Subject: YAMS: more on small separations
From: Dave Monet
Submitted: Sun, 9 Mar 2003 09:17:53 -0700
Message number: 92
(previous: 91,
next: 93
up: Index)
For reasons that are not obvious, I have completed the astrometric
solutions for all 72 Megacam amplifiers on the 3 60-sec images discussed
in the previous note (which was based on data from 18 amplifiers).
No big changes, but the larger number of pairs (about 2.5 million) gives
better statistics. My desire was to look at the small separation
results to see if there was some sort of astrometric "floor". To do this,
I changed the bin size from 30 arcsec to 5 arcsec, and I have given the
full table as an appendix to this message. Conclusions that make some
sense this morning are the following.
1) The astrometric "floor" appears to be a median error of about
3 mas. Subtracting this in quadrature from the value at 200 arcsec
(about 7 mas) gets these results pretty close to the factor of
7.8 predicted by SQRT(60/1) from the exposure times.
2) The upturn in the first bin (0-5 arcsec separations) is probably
due to image blending. Even in good seeing, images this close
will need a different astrometric algorithm than the isolated,
single star routine that I used.
3) I think that the turn-over seen at large separations (> 200 arcsec
which is about 1000 pixels) comes from the least squares fit needed
to remove the effects of the optics. Least squares always
emphasizes the end points, and the lack of high-frequency terms
in a linear or quadratic fit ought to minimize the errors at
something on the scale of the dimensions of the CCD. I think that
the "correct" answer is the one seen in the 1-sec undithered
data where the astrometric error reaches an asymptotic limit
and then stays there.
4) I am beginning to think that we don't need to re-do the full-field
astrometric solution in favor of many small-field solutions, at
least if Mauna Kea is the site. The Flagstaff data taken quite a
while ago showed a much larger difference between the small-field
and large-field errors, but now I am in the process of convincing
myself that this comes from the quality (or lack thereof) of the site.
If all we gain is a factor of 2 between the large-field solution
and a plausible small-field size of 30 arcsec, and if we really get
something like 7 mas per visit, then we really ought to be able
to get something like 1 mas per year (==SQRT(1 visit/week)).
Maybe I can get the Megacam engineering folks to take 3 undithered
frames of the same field a few more times to see if SQRT(Nvisit) has
anything to do with the astrometric results. Again, many thanks to
those who provided the data: this is just what I think we need to plan
the astrometric side of the Pipeline.
-Dave
=====================
Middle of N 25% 50% 75%
bin (arcsec) <--------- mas ------->
-------------------------------------------------
2.50 1172 1.9 3.4 5.6
7.50 5719 1.8 3.2 5.0
12.50 9222 2.0 3.4 5.3
17.50 12392 2.2 3.6 5.7
22.50 15510 2.3 3.9 5.9
27.50 18562 2.5 4.1 6.4
