In this second weekly science and analysis working group meeting, we
continued with the status reports that we didn't get to last week.
- D. Kirkpatrick:
Davy has been comparing the NLTT
proper motion catalog of Luyten with the 92-94 2MASS database,
searching for positional matches. The purpose is to find high proper
motion (ie. very nearby) stars, and very low mass stars in the 2MASS
database and to apply the knowledge learned in the process to hunt
through the entire 2MASS database for previously unknown nearby stars
and low mass stars. He has obtained a match rate of 96% for the 329
Luyten stars within the areas scanned by the prototype camera prior to
the May 1995 run; all the unmatched stars have good excuses (large
positional uncertainties in the Luyten catalog; too faint to 2MASS).
From an R-Ks vs Ks color magnitude diagram for these stars (R band data
derived from matching to the APM catalog), Davy is able to identify a
handful of new candidate nearby (D<25pc) stars which should be followed
up at the telescope.
Davy and Chas have completed the clean-up and bandmerging of the 92-94
data, and T. Evans reported that this is now in sybase and will be
installed in the 2MASS-private version of XCATSCAN by B. Narron next
week.
- C. Lonsdale:
Carol is studying the robustness of
the various GALWORKS galaxy photometric measurements in the 1995 Coma
and Hercules cluster data, with a view to understanding how 2MASS can
best derive Tully-Fisher magnitudes. She has compared the circular
aperture 2MASS H band growth curves for the approximately 120 galaxies
that appear in the RC3 to standard H band growth curves derived by
Tormen and Burstein (1995) from the accumulated Aaronson et al
Tully-Fisher data set. The 2MASS curves for these relatively bright
galaxies behave well, even though a significant proportion of them are
flagged as having been affected by nearby stars. The analysis brought
to light an error in the adaptive circular aperture magnitudes derived
by GALWORKS, which has now been corrected for the Coma data set by T.
Jarrett. Tom has also added a new isophotal measurement at 21st
magnitude per square arcsecond, to complement the existing 20th
magnitude per square arcsec measurement. Carol has a few more analyses
with this data in mind before she writes up a detailed memo for
circulation.
- T. Chester:
reported on the investigation he and
T. Jarrett have made into the idea of using a super-coadd for more
complete galaxy detection and characterization. The two Toms are on
the verge of distributing a memo with full details; the highlights are
that (1) the best improvement in SNR that one can expect using a
super-coadd (a coaddition of the J,H and Ks coadds) is slightly better
than sqrt(2) because galaxies are always detected at significantly
higher SNR at J and H than they are at Ks. The actual improvement one
can reach depends on the relative weighting of the three images, which
have different SNRs, and galaxy color. The process becomes useless at
high extinction levels because the SNR at J becomes less than the
originally lower SNR at Ks; above a certain extinction level
super-coadds actually do worse than single Ks band detection.
- T. Jarrett and S. Terebey:
Sue and Tom have been
looking at the efficiency of extended source detection of GALWORKS in
the '95 rho Oph data. GALWORKS misses bright stars with nebulosity
because its thresholds are set to avoid bright stars. Tom, Roc and Sue
came up with a new algorithm using an annular region around the star to
try to recover these 'extended' stars, and Tom has had some success
with it for ~6th magnitude stars and fainter. Sue reported that
GALWORKS mistakenly extracted many false extended sources in the
nebulosity in rho Oph; most of them single pixel blips in a background
of extended emission. Tom J. came up with a new rejection filter for
GALWORKS, based on this discovery, that successfully identified and
rejected these false sources.
A issue that came clearly to light in this analysis is that coadds in
regions of nebulosity do not mosaic well - the background levels are
set differently from coadd-to-coadd, resulting in a patchwork mosaic.
This is because the average background value assigned to a given region
can be strongly affected by the nebulosity in that region. A better
method would be to match the background between successive coadds such
that the average background matches for the nearest halves of each
coadd, instead of matching the average background of the entire coadd.
- B. Light and S. Wheelock:
Bob and Sherry showed a
coadd image with a couple of peculiar faint transient images (present
on odd-numbered scans but not even-numbered) close to a brighter star
that they could not find a definitive explanation for. It turned out
that there was a bright persistence image also present in the coadd
from a bright star, and that both of the fuzzy objects seen in the odd
scans were due to persistence from another star slightly above and to
one side of the bright star that left the brighter residuals seen.
That second star produced another set of persistence features that were
nearly coincident with the first, due to the second star having nearly
the same X and almost exactly 2 in-scan steps difference in Y.
This curious circumstances with its difficult diagnosis points out the
desirability of carrying EVERY detection through the whole pipe,
flagging bad objects as seen. Having all of this in a single file
would make the job of sorting out situations like this much easier.