===========================================================================
COSMOS HST/NICMOS Data						Version 2.5
===========================================================================

This collection contains the HST/NICMOS Cycle-12+13 data for the COSMOS 
project, a 2-square degree survey of an equatorial field using primarily 
the ACS instrument aboard HST. The survey encompasses some 600 telescope 
pointings, arranged in roughly a square centered on the position:

	RA  =  10h 00m 28.6s
	DEC = +02d 12m 21.0s

The NICMOS data collection consist of parallel observations of the COSMOS
field using the F160W (HU) filter and the NIC3 camera. These data are from
cycle 12 + 13 HST observations. Note that the NICMOS parallel imagery do not
provide contiguous coverage of the COSMOS field.

The data have been reduced by James Colbert using a combination of IRAF/STSDAS
routines and his own code.

The files here follow the standard COSMOS naming scheme, e.g.

	nicmos_H_sci_100142+0223_EXT_25.fits

refers to an HST/NICMOS H-band (F160W) science image image centered at
RA=10:01:42 and DEC=02:23, version 2.5. EXT will either be "_sci_" for
the science images or "_wht_" for the weighting images.
	
There are 474 COSMOS NICMOS fields provided. Each is a combination of
at least two dithered images at the same pointing, which we chose as
the minimum criteria for inclusion. There were also 24 single data
frame fields that are not included in this data set, as the total lack
of pixel redundancy made reliable data reduction impossible. Of those
474 images, 203 are from Cycle 12 and 271 are from Cycle 13. All Cycle
12 fields overlap with ACS, while 216 of the Cycle 13 fields do. The
fields have been registered to the ACS images where possible. The 55
remaining fields were matched using Subaru I-band astrometry.

All data images have been set to the same size, 556 x 556
pixels. Extra area around the outside edge, which contains NO data,
has been set to the value of 0.0. The pixel scale is 0.101 arcsec,
making each image 0.88 square arcminutes in size although that
includes the regions with no data. The central "usuable" region tends
to be more like 0.7 square arcminutes.

The images are in the units of counts/sec. A direct conversion to
magnitudes, not accounting for any aperture, gives F160W zero points
of 23.172 (AB) or 21.836 (Vega). This can be derived by the user
themselves directly from the PHOTOMETRY KEYWORDS in the header, using
the methods described in Section 5.3 of the NICMOS DATA HANDBOOK:
http://www.stsci.edu/hst/nicmos/documents/handbooks

Each NICMOS image is a combination of 2-8 frames of 512 seconds
each. The keyword NDRIZIM tells you exactly how many images went into
each image. The majority of combined integration times for each field
are either 1024 (18%), 1536 (49%), or 2048 (31%) seconds, with only 12
fields imaged for longer than a single orbit.

If the user is interested in the exact input files used to create each
image, they are identified in the image header as D00xDATA, where x is
the order the images were processed. I.e., D003DATA would be the third
file that was combined. However the extension on that file varies
between several internal pipeline file extensions ("_flt2", "_flt",
"_med"), none of which are available from HST. We took the "_cal" file
that is available from HST and applied a SAA correction, a quadrant
offset correction, a sky subtraction, and a median filter to get to
the input file listed in the header.
 
WARNING: Two fields were unusual, but have been included for sake of
completeness.  One should only use them with caution:

095814+0127: No dither between images, making it almost impossible
to get rid of most bad pixels and other artifacts.

100005+0254: The images are smeared out a bit in one direction,
cause unknown (vibration of some sort?). Fluxes may be all right.

Version History

1.0  Cycle 12 data only
2.0  Initial Cycle 12+13 images. Fixed over-clipping issue in some
images. Created more uniform format and size.
2.5  Properly registered to ACS and Subaru data. Fixed a couple of
images that had been  incorrectly reduced.

===========================================================================
For more details on the data, including exposure times, filter
choices, etc., as well as on the COSMOS project as a whole, see the
COSMOS web pages at:

	http://cosmos.astro.caltech.edu/

The COSMOS Team, Patrick Shopbell (Archive Lead), and James Colbert
(colbert@ipac.caltech.edu).
July, 2008