From jwf@ipac.caltech.eduWed Sep 10 14:42:27 1997
Date: Fri, 30 May 1997 17:57:02 -0700 (PDT)
Subject: WG Mtg #124 Minutes

           IPAC 2MASS Working Group Meeting #124 Minutes

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


1.) Band-to-Band Position Anomalies
2.) Unset Par File Parameter
3.) New Dithering Pattern
4.) 2MAPPS Part of Next Week's Review


1.) Band-to-Band Position Anomalies

    H. McCallon presented evidence that the H array moves slightly relative to
the J and K arrays during scans and slews. It is not yet determined that the H
array alone is involved, but further tests should reveal whether it is. The
latest version of PFPREP computes a 5-parameter band-to-band alignment model
for all band pairs on a single-frame basis. For sufficiently dense scans, the
parameters are relatively well determined at the frame level. The five
parameters for any given band pair are: offsets in X and Y, scale factors in X
and Y, and rotation angle.
    Data from 970514 were used in the study presented. No anomalies have been
observed in the scale factors or rotation angle, but the X and Y offsets show
statistically significant variations over the night and during most 6-degree
scans. Scan 036 in particular showed an unusually large change of almost a
full arcsecond in J-H offsets between the beginning and end of the scan.
    Whereas some band-to-band alignment change due to flexure has been
expected, it was not supposed to be large enough to make the 5-parameter model
inapplicable to a complete scan, and so the pointing reconstruction does not
currently include (e.g.) a linear fit to these parameters over the duration of
a scan (in fact, sparse scans could not supply enough data points for reliable
linear fits). The changes in offset over scan 036 are clearly nonlinear; a
linear correction would remove most of the error, however (a visual estimate
based on the plots would be about 80%).
    As it stands, scan 036 would have J-H and H-K alignment errors of about
plus or minus 0.5 arcsec over the scan. Some of this would be reduced by band
merging in the case of multi-band sources successfully merged (0.5 arcsecond
error would not make a serious impact on band-merging accuracy except in
confused cases). It is not known whether scan 036 is as bad as the phenomenon
ever gets, however. Plots of excursions over the entire night show what
appears to be a random walk covering several arcseconds, and the possibility
that such a large excursion could occur during a scan cannot be ruled out.
    The large effects are seen in J-H and H-K, but J-K also shows a
statistically significant effect that is much smaller and appears correlated
with the others over the night. Further study of the phenomenon and possible
software enhancements will be pursued.

NOTE ADDED IN PROOF: H. McCallon later established high correlations between
all offset variations (i.e., all band pairs and both axes) and identified the
H-pickoff dichroic as a possible source for the effect, since it would be
capable of producing the large J-H and H-K variations with the small
correlated J-K variations. Information has been set up on the 2MASS web site
discussing this study.

2.) Unset Par File Parameter

    J. Fowler reported that during investigation of the anomalies discussed in
section (1.) above, he and H. McCallon discovered that the array temperatures
were always 0 deg C as reported in the "par" file sent from the observatory.
Since the origin of the anomalies was a mystery at the time, correlation with
any and all available parameters was being sought (e.g., zenith angle, time,
temperature, etc.). It was obvious that the array temperature parameters were
not being set by the output-generating software at the observatory.
    This prompted a check of other parameters in the par files that might have
apparent values that were not actually resulting from measurements, and this
yielded the following "group 2" (supposed to be measured at scan start time
and again at scan end time) parameters: wind speed, wind direction, and
seeing. The barometric pressure, which is needed for the refraction model in
POSMAN, was obviously being set on a one-value-per-night basis, which implies
dubious usefulness, but at least its value changed from night to night.
    The working group voted to recommend that the par files not include any
parameters that are not actually measured (since they appear to be measured
quantities in the downstream data flow, creating confusion and possible false
conclusions), that the array temperature and barometric pressure parameters be
actually measured and recorded in the par files at the start and end of each
scan, and that if corresponding changes to the observatory software are not
feasible, then the RDFRAME module will be modified to prevent erroneous
parameter specifications from being included in the downstream data flow or at
the very least attach FITS header comments that identify such parameters as
received from the observatory but not to be used because of not having been
actually measured.

3.) New Dithering Pattern

    T. Chester reported that a new dithering pattern was supposed to be in
effect as of 970514, and he requested that analysis be done to verify that the
new pattern is being executed and that it yields the desired improvements. The
modifications for reducing the electrical noise have also been made, and these
have already been found to be successful.

4.) 2MAPPS Part of Next Week's Review

    R. Cutri reported that although 2MAPPS is not the main subject to be dealt
with in the upcoming review, its products will nevertheless play an important
role in judging the operational readiness of the observatory. The status of
2MAPPS itself will be discussed in the morning of the second day. Final lien
schedules should be sent to J. Fowler for inclusion in the document to be
posted on the web for the Science Team to read before coming to the review.