Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.stsci.edu/ftp/science/hdf/archive/shortcuts.html
Дата изменения: Thu Feb 13 21:56:48 1997
Дата индексирования: Sun Dec 23 00:44:45 2007
Кодировка:
Поисковые слова: annular
The Hubble Deep Field -- version 1 shortcuts

The Hubble Deep Field Version 1: Shortcuts in the Data Processing:

The version 1 data reflect processing up to January 15, 1996. A number of shortcuts were taken to make these data available within three weeks of the last observation. These are outlined below.

  • Hot pixels were masked once and for all. Some of these will be later recovered with a time-dependent hot pixel mask.

  • Except for F300W, most frames with a visible scattered light pattern were not used. This constitutes about 25% of the observing time in F450W and F814W; we expect the limiting depth to be about 0.1 mag fainter when those frames are included in the final stacks.

  • Moving targets are noted in the log sheet, but no special procedures have been followed to remove them. In many cases these streaks are not completely removed by the cosmic-ray rejection software. They are not individually visible in the final drizzled images, but nevertheless contribute to the overall noise.

  • Images have been combined with exposure-time weighting. This is close to optimal for F606W and F814W, which are generally sky-dominated, but is not optimal for F300W and F450W. This will be changed in version 2.

  • Small rotations between the WFPC chips have not been removed.

  • Also not included in the drizzled images are shifts introduced by the filters and variations in pixel scale among the WF chips. These cause misalignments of up to several tenths of an arcsec when comparing images.

  • There are lingering doubts about the image registration at a level of about 0.02 arcsec. The doubts involve image stacks that were nominally at the same dither position, but had a guide-star acquisition or reacquisition between individual frames.

  • We have not completed the tests of the drizzling procedure on simulated data. Our tests to date suggest that there may be systematic errors of about 1.3% in the photometry in the final images. The simulations also indicate that the pixel-to-pixel variance is correlated on scales of up to about 0.3 arcsec, which is somewhat larger than anticipated and is not completely understood.

    Copyright © 1997 The Association of Universities for Research in Astronomy, Inc. All Rights Reserved.

    Harry Ferguson ferguson@stsci.edu 1/14/96