TN 0019: Photometric properties of Vermont triplet: Aug 21, 1996

Author: Michael Richmond
Date: 960903
Revision: #1 960903
Key Words: photometry, filters, catalogs

Contents

Tables

Introduction

This technical note contains an analysis of a very small portion of the data collected by triplet #3, located in East Braintree, Vermont (longitude 72:40, latitude +44:03, altitude 1520 feet), on August 21, 1996. It is concerned primarily with the photometric properties of the TASS cameras, and their relationship to the standard Johnson-Cousins magnitude scale.

Acquisition and reduction of Landolt fields

During the night, the cameras scanned over a number of fields in which Arlo Landolt has measured photometry in the Johnson-Cousins UBVRI system. See his papers

There are fields containing many standard stars in the Selected Areas on the celestial equator. These areas occur roughly one hour apart, and cover about one square degree on the sky. Each one fits easily inside a TASS strip 3 degrees wide.

I concentrated on the following Selected Areas:

name          RA (2000.0)     Dec (2000.0)    number of Landolt stars

SA 113          21:41            +00:30              46
SA 114          22:41            +01:00              13
SA 115          23:42            +01:00               9

I was able to get I-V-I coverage of SA113 and SA115, but only V-I for SA114 (some house lights that came on briefly in the middle of the night obscured stars in the East I-band frame of SA114). I used the XVista programs to process the frames, which I broke up into 1100-row segments (thus, each segment was about 15 minutes long, or about 4 degrees, and 3 degrees wide).

When I tried to compare the detected star lists with Landolt's catalogs, I discovered that many of the Landolt stars were not detected; a few were too bright, many were too close to other stars to be measured separately, and many others were fainter than could be seen on the TASS images. In order to get the largest possible sample for comparing Landolt and TASS magnitudes, I displayed each image and compared it by eye with finding charts for the Landolt fields; I measured manually aperture magnitudes for each unblended, unsaturated star.

My image reduction and measurement routines need some work; improvements to the methods, and tuning the parameters to match the TASS images, should (I hope) allow me to use the automatically generated stars lists in the future.

After comparing the TASS images with Landolt fields, I was able to identify a number of Landolt's standards in each image. The Table below describes the properties of each image and the number of stars I was able to measure:

             Table 1. Properties of TASS images of Landolt fields

                             sky  skysig     FWHM       aperture radius
field   file  dir  filter   (ADU) (ADU)    pix (arcsec)   pix (arcsec)   Nstar
-------------------------------------------------------------------------------
SA113   n_4c   E     I      1363   28    3.5-4.4 (49-61)   4.0 (55.5)     14
        n_5b   S     V       228   16    1.5-2.1 (21-29)   2.0 (27.8)     23
        n_6a   W     I      1888   32    2.8     (39)      2.8 (38.9)     19

SA114   n_6b   S     V       198   16    1.5-2.0 (21-28)   2.0 (27.8)     12
        n_7a   W     I      1820   41*   2.8     (39)      2.8 (38.9)      7

SA115   n_6c   E     I      1198   28    3.7-4.3 (51-57)   4.5 (62.5)      9
        n_7b   S     V       193   17    1.4-1.9 (19-26)   2.0 (27.8)      8
        n_8a   W     I      1618   29    2.8     (39)      2.8 (38.9)      3

-------------------------------------------------------------------------------
Notes: assumes pixel scale=13.88 arcsec/pixel
       sky varies in frame n_7a: brighter to East
       focus is worse on N side of frame n_6c
       focus is better on S side of frame n_8a
       most of the SA115 stars fell outside of frame n_8a

Detection thresholds in Landolt fields

Some stars were barely visible in the TASS images, so faint that it was impossible to make any accurate measurement of their brightness. Nevertheless, these stars were clearly present. It is possible that objects falling into this No-Man's-Land could be useful under special circumstances: if a nova is discovered inside the area covered by TASS, one could go back to pre-discovery images and search the known position of the nova for evidence of its presence. There is a definite difference in the limiting magnitude of an image, depending on the manner in which one defines the phrase:

For most purposes, it is the first definition which is more useful.