32.50 20894 2.6 4.3 6.6
37.50 23546 2.7 4.5 6.9
42.50 26033 2.8 4.7 7.2
47.50 28076 3.0 4.9 7.5
52.50 29842 3.1 5.0 7.6
57.50 31884 3.1 5.2 7.9
62.50 33188 3.2 5.3 8.0
67.50 34796 3.3 5.4 8.2
72.50 36024 3.4 5.5 8.4
77.50 37470 3.5 5.6 8.5
82.50 38175 3.5 5.7 8.5
87.50 38737 3.5 5.7 8.6
92.50 39446 3.6 5.8 8.8
97.50 39550 3.6 5.9 9.0
102.50 39861 3.7 6.0 9.0
107.50 39887 3.7 6.0 9.2
112.50 40144 3.8 6.1 9.3
117.50 39746 3.8 6.2 9.4
122.50 39326 3.8 6.2 9.5
127.50 39076 3.8 6.3 9.6
132.50 38259 3.8 6.4 9.7
137.50 37195 3.9 6.4 9.7
142.50 36919 3.9 6.5 9.8
147.50 35844 4.0 6.5 9.9
152.50 34813 4.0 6.7 10.0
157.50 33400 4.1 6.7 10.2
162.50 32092 4.2 6.8 10.2
167.50 31243 4.2 6.9 10.2
172.50 30413 4.3 6.9 10.3
177.50 29620 4.3 7.0 10.3
182.50 28670 4.4 7.0 10.4
187.50 28554 4.4 7.1 10.5
192.50 27827 4.5 7.1 10.4
197.50 27119 4.4 7.1 10.4
202.50 26720 4.5 7.1 10.5
207.50 26035 4.5 7.2 10.5
212.50 25520 4.5 7.2 10.6
217.50 24897 4.5 7.2 10.5
222.50 24635 4.6 7.2 10.5
227.50 24237 4.6 7.3 10.6
232.50 24077 4.5 7.2 10.5
237.50 23601 4.6 7.2 10.5
242.50 23181 4.5 7.2 10.5
247.50 23131 4.6 7.1 10.4
252.50 22545 4.6 7.2 10.4
257.50 22048 4.5 7.2 10.4
262.50 21728 4.5 7.1 10.3
267.50 21661 4.6 7.2 10.4
272.50 21160 4.5 7.1 10.2
277.50 21090 4.5 7.0 10.2
282.50 20630 4.5 7.0 10.1
287.50 20536 4.5 7.0 10.1
292.50 20094 4.4 6.9 10.0
297.50 19979 4.4 6.9 9.9
302.50 19946 4.4 6.9 9.9
307.50 19069 4.3 6.8 9.8
312.50 19115 4.3 6.8 9.8
317.50 18871 4.3 6.7 9.8
322.50 18524 4.3 6.6 9.6
327.50 18042 4.3 6.7 9.7
332.50 17967 4.2 6.6 9.6
337.50 17780 4.2 6.6 9.5
342.50 17538 4.2 6.6 9.6
347.50 17407 4.1 6.5 9.5
352.50 17228 4.2 6.6 9.4
357.50 17003 4.2 6.6 9.5
362.50 16847 4.2 6.5 9.4
367.50 16361 4.1 6.5 9.4
372.50 16333 4.1 6.5 9.2
377.50 15880 4.1 6.4 9.3
382.50 15768 4.1 6.5 9.3
387.50 15629 4.1 6.4 9.2
392.50 15475 4.1 6.4 9.2
397.50 15255 4.1 6.5 9.2
402.50 15061 4.1 6.4 9.1
407.50 14981 4.1 6.3 9.0
412.50 14579 4.0 6.4 9.1
417.50 14441 4.1 6.3 9.0
422.50 14175 4.1 6.3 9.1
427.50 13891 4.0 6.3 9.1
432.50 13838 4.0 6.3 9.1
437.50 13479 4.1 6.3 9.0
442.50 13343 4.0 6.3 9.0
447.50 12916 4.0 6.2 9.0
452.50 13033 4.1 6.3 9.0
457.50 12861 4.0 6.2 8.9
462.50 12687 4.0 6.3 8.9
467.50 12293 3.9 6.1 8.8
472.50 12106 4.0 6.2 8.8
477.50 11949 3.9 6.1 8.8
482.50 11701 4.0 6.2 8.9
487.50 11604 3.9 6.1 8.8
492.50 11322 3.9 6.1 8.8
497.50 11103 3.9 6.1 8.8
LSST LSST LSST LSST LSST Mailing List Server LSST LSST LSST LSST LSST LSST
LSST
LSST This is message 92 in the lsst-general archive, URL
LSST http://www.astro.princeton.edu/~dss/LSST/lsst-general/msg.92.html
LSST http://www.astro.princeton.edu/cgi-bin/LSSTmailinglists.pl/show_subscription?list=lsst-general
LSST The index is at http://www.astro.princeton.edu/~dss/LSST/lsst-general/INDEX.html
LSST To join/leave the list, send mail to lsst-request@astro.princeton.edu
LSST To post a message, mail it to lsst-general@astro.princeton.edu
LSST
LSST LSST LSST LSST LSST LSST LSST LSST LSST LSST LSST LSST LSST LSST LSST LSST