From jwf@ipac.caltech.eduMon Apr 14 15:17:00 1997
Date: Thu, 10 Apr 1997 12:04:19 -0700 (PDT)
Subject: WG Mtg #120 Minutes

           IPAC 2MASS Working Group Meeting #120 Minutes

Attendees: R. Beck, R. Cutri, T. Evans, J. Fowler, L. Fullmer, T. Jarrett,
           G. Kopan, B. Light, H. McCallon, S. Wheelock, J. White


1.) Telescope Alignment
2.) 2MAPPS Dry Run
3.) Position Reconstruction Biases
4.) DBMAN Processing
5.) QUALITY Input Files
6.) DLT Hardware Move


1.) Telescope Alignment

    R. Cutri reported that the alignment of the northern-hemisphere
telescope's optical axis with respect to the declination-drive axis was now
close enough to perpendicular that further alignment may be unnecessary. The
misalignment is currently 243 arcsec. Within a couple of degrees of a pole,
the pixel subsampling will not be very good unless the telescope control
program is allowed to use the pointing model to do a lot of work to compensate
for the misalignment. To align more nearly to perpendicularity would require
tilting the primary mirror. A general request for input on whether this is
needed has been issued.

2.) 2MAPPS Dry Run

    The 2MAPPS dry run scheduled for today is underway as the meeting goes on.
R. Beck reported that the DARKS subsystem is running within the pipeline for
the first time; this was tested yesterday and appeared to work correctly, but
a thorough verification has not yet been carried out. G. Kopan reported that
the PIXPHOT NAMELIST file being used is not the one that nulls out the 
position offsets between FREXAS extractions and PROPHOT extractions (see item
3 below and the addendum to these minutes); since the dry run had started 
recently, R. Beck cancelled it and restarted it with this file.

3.) Position Reconstruction Biases

    The position bias problems reported in the last two meeting minutes have
been studied further with some significant progress. Three sources of position
bias have been identified. G. Kopan has found that the zero points implicit in
the FREXAS PSFs and those of PROPHOT are slightly different and give rise to a
constant direction-independent position bias between the point sources used
for frame-offset computation and those intended for the catalog. Gene provided
a report which is appended to these minutes; this report shows that these
biases can be controlled and therefore nulled out.
    A second source of position bias has been found by H. McCallon. This bias
is direction-dependent and stems from an algorithm in which a linear
correction is fit to the scan and turns out to need to be iterated. Just two
iterations is enough to eliminate any bias between source positions from scans
of opposite direction at the point in the pipeline between MAPCOR and
BANDMERGE. After BANDMERGE, however, a residual bias is seen which is about
half the previous amplitude. J. Fowler and H. McCallon will investigate the
position-related code in BANDMERGE to see if the origin of this bias can be
found. J. Fowler reported that he has checked for biases after position
refinement and found none within the U-Scan coordinates used throughout
BANDMERGE, and so the suspect area is the transformation between band-scan
coordinates and U-Scan coordinates.

[Note added in proof: It was determined that the usage of the band-scan-to-U-
Scan transformation parameters in BANDMERGE was obsolete. BANDMERGE was the
first code to include this transformation and had to be written before work on
POSFRM could be undertaken by H. McCallon. Installation of the up-to-date
prescription for this transformation has eliminated the last bit of position
bias. It is encouraging to note that the sum of these three small sources of
position bias would have been eliminated to a satisafactory extent by the 
POSPTS final position solution, since the contributors were all either 
constant or approximately linearly developing over the scan.]

4.) DBMAN Processing

    T. Evans reported that DBMAN is now operating in the pipeline as of the
dry run. Since interfaces with DBMAN are now no longer hypothetical, she
requested that close coordination with her be followed when any interface
changes affecting the catalog need to be made. One such change coming up in
the next few weeks involves the association parameters in the point source
records after POSPTS processing. This affects the SIS's POS05 and GAL08.
    T. Evans also requested that subsystem engineers whose products go into
the data base access such products therein and check for accuracy of
representation. She will distribute information via email on how to perform
such data base accesses. 

5.) QUALITY Input Files

    L. Fullmer reported that she is still not receiving QUALITY input files
from most subsystems. Discussion revealed that most subsystems are producing
files for input to QUALITY, but the wrapper scripts have not been updated to
move these files to the proper location. Linda will distribute information on
where to move files for scan-oriented and date-oriented quality checking.

6.) DLT Hardware Move

    J. White reported that attempts to use the DLT hardware on rodan have
encountered some problems (the DLT hardware was moved from lugosi to rodan
yesterday to support the dry run). A simulated observatory tape that was
processed correctly by the DLT hardware on lugosi is not able to be processed
correctly on rodan. As a result, the dry run has not be able to begin at the
DLTLOAD program, and the output from that point has been set up by hand.


From:    G. Kopan
Subject: frexas/prophot position biases

In investigating the bias between ascending and decending scans,
a bias was found between the frexas and prophot source positions. This bias
was traced to differences in centering of the respective psfs in the 
protocamera psfs of frexas and prophot. A bias of 0.025+/-0.01 pixel in
cross-scan and 0.076+/-0.015 pixel in in-scan was found, but in the same
direction independent of scan direction. A test run was made on the M92
scans with corrections in pixphot for these bias was made, and as expected,
the bias between ascending and decending scans was not fixed, although 
all sources were moved in the same direction by the expected amount. These
corrections will be included in future runs on the protocamera data.

Special attention will be given so that these biases will be removed
when the survey camera data psfs are created. Checks will be made on the
results, and provision will be made to remove any residual biases so that 
the r1 and r2 source positions will agree with each other and with the 
coadded images.