gaps and cover the distorted image completely, the method is rigorously flux-conserving. The improvement in quality is most apparent for smaller formats where the small number of visible Reseau marks prevented the determination of a good model.
Figure 6.2: Pixel Transformation
If you need the finest possible spatial resolution, bear in mind that this method applies a position-dependent smoothing to the data. You might find working with the raw uncorrected data more profitable if you need to preserve every detail.The keywords PXFORMT, SAMPPLN, LINEPFM, SAMPOFF, and LINEOFF indicate the format in which the image was taken and therefore determine the appropriate geometric correction file. Geometric correction files exist for most formats listed in Table 4.2 and Table 4.3. The keyword GEOCORR tracks the execution of this step and is set to COMPLETE upon creating the .c0h file. In addition, the keyword GEOHFILE lists the geometric correction file that was applied to the image, which can be useful in making sure that the proper correction was applied.
Status of GEOHFILEThe new geometric correction files have been used in the calibration pipeline for f/96 data since March 19, 1995 (f/48 geometric correction files are still based on Reseau marks). Only observers who want sub-pixel accuracy in position measurements or those who have used the 256 x 256 format should even consider reprocessing their old data with the new geometric correction files. For most observers, the improvements will not significantly affect positions or photometry.
6.2.4 Flatfield Correction (UNICORR)This correction is referred to as the uniform detective efficiency (UNI) correction. It attempts to remove the effects of non-uniform efficiency of the detector, and its complicated name is really just another way of saying "flatfielding." The procedure first selects the appropriate correction file on the basis of wavelength. The pivot wavelength of the bandpass (OTA + filters + detector) is used to select the correction file with the closest wavelength (in the geometric sense). The UNI correction files are 1024 x 1024 images from which the appropriate sub-image is extracted to match the image format of the science image. For example, a science image taken in the standard centered 512 x 512 format will use the center 512 x 512 of the appropriate UNI correction file whereas a 512 zoomed x 1024 format science image will use the whole UNI correction file (recall that the science image has been dezoomed in the pipeline process). These files are heavily smoothed to correct large-scale features extending more than 20 pixels and are geometrically corrected with the current geometric correction file. The calibration is then performed by multiplying the science image by the UNI sub-image given by the keyword UNIHFILE. The UNICORR keyword is then set to COMPLETE when this step is finished, and OMIT if it is not executed.
Status of UNIHFILEThe current UNI files are derived from the same observations that produced the pre-COSTAR corrections; external observations taken at 1360Å, and internal flatfields at 4800Å, 5600Å, and 6600Å. The difference between the pre-COSTAR and post-COSTAR UNI files lies in the geometric correction. The installation of COSTAR changed the optical distortion of the f/96 field, which is rectified in the geometric correction step. The latest UNI files, for COSTAR-corrected f/96 observations, are the pre-COSTAR UNI files corrected with the post-COSTAR geometric correction file.
The pre-COSTAR UNI files are derived from pre-COSTAR observations taken at 1360Å, 4800Å, 5600Å, and 6600Å, to which we have applied the pre-COSTAR geometric correction file.
Copyright © 1997, Association of Universities for Research in Astronomy. All rights reserved.
Last updated: 11/13/97 16:43:34