From: Walt Morgan
Date: Sun May 22, 2005  1:50 pm
Subject: April 15 Ups Gem graze analysis 	


The analysis done by Michael Richmond for our April 15-16 graze by
Ups Geminorum (magnitude 4.1) is extremely detailed and precise, and
I would like to think that such analysis could be routine in the
future. Since so much time was put into it in this case, it seems
appropriate for me, as expedition leader, to add information not
readily available to Michael.

Frank Anet had suggested that the plots should use linear amplitude
displays instead of the logarithmic displays used in earlier plots,
and I believe that is a significant improvement. Michael's revised
report was announced in message #11869 (May 18, 2005), and may be
found at:

http://spiff.rit.edu/richmond/occult/ups_gem/ups_gem.html

There were a total of 15 stations in Fremont, CA for that graze, with
13 of them obtaining useful data, 7 of them using video. The
detailed analysis by Michael was for just three of the video stations
(the ones with the most events, and all that were readily available
to me), as follows:
J – Walt Morgan 499m south
K – Derek Breit 564 m south
P – Ed Morana 954 m south

Since those observers were not all on a common line perpendicular to
the limit, central graze time differed between them, with the higher-
lettered stations falling later. Since K had the most data, and used
McAfee VTI, it will be convenient to correct the other stations to
that:
Add 0.13 sec to the times at J to adjust to K
Subtract 0.18 sec from the times at P to adjust to K

The times in Michael's plots do not take these differences into
account, but I will include them in what follows. I will, however,
round all times to 0.1 seconds. Note that in all cases times are
given in seconds following 4 hours 38 minutes UT. This was a
northern limit graze, so it is to be expected that peaks will be
shortest at station J, and valleys will be shortest at station P.

I have gone through all of the plots for stations J, K and P, noting
the 1/3 amplitude points for each event. My understanding is that
for a star of this type, of intermediate diameter, that the correct
transition time is at about 30% of the un-occulted brightness.
However, the un-occulted images were saturated in all cases, so the
transition time should probably be at a higher level on the plots,
perhaps even one-half, or higher. My choice of 1/3 may therefore not
be quite correct.

Prior to sending the data to Michael, the observers at each of the
three stations had done their own determination of times by viewing
frame by frame. At J the record was moved from VHS tape to the
computer by using Dazzle's Hollywood Bridge, which I have shown to
have a more or less random audio lag of from 1 to 9 frames. A lag of
five frames is the average that I use: the time for a transition
frame is established by interpolating from WWV audio references, then
five frames are subtracted, the audio lag being treated like a
Personal Equation. Note that the uncertainty in this is more than
0.1 seconds.

The station J data processed by Michael were moved from VHS tape to a
digital record by using Ed Morana's Adaptec A/D converter, and there
is reason to believe that it does a better job than the Dazzle gear.
In comparing station J times from the plots with the times obtained
using the Dazzle equipment, there is a consistent difference of 0.3
(+/- 0.1) seconds. This inconsistency is not understood; all station
J times that follow will use times from the plots—because I really
don't trust the Dazzle A/D conversion approach.

The identification of events in the following use the names assigned
by Michael.

EVENTS D1 & R3
The leading edge of the moon was quite steep, so there were no
leading-edge multiple events, and the first disappearance for all
observers (spanning nearly 2000 metres across the limit line)
occurred in an approximately linear sequence in 31.3 seconds (after
adjusting to a common central graze time). Because of the varying
depths/heights of the intermediate valleys/peaks, the common features
for the three stations J, K and P were only the first D (D1) and the
last R (R3). (R3 was the last event at J and K, but P saw later
events.) Their times were:
Sta D1 R3
J – 28.0 - 139.1
K – 27.8 - 141.1
P – 26.2 - 144.5

These times are not only self-consistent, they are consistent with
other observers. With this frame of reference, we can be quite
confident about intermediate times.

EVENT F
This "flash" came to full brightness for J, but only to about 3%
brightness for K.
For J the R was at 57.6 and the D at 58.2.
For K the center of the flash was at 58.0.
The K flash is unique in my experience. I first saw it (via a VHS
copy) only because I was checking for something that might match the
J record near 58 seconds. It was not visible on the Dazzle-produced
digital record, and it could not always be seen in playback of the
VHS tape. The record produced by Michael certainly argues for its
presence, and the coincidence of times between J and K is a
confirmation of that.
A visual observer at 530 m south did not notice this flash, but it
was surely less than one-half second in duration for him, and
probably reduced in brightness, too. This of course is a powerful
argument for the video approach.


EVENTS R1 & D2
For J the R was at 66.6 and the D at 91.1. (net 24.5 sec)
For K the R was at 67.5 and the D at 90.7. (net 23.2 sec)
For P there was a flash centered at 69.4, peaking at 40% of full
brightness.
The implication is that the valley barely reached to P.
Problem: Observer P, based on his frame-by-frame analysis, reported
a D at 72.6 and an R at 74.0. The plot for P could be reported as an
R at 69.0 and a D at 69.6 instead of a flash, but there is still an
inconsistency.

EVENT R2
J – 95.5
K – 96.1

TIME 100
Observer P, based on his frame-by-frame analysis, reported a flash
at 100.3.

EVENT P
This "blink" was observed only at K, implying a peak that reached
to K but not to J.
K – Drop to 40% at 122.9;
K – Return to full at 123.3.
The fact that the brightness did not approach zero could be
attributed to a partial occultation. Most of the observers noted
many slow transitions, supporting the concept that the star diameter
was significant.

TIME 127
Station P showed a 5% increase in brightness between 126.9 and
127.9. It is difficult to associate this with a valley because the
times do not correlate with other stations and the level is such a
small increase above the noise. The observer, based on his frame-by-
frame analysis, reported a flash at 128.0.

EVENT D3
J – 135.0
K – 134.3

EVENT R3 was given above with D1, and shows good consistency between
J, K and P. Just prior to R3, however, there are indications of low
level flashes for J and K.
J – 137.0 to 3%
K – 139.8 to 3%
These both look real in Michael's plots, but they are very low level
and the times do not correlate well with each other. My belief is
that these flashes are more likely demonstrating that these
inexpensive video systems are not research grade instruments. They
are giving us wonderfully precise data about the moon, but they may
well also be subject to stray instrumental noise pulses. The same
statement might apply to the TIME 127 event for station P.


SUMMARY
There has been extensive discussion about the possibility that some
of the video records for this graze reveal a faint secondary. That
may be a correct interpretation, but my discussion above ignores that
possibility, dealing only with the more fundamental questions of
outlining peaks and valleys by assuming there is only a single star.
It is believed that given that premise, stations J, K and P are
consistent in positively identifying three peaks and two valleys to
an accuracy of one-quarter second or better.

The ability to do photometry on video records spanning a few
minutes in time is a very powerful tool, one it would be desirable to
have available to more persons.

Walt Morgan