Author: Michael Richmond Date: 970330 Revision: #1 970330 Key Words: CCD, astrometry, observation, techniques
Tom Droege acquired a set of images with a triplet at his home near Chicago (longitude approx 88:20, latitude approx 41:50) during the month of February, 1997. He posted a message describing his dataset to the TASS mailing list, and urged members to try their hand at reducing the images. The images are largely I-band, with a few V-band; unfortunately, there are no fields covered in both passbands. You can find more information about this dataset at ../tom_feb97.html
Three people took up the challenge, reduced the images, detected stars in each, and measured the position and brightness of each. They are
This document describes my attempts to compare the positions and brightnesses measured by each person for a subset of Tom's images. The executive summary is encouraging: the three methods yield consistent relative magnitudes (with one caveat), but there are certainly small systematic differences in astrometry.
Note to Mike and Glenn: if I have any details wrong, please let me know, and I'll make the necessary corrections.
Tom distributed a set of 20 images, consisting of
Table 1. Image names, positions, times
# H = high galactic latitude # L = low galactic latitude # # the "RA" and "Dec" values are equinox 2000 coordinates # at the rough center of each image # pixel # name date time camera RA Dec sky skysig FWHM g0483967.fts 03-02-1997 11:13:48 0 I H 202.34 -1.90 -27918 55 3.4 g1493946.fts 13-02-1997 10:42:37 1 V H 219.13 -1.04 -26547 46 2.6 g2493623.fts 13-02-1997 02:57:18 2 I L 101.40 -0.76 -26748 82 4.4
Each of us used different software to perform the tasks necessary to remove the instrumental signature from the raw images:
In fact, we probably made rather different assumptions about the proper way to generate dark and flatfield vectors. It is very likely, therefore, that our corrected images were not quite identical. Even if we had used identical software to find and measure stars, the small differences in our corrected images would have produced some variations in our results.
The three of us each used different software to find and measure the brightness of stars in each image. Here's a quick summary of each approach:
Since my comparisons were made on calibrated quantities (RA and Dec for position, 'standard' I and V magnitudes), this was a crucial step in each reduction. You'll see below that I had to "cheat" a bit.
My calibrations went in two steps: first, I compared the positions of a subset of bright stars in my raw photometry output against a subset of bright stars from the Hubble Guide Star Catalog. I used triangle-matching code to find the stars in common in each set of stars, typically settling on 6-15 stars across each frame. Based on those matching stars, I then calculated a linear transformation of the form
x = A + B*row + C*col
y = D + E*row + F*col
between the measured (row, col) position of a star on the frame,
and (x, y) the spherical coordinates, projected onto a plane at
the center of the frame.
I applied this transformation to each detected star.
Next, to calibrate the magnitudes, I compared the positions of
stars against a set of Landolt standard stars in a band around
the equator.
Because I didn't use a small threshold when finding stars, I found
only a very few stars in both my star lists and in the Landolt catalog.
I averaged the difference between Landolt and instrumental magnitude
for each field, then subtracted the average from all instrumental
magnitudes.
Mike Gutzwiller provided star lists which tabulated the (row, col) position on the raw image as well as the (RA, Dec) position on the sky. I speculate that he used the SAO catalog to determine positions. The magnitudes in his star lists appeared close to those on the standard Johnson-Cousins scale, but I don't know how to did the photometric calibration.
Glenn Gombert provided two sets of stars lists: one had raw quantities, (row, col) plus instrumental magnitude, and the other calibrated quantities, (RA, Dec) and 'standard' magnitude. I confess that I was unable to find the correspondence between the two lists for the first field I examined (g0483967). After a bit of work, I was able to determine the correspondence between his raw quantities and those of Richmond and Gutzwiller:
Richmond "row" = Gutzwiller "Y" = Gombert "Y"-1
Richmond "col" = Gutzwiller "X" = Gombert "X"+15
I ran my own astrometric procedure on his raw star lists to
calculate (RA, Dec), and then ran my photometric calibration procedure
to yield V-band and I-band magnitudes on the Landolt system.
Actually, I cheated a bit in this step for one image:
I was unable to make a good match between his raw stars
and Landolt stars in the low-latitude I-band field g2493623,
so I calibrated his raw magnitudes against my calibrated magnitudes
in that field. I used 201 stars, all brighter than I=10.0,
to find an offset of 0.42 +/- 0.05 mag, then applied that to
his raw starlist.
Okay, enough of the preliminaries, let's get to the good stuff.
I present here a comparison of the positions and magnitudes of stars in each of three fields. Since I suspect that there could be systematic differences in our aperture photometry as a function of magnitudes, I decided to tabulate magnitude differences in three groups:
Here's a table of the photometric differences:
Table 2: Photometry (Gutzwiller - Richmond)
g0483967 g1493946 g2493623
I-band, sparse V-band, sparse I-band, crowded
N mean stdev N mean stdev N mean stdev
------------------------------------------------------------------------------
mag < 9 35 -0.04 0.01 10 -0.36 0.02 143 -0.47 0.03
9 < mag < 11 132 -0.03 0.03 88 -0.37 0.02 676 -0.46 0.11
mag > 11 255 -0.02 0.12 297 -0.36 0.10 305 -0.49 0.24
Note that there are no large trends of offset with magnitude, which is good. The important column in this table is the one labelled "stdev" -- it shows the scatter between measurements of stars of similar magnitude. At high galactic latitudes, where the stars are few and far between, the scatter is about three percent down to eleventh magnitude. I think that's very promising.
Here are plots showing the above results in graphical form:
I also made plots showing the difference in position of stars matched in the two datasets. I included only stars whose total difference in position was less than or equal to 10 arcseconds. Below I show histograms of the differences, for each image:
Finally, plots which reveal any trend in the differences as a function of RA or Dec. If the points in the diagrams below do not lie along a perfectly horizontal line, then there is some systematic difference in the astrometry.
You can see that, indeed, there are some trends in the differences of position. I do not understand the source of the trends yet.
Here's a table of the photometric differences. Remember that I calibrated Gombert's raw measurements of g2483923 against my own calibrated measurements, forcing the offset to be 0.00 (at least for the bright stars).
Table 3: Photometry (Gombert - Richmond)
g0483967 g1493946 g2493623
I-band, sparse V-band, sparse I-band, crowded
N mean stdev N mean stdev N mean stdev
------------------------------------------------------------------------------
mag < 9 40 -0.01 0.02 10 -0.01 0.06 81 0.00 0.04
9 < mag < 11 125 -0.02 0.03 88 -0.03 0.04 302 -0.02 0.08
mag > 11 205 -0.05 0.09 318 -0.04 0.21 161 -0.20 0.18
Note that there is only one entry which shows a large trend of offset with magnitude: the faint stars in the crowded I-band field. You can see the obvious curvature in the third plot below.
Here are plots showing the above results in graphical form:
I also made plots showing the difference in position of stars matched in the two datasets. I included only stars whose total difference in position was less than or equal to 10 arcseconds. Below I show histograms of the differences, for each image:
Finally, plots which reveal any trend in the differences as a function of RA or Dec. If the points in the diagrams below do not lie along a perfectly horizontal line, then there is some systematic difference in the astrometry.
You can see that, again, there are some trends in the differences of position.
Although I haven't had time to digest this information very thoroughly, I can make some quick statements:
Overall, I'm pleased by the consistency between the three independent methods of measurement ... and I think we can improve a bit.
You can find pointers to the raw images upon which this work was based, and the calibrated star lists created by all three users, in Data Archive for Tom Droege's February 1997 images.
Back to Index of Technical Notes
Back to TASS home page