When I tried to measure the brightness of Landolt's standard stars, I found the following limiting magnitudes:

You can see for yourself in the following image subsections, which are taken from the field of SA 113. Each subsection is oriented North at left, East up, and is about 24 arcminutes E/W by 30 arcminutes N/S. The images are centered on star SA 113_366, which has coordinates RA=21:41:54, Dec=+00:29:21 (J2000). First, the V-band image, displayed at 2 contract settings:


Now, the West I-band image:


Here's a table of the stellar magnitudes in this area, taken from the Landolt papers:

  star        RA       Dec       V        I
SA113_366  21:41:54 +00:29:21  13.537  12.326 
SA113_372  21:42:02 +00:28:36  13.681  12.915 
SA113_260  21:41:48 +00:23:52  12.406  11.800 
SA113_267  21:41:56 +00:20:44   7.653   7.071 
SA113_475  21:41:51 +00:39:19  10.306   9.208 

As for the brightest stars that can be measured accurately, I discovered that a few of the brightest stars in the fields didn't match the Landolt magnitudes very well. Most of these stars showed evidence for saturation in their radial profiles -- that is, a plot of intensity vs. radius revealed that the innermost part of the profile was flat. The stars which appeared to be saturated had magnitudes

Summing up, I conclude that the range over which these images showed stars which could be calibrated photometrically is

I do not know the accuracy with which one can measure the brightness of saturated objects, but it may be possible to use photometry of stars which are only slightly saturated.

Sky brightness

Once I had identified standard stars in the Landolt fields, I could calculate the sky brightness. First, I used a large aperture (radius = 10 pixels = 139 arcsec) to integrate the light around a bright, relatively isolated star; that yielded the count/magnitude relationship for the image. Next, I measured the sky value in the image, fitting a Gaussian to a histogram of binned pixel values. The fitted sky values are shown in Table 1. I then converted the sky value from counts per pixel to counts per square arcsecond, using a pixel size of 13.88 arcseconds per pixel, close to that found in TASS Technical Note 9 and TASS Technical Note 10. Finally, given the ratio of brightness between the integrated light from the star and the light in a single square arcsecond, I could calculate the magnitude of that square arcsecond. The results are shown below.

                 Table 2. Measurements of Sky Brightness 

field  file   dir  filter   sky (mag/sq. arcsec)  approx EDT    zenith angle
------------------------------------------------------------------------------
SA113  n_4c    E     I          18.1                 23:30       44
SA115  n_6c    E     I          18.2                 01:30       44

SA113  n_5b    S     V          20.8                 00:30       46
SA114  n_6b    S     V          20.9                 01:30       46
SA115  n_7b    S     V          20.9                 02:30       46

SA113  n_6a    W     I          18.5                 01:30       44
SA114  n_7a    W     I          18.5                 02:30       44
SA115  n_8a    W     I          18.6                 03:30       44
------------------------------------------------------------------------------
Notes: sunset  occurred at 19:46, sunrise  at 06:00 
       moonset             23:46, moonrise at 13:18

For comparison, Garstang (Publications of the Astronomical Society of the Pacific, vol 101, p. 306 [1989]) lists the following values of sky brightness at several observatory sites (his list includes many more sites, too).

                    sky brightness in V (mag/sq.arcsec)
site        year      zenith angle=0    zenith angle=45

Lick        1966          20.8              20.4
Palomar     1972          21.5              21.7
Kitt Peak   1987          21.9
Sac Peak    1978          21.9              21.7
Lowell      1982          21.1              20.8
CTIO        1988          21.6 

I'm very happy with the measured value of V=20.9 (at zenith angle 45 degrees) for the TASS site in Vermont. It's not world-class, by any means, but it's better than the brightness at a few professional observatories, and certainly much better than that in New Jersey.

Color terms for the TASS cameras

My main goal in this analysis was to find the color terms for the TASS cameras; that is, the corrections one must apply to convert raw, instrumental magnitudes measured with the TASS equipment to the standard Johnson-Cousins UBVRI magnitude system. In the simplest form, these equations would be

        V = v + k1*(v - i)       (1a)
        I = i + k2*(v - i)       (1b)
where
        V,I    are the magnitudes in the Johnson-Cousins system
        v,i    are the instrumental magntitudes

Note that one can also write the color terms in a slightly different form, using the known magnitudes as the independent variables:

        v = V + k1'*(V - I)      (2a)
        i = I + k2'*(V - I)      (2b)
Given a set of standard