Date: Wed, 10 Dec 1997 18:22:43 -0800 (PST)
From: jwf@ipac.caltech.edu
To: 2mass@ipac.caltech.edu
Cc: chas@ipac.caltech.edu, sstrom@donald.phast.umass.edu,
    stiening@ipac.caltech.edu
Subject: 2MASS WG Mtg #138 Minutes

           IPAC 2MASS Working Group Meeting #138 Minutes
                            12/09/97

Attendees: T. Chester, R. Cutri, D. Engler, T. Evans,
           J. Fowler, T. Jarrett, G. Kopan, J. Mazzarella,
           H. McCallon, B. Wheaton, S. Wheelock, J. White


AGENDA

1.) DARKS Design Changes
2.) FREXCLEAN Redelivered
3.) Southern Observatory Status
4.) Band-to-Band Position Discrepancies
5.) 2MAPPS Version 2.0 Expectations
6.) Data Base Status



DISCUSSION


1.) DARKS Design Changes

    A telecon was held earlier today to discuss design changes to
the DARKS subsystem. The participants were R. Cutri, J. Fowler,
G. Kopan, and M. Skrutskie. In the last year, a better
understanding of the hardware behavior has been obtained, and
some of the assumptions underlying the design of DARKS have been
found to be incorrect. These mostly involve the stability of the
dark-current behavior and the responsivity. Both have been
observed to drift on time scales of days to months, and the
J-band dark behavior exhibits sudden spontaneous jumps. The
drifts will require a sliding-window averaging of the
responsivity images, and the spontaneous dark jumps will require
more elaborate rejection tests for dark sequences. Another change
will be the freezing of the masks, which will henceforth be
changeable only through human intervention; this can be done with
a NAMELIST change. Different utilization of morning/evening
combinations of dark and flat sequences will be made, and some
quality-diagnostic output files will be modified or added. The
detailed descriptions of the changes will be available in the
DARKS SDS.


2.) FREXCLEAN Redelivered

    T. Evans reported that FREXCLEAN had been redelivered. The
new version updates not only the FRX01 files (see the minutes of
meeting number 137), but also the FRX02 and FRX03 files, which
contain information relating to the number of detections per
frame and per scan. These changes are needed by PFPREP and
QUALITY.


3.) Southern Observatory Status

    R. Cutri reported that the southern telescope had obtained
"first light" and that collimation was underway. All three chips
in the camera have been swapped out, and the J chip was replaced
yet again because of a bad column. When the camera was opened to
make these changes, it was discovered that the mounting screws
for the chips had no locking nuts, and one of the chips was
actually loose enough to wobble a bit.

4.) Band-to-Band Position Discrepancies

    H. McCallon showed the results of some analysis based on the
recently completed POSMAN summary-file processing capabilities.
These related to the band-to-band alignment anomalies discussed
in recent meetings. Although these results shed no light on the K
anomalies, they confirmed a theory proposed some months ago that
movement of the H chip was the cause of band alignment
variations. POSFRM solves for the band-to-band alignment for each
scan, and with the exception of two anomalous nights in which all
three bands showed mutual alignment anomalies, all available data
showed relatively consistent alignment between the J and K chips,
both of which varied significantly with respect to H.


5.) 2MAPPS Version 2.0 Expectations

    Version 2.0 of 2MAPPS is due on February 2, 1998. The goal of
this version is to be capable of processing all observations and
generating products that satisfy the Level 1 requirements,
without needing redeliveries before Version 3.0 at the end of
data acquisition. To do this, it is desired for all subsystem
capabilities currently in the design documentation to be
implemented, tested, and tuned in Version 2.0. Because of design
changes necessitated by unanticipated hardware problems, and also
because of periods of temporary reassignment of 2MAPPS software
developers to other projects, it does not appear feasible to have
all capabilities currently in the design implemented, tested, and
tuned by the nominal delivery date for Version 2.0, and so some
discussion of the more significant areas of concern was held.
    Two main areas were identified: (a.) the POSMAN capabilities
that have not yet been implemented, specifically, linear band-to-
band alignment change during scan, position uncertainty model
refinement, refraction modelling, and distortion modelling; (b.)
PSF modelling and activities dependent upon it.
    There appears to be no possibility of implementing all the
remaining POSMAN capabilities by next February, and the team
expressed doubt that all were necessary for meeting the Level 1
requirements. Prioritization was therefore in order, and the
decision was to implement the band-to-band linear alignment
change first. This might be the only addition to POSMAN possible,
but if time permits, the next task would be refinement of the
position uncertainty model. It was noted that the lack of this
refinement did not appear to be much of a threat to the Level 1
requirements, since the position uncertainties used by BANDMERGE
are dominated by source-extraction position errors (for which the
uncertainties appear to be reasonable for clean sources, if
somewhat overestimated), and the optical-catalog association
processing no longer uses the position uncertainties at all (only
coarse-window matching is done). The final tests of position
discrepancies with respect to the astrometric reference catalogs
will provide the overall position uncertainties, and results to
date indicate that the Level 1 requirements are already being
safely met, which also suggests that the refraction and
distortion models will contribute more to enhancing the products
than to permitting the requirements to be met.

     [Note added in proof: subsequent discussion between H.
     McCallon and R. Cutri resulted in a reorder. Top priority
     will be refracttion and distortion, since these are the most
     significant effects. Second is the band-band linear fit,
     and, time allowing, the uncertainty modelling.]

    The other main area involves fully populating the PSF models.
B. Wheaton reported that the goals for Dec. 1 had been met, but
G. Kopan reported that a considerable amount of tuning with the
full model needs to be done before the dependence of the
detection thresholds on PSF FWHM will be on solid ground. The
imminent arrival of data from the Southern observatory may
introduce some surprises, but at the least, a lot of tuning for
its peculiarities will be needed.
    A conflict was clearly recognized between the need for a
stable version of 2MAPPS in February and the lessons of
experience dictating that redeliveries will not cease to be
necessary at that time. It was decided that the best way to
accomodate both considerations was to have a "development
pipeline" running a controlled RTB concurrently with the
production pipeline. The former would have to be isolated from
the latter to a sufficient degree to ensure that no accidental
cross-talk would be likely, which would be easiest to guarantee
if a separate machine were dedicated to this task. The RTB would
have to be large enough to provide statistical likelihood that
all known types of problem scans were included, but small enough
to permit the entire RTB to run in a few days or about one week
at most. As new problems are encountered in new data, a
representative scan would be added to the RTB. This would
eliminate any need to test new code in the production pipeline,
but would permit full testing of all software fixes and
enhancements. If fixes to the production software were to become
unavoidable, thorough testing beforehand would be possible.
Furthermore, considerable lead time for developing and testing
Version 3.0 would be provided.
    Two concerns were recognized: (a.) a separate machine would
be highly desirable, but resources would have to be allocated;
(b.) many of the current software developers are not scheduled to
be working on 2MASS even through the end of 1998, much less when
version 3.0 would be delivered. The latter concern is a problem
in any event; the only clear path to alleviating it is to
minimize the shortcomings of version 2.0, which every effort will
be made to do.

6.) Data Base Status

    J. Mazzarella reported that the query-by-table capability of
the data base has been implemented and that the design for
accessing overlapping images from different surveys has been
done.