|
•
|
The pixel position on the drizzled images ( drz.fits) created by multidrizzle which corresponds to an undistorted pixel position on a tangent plane projection of the sky.
|
A pixel position on a drizzled image (drz.fits) may be transformed to a position on the celestial sphere (RA, Dec.) using the task
xy2rd (in the
stsdas.toolbox.imgtools package). There is a corresponding task,
rd2xy, that transforms a RA, Dec. position to a pixel position on a drizzled image. (These tools are only meant for drizzled images; they cannot be used for
raw.fits and
flt.fits files as they do not include the very large effects of geometric distortion).
The transformation between pixel positions on a distorted image (raw.fits or
flt.fits) and drizzled (
drz.fits) image may be performed using the
dither package tasks
traxy and
tranback; these tasks implement the same geometric mapping as
drizzle and
blot, respectively. For more details on how to use them, the
Multidrizzle Handbook or the
dither package help files should be consulted.
However, there is a more convenient high-level wrapper for determining pixel position transformations between distorted and undistorted images. The
tran task in the
dither package, which was released as part of the late-2004 version of STSDAS, allows the mapping of X,Y positions between
flt.fits and
drz.fits images. This task requires the following to enable processing:
|
•
|
A set of geometric distortion coefficients files (“coeff”). These are created when multidrizzle is run to combine a set of flt.fits images on your home machine. ( “coeff” files are not available from the archive or from the STSDAS Web site; they can only be created by running multidrizzle on the flt.fits files on your local machine.)
|
|
•
|
In order to run multidrizzle, you need to retrieve the IDCTAB distortion reference table (idc.fits) and the DGEOFILE distortion correction reference image ( dxy.fits) associated with those images. These filenames are given by the IDCTAB and DGEOFILE image header keywords, and they can be downloaded from the ACS Reference Files Web page .
|
Work is now underway at STScI to substantially simplify and improve the way astrometric information is represented and used by MultiDrizzle. This should allow users to more easily handle astrometry, and align images to each other and to external catalogs.
Please check the ACS Web site and
STSDAS MultiDrizzle Web site for updates.
Example: run tran to translate the x,y position on a
flt.fits file to its location in a
drz.fits file. This task requires that
multidrizzle has processed the
flt.fits images on your local machine, in order to use the coefficient files (
coeffs*.dat) generated by
multidrizzle.
# List of input flt.fits images for multidrizzle, and resulting drizzle-combined image.
|
The reverse operation, translating a position on a drz.fits image to a flt.fits image, can be applied as follows:
Finally the xytosky task, which is part of the
dither package (as of May 2011), will convert a pixel position on a distorted
flt.fits file directly to a sky position, by applying the distortion correction from the
IDCTAB reference table and using the world coordinate information from the header. It may be used as follows:
This task doesn’t require the coefficient files but, like multidrizzle, it requires a copy of the IDCTAB to be available. Both
tran and
xytosky have options for lists of positions to be supplied as text files to allow multiple positions to be efficiently transformed. The task
xytosky does not currently support DGEO
FILE reference images. If accuracy at a level better than 0.1 pixels is needed, then we recommend the following:
run
tran to get the corresponding X,Y pixel position on the drizzled (
drz.fits) image, followed by
xy2rd.
The astrometric information in the header of an ACS image comes indirectly from the positions of the guide stars used
during the observations. As a result, the absolute astrometry attainable by using the image header world coordinate system directly is limited by two sources of error. First, the positions of guide stars are not known to better than about 0.3 arcseconds. Second, the mapping from the guide star to the instrument aperture introduces a smaller, but significant error.
Although absolute astrometry cannot be done to high accuracy without additional knowledge, relative astrometry with ACS is possible to a much higher accuracy. In this case the limitations are primarily the accuracy with which the geometric distortion of the camera has been characterized. This is discussed in detail in the
Multidrizzle Handbook. With the recent inclusion of a time-dependent skew into the model of the ACS distortion used by MultiDrizzle, typical accuracy in aligning two ACS images is a few hundredths of a pixel or better.
Accurate astrometric measurements, especially for faint sources, should take into account effects of CTE, as described in
ACS ISR 2007-04. The institute is monitoring the variations of the linear skew terms and updating the corresponding astrometric reference files derived from the above-mentioned ISR.
The normal guiding mode uses two guide stars that are tracked by two of HST’s Fine Guidance Sensors (FGSs). O
n some occasions, when two suitable guide stars are not available, single-star guiding is used with the telescope roll controlled by the gyros. These observations will suffer from small drift rates. To determine the quality of tracking during these observations, please refer to the
Introduction to the HST Data Handbooks for information about jitter files.
In single guide star guiding, typical gyro drift rates produce a roll about the guide star of 1.0 - 1.5 mas/sec., which in turn introduces a translational drift of the target on the detector. This roll is not reset and continues to build over multiple orbits and reacquisitions, until the next full guide star acquisition.
The exact size of the drift depends on the exact roll drift rate and distance from the single guide star to the target in the HST field of view. Additional information is available in
ISR TEL 2005-02. For ACS with single-star guiding, the typical and maximum drift rate of the target on the detector are shown in
Table 5.7.
The drift over an orbital visibility period can be calculated from the values in Table 5.7. The typical visibility period in an orbit (outside the Continuous Viewing Zone [CVZ]) ranges from 52 to 60 minutes, depending on target declination. The drifts inherent to single-star guiding are
not represented in the image header astrometric information, and have two important consequences:
|
•
|
There will be a slight drift of the target on the detector within a given exposure. For the majority of observations and scientific applications this will not degrade the data (especially if the exposures are not very long). The drift is smaller than the FWHM of the point spread function (PSF). Also, the typical jitter of the telescope during an HST observation is 0.003 - 0.005 arcsec, even when two guide stars are used.
|
|
•
|
There will be small shifts between consecutive exposures. These shifts can build up between orbits in the same visit. This will affect the MultiDrizzle products from the pipeline because it depends on the header WCS (predicted) positions to determine image offsets when combining dithered images. Therefore, the structure of sources in the image will be degraded during the cosmic ray rejection routine. This problem can, however, be addressed during post-processing by running multidrizzle, off-line, using manually-determined image offsets.
|
Even when two guide stars are used, there is often a slow drift of the telescope up to 0.01
arcsec/orbit due to thermal effects. So, it is generally advisable to check the image shifts, and if necessary, measure them to improve the alignment of exposures before running
multidrizzle off-line to perform the cosmic ray rejection and image combination.
In summary, for most scientific applications, single-star guiding will not degrade the usefulness of ACS data, provided that the shifts are measured post-facto and
multidrizzle is re-run offline using these shifts. However, we do not recommend single-star guiding for the following applications:
Observers who are particularly concerned about the effect of pointing accuracy on the PSF can obtain quantitative insight using the
TinyTim software package. While this does not have an option to simulate the effect of a linear drift, it can calculate the effect of jitter of a specified RMS value.
