- Telescope Status --
R. Cutri and C. Beichman reported that agreements between the project, the
telescope manufacturers, and the bonding company have led to the decision that
the telescope assembly can be kept on the same schedule and budget as previ-
ously planned. The use of the new telescope for the northern-hemisphere sur-
vey is now as likely as the use of the Kitt Peak 50-inch telescope. [Note
Added in Proof: M. Skrutskie stated later that '...the new telescope on Mt.
Hopkins has revived from being a longshot to being a 50/50 competitor with the
50".'] The 2MAPPS schedule remains compatible with reception of data from the
new telescope in September 1996.
- PIXCAL Tests and Timing --
J. Fowler and G. Kopan reported that the 2MAPPS processor consisting of
the integrated subsystems PIXCAL, RDFRAME, DFLAT, and FREXAS has successfully
executed in a test environment. The interfaces between all four subsystems
(and that between them and the PCP subsystem, which will run this processor)
appear to be functioning correctly, and the test environment and input data
formats duplicate those of 2MAPPS. Output data appear to be correct. Future
input from the observatory should be able to be processed by this program.
Some anomalies were observed in the timing measurements, however. The
variation of CPU and I/O time with processing load was inconsistent, with
lighter processing loads sometimes taking more time, and with identical execu-
tions taking different times. Apparently the way in which CPU and I/O times
are measured by the operating system is not as straightforward as expected,
and variations of about 30% are encountered, probably depending on other
activities on the machine. Even so, the execution times are close enough to
the FDD allocations to eliminate concern. Comparable processing in the
proto-pipeline takes about ten times longer; the main speedup is believed
due to the greater efficiency in FREXAS relative to DAOPHOT.
- Distortion Modeling --
J. Fowler, G. Kopan, and B. Light reported that the baseline 2MAPPS
approach to handling optical distortion is currently to retain documentation
of a plan for implementing it if needed, to derive certain statistical para-
meters in POSFRM that will reveal whether optical distortion effects are
present to a significant extent, but otherwise to assume that distortion
corrections are not needed unless evidence to the contrary arises. The PROPHOT
and PICMAN SDS's will carry liens flagging the fact that distortion correc-
tions are not being made.
The decision to adopt this approach is based on:
- (a.) measurements of
optical distortion in the latest protocamera data have failed to reveal
any effects larger than about 0.1 pixel (it must be noted that these
measurements are spot checks, not exhaustive searches);
- (b.) the
resources required to implement a distortion model (i.e., an
algorithmic design, a calibration method and implementation, and code
to use the model in PROPHOT and PICMAN) are too large to be written off
as worthwhile without a clearcut need.
POSFRM will accumulate statistics on the means and variances of the
position separations between point source apparitions in each panel and
refined position estimates (i.e., based on all apparitions). This will
be done in a grid over focal plane position. Significant distortion
should produce significant mean separations, and the dependence on
focal plane position should allow a distortion correction model to be
derived if needed.
This model would be implemented in PROPHOT as a subroutine that
converts apparent pixel coordinates to non-distorted coordinates, so
that PROPHOT can stack apparitions optimally. It would not be necessary
for PROPHOT to depend in any way on the internal details of the
correction subroutine.
PICMAN would have to implement the correction in a more complicated
way. Whereas PROPHOT spends relatively little time doing
position dependent computations, PICMAN would need the distortion
corrections inside its most deeply nested loops. This would probably
require model dependent coding to implement the corrections in the most
computationally efficient way.
No claim is made that the cost of implementing a distortion correction
is prohibitive, only that it is too high to undertake without a clear
need. Furthermore, the discussion above should not be considered a
final description of the issues involved or the methods to be
developed. It should also be noted that this discussion does not
address differential refraction, effective scale factor variation
during scans, or other questions to be investigated by H. McCallon.
- Karloff Disks Cross-Mounted to Lugosi --
R. Cutri
reported that D. Wittman is in the process of cross-mounting karloff's
disks to lugosi. Since karloff's CPUs are being used heavily, use of
lugosi's CPUs for processing data on karloff's disks is currently
recommended.
- Dependence of (R2-R1)-R1 Magnitudes on Scan Direction --
T. Evans reported that the differences in aperture magnitudes
for Read1 and Read2-Read1 detections of the same sources show a
dependence on scan direction. She displayed graphs of this phenomenon
and supplied the following summary which constitutes the remainder of
this section.
The R1 and R2-R1 (.ph) sources in each of the HHE survey scans
taken on photometric nights have been matched to each other and
compared. Only the sources with both R1 and aperture R2-R1
magnitudes between certain cutoffs were used:
J: 9.5 <= mj <= 11.0
H: 9.0 <= mh <= 10.5
K: 8.5 <= mk <= 10.0
Note that the R1 magnitudes were calibrated to the aperture R2-R1
magnitudes by only one R1 offset per band over the whole May '95
observing run.
The plots of R1-(R2-R1) magnitudes (DM) vs. X (and, shown in
previous meetings, Y) positions show no significant trends with
position, but the average DM for ascending scans is slightly more
positive [m(R1)>m(R2)] than for descending scans. Here are the
approximate values (meaning the fitted line intercept at
X=0):
J: asc: .044 +/- .003 desc: .019 +/- .004
H: asc: .014 +/- .001 desc:-.013 +/- .002
K: asc: .017 +/- .002 desc:-.009 +/- .003
The difference offset of from 0 is usually more statistically
significant for the ascending scans than the descending scans.
The difference between the two averages for each band is about
.026 mag. It is assumed that this difference is caused by
differences in the chopping secondary action or settling time for
ascending and descending scans, or some other telescope dependent
effect. Some careful testing of the survey telescopes is needed;
but note that the survey R1 photometry will be calibrated to the
aperture R2-R1 magnitudes on a scan-by-scan basis. If the problem
is only with the R1 data, this calibration scheme should
theoretically remove the effects.