I have put all the existing Mark III data into the database on the computer stupendous.rit.edu. I am currently examining various properties of the data, to find out if we can use it as-is, or if we need to make additional corrections to it. I welcome your input.
The "current data" consist of measurements from three sites:
All data has been entered into the database exactly as it was sent to me from each site. I have not modified the reported magnitudes in any way; one purpose of this note is to find out if any such modification is necessary.
Let me make a number of different comparisons between the TASS data and other sources of information.
TASS versus Landolt on field-by-field basis
The continuous images from the TASS Mark III cameras are chopped into "fields" for convenient processing. Each "field" is about four degrees long in the Right Ascension direction. In this section, I work one field at a time: for each field, I find stars detected by the TASS camera which match stars in Landolt's catalogs of standard stars. I then calculate the mean difference
Landolt mag - TASS mag
from all matching pairs in the field, and also calculate
the standard deviation from the mean.
A "match" occurs when a TASS detection falls within 5 arcseconds of a Landolt star. I make no checks for similar magnitudes.
First, TASS data from site "B" (Glenn Gombert's triplet).
Next, TASS data from site "D" (Mike Gutzwiller's triplet):
Next, TASS data from site "H" (Tom Droege's triplet):
What does this show?
TASS versus LONEOS on field-by-field basis
Brian Skiff has collected a very large set of photometry published in a wide selection of sources. His collection is not homogeneous, but does cover a large fraction of the entire sky. He states that the collection contains "low-precision" magnitudes, which might be good to 0.10 mag or so. The main purpose of this collection is to support the LONEOS project. I will refer to Brian's collection as the LONEOS catalog, for short.
In this section, I work one field at a time: for each field, I find stars detected by the TASS camera which match stars in the LONEOS catalog. I then calculate the mean difference
LONEOS mag - TASS mag
A "match" occurs when a TASS detection falls within 5 arcseconds of a LONEOS star. I make no checks for similar magnitudes.
First, TASS data from site "B" (Glenn Gombert's triplet).
Next, TASS data from site "D" (Mike Gutzwiller's triplet):
Next, TASS data from site "H" (Tom Droege's triplet):
What does this show?
Are field-to-field variations correlated?
One danger of combining measurements made by different instruments and on different nights is failing to set the zero-point of each set of measurements properly. Suppose that one has measurements of 1000 stars on night A, and the same 1000 stars on night B. Even if the relative photometry between stars is the same on each night, the zero-point might be different. One might discover such an error by noting that
One way to check for zero-point errors is to pick a subset of stars which are close to each other on the sky, and plot their magnitudes vs. time (or night, or whatever). If there are zero-point errors, the plot will show the stars varying "in step": all fainter on the second night, then all brighter on the third night, and so forth.
I picked three subsets of stars to check for this problem, and below show the graphs of magnitude vs. field number (not time) for each set.
Set 1 is a set of stars with V-band magnitudes between 9.0 and 9.5, in a small patch of the sky near RA = 10 degrees and Dec = 0 degrees. Note that the graph uses "millimags" on the vertical axis.
Set 2 is another group of stars in the same region, but somewhat fainter: V-band magnitudes between 9.5 and 10.0. Note that the graph uses "millimags" on the vertical axis.
Set 3 is a group of stars of roughly the same spectral type: their types are F2, F2, F8, G1, G1 and G8. These stars are relatively unlikely to vary with large amplitudes. All the stars have V-band magnitudes between 9.1 and 10.5. Below are the graphs of their V-band magnitudes, and then their I-band magnitudes, versus "field" number.
What does this show?
TASS camera values vs. other TASS camera values, on star-by-star basis
One way to characterize the uniformity of data from different devices is to compare the measurements each makes of identical stars. In this section, I pick stars which were observed by two different TASS cameras in the same passband. For each camera, I calculate the mean of all magnitude measurements; I then compare the mean value from one camera against the mean value from the other camera.
All stars considered in this section are good; they must have been detected on at least 10 occasions (not by each camera, but just overall).
First, comparison of the V-band cameras: there are only 3 cameras, so I can compare each pair of cameras against each other on a single plot with 3 panels. On both axes, magnitudes are expressed in millimags.
Now, the I-band cameras: there are 5 cameras, which means many pairs. Below are plots in which the same camera (B0) is used as one member of the pair.
One can calculate numerical values for these differences, as well as plotting them. I discovered that there were a number of outliers in the distribution of differences; they might have been due to intrinsic stellar variability, or "junk" such as airplanes and satellites. I therefore ignored all differences such that
abs(difference between cameras) > 1.0 mag
I then split the stars into two groups:
Note that values in the tables below, and on the graphs above, are in millimagnitudes (0.001 mag).
V-band differences
cameras N mean offset stdev
-------------------------------------------------------------------------
B1 - D1 bright 6,584 1 mmag 65 mmag
faint 46,512 124 253
B1 - H0 bright 3,156 7 74
faint 22,337 68 247
D1 - H0 bright 7,421 6 58
faint 52,528 -47 211
I-band differences
cameras N mean offset stdev
-------------------------------------------------------------------------
B0 - B2 bright 41,445 36 mmag 138 mmag
faint 127,679 -2 256
B0 - D0 bright 4,664 -9 171
faint 8,139 38 290
B0 - D2 bright 6,899 28 174
faint 15,339 127 304
B0 - H2 bright 2,687 -43 172
faint 6,657 -23 301
What does this show?
TASS camera values vs. Landolt values, as a function of color
In this section, I compare magnitudes on a star-by-star basis from individual TASS cameras against magnitudes from the Landolt catalogs. I plot the residuals versus Landolt color (V-I) to see if there are significant color terms in the data from each camera.
All stars considered in this section are good; they must have been detected on at least 10 occasions (not by each camera, but just overall).
All graphs show magnitudes in units of millimags. Plus symbols (+) denote stars which are brighter than mag 11, and small circles (o) denote stars which are fainter than mag 11.
First, V-band residuals versus color (V-I).
Next, R-band residuals versus color (V-I).
Now I-band residuals versus color (V-I); the first graph includes 4 of the cameras, and the second graph the last of the 5 I-band cameras.
What does this show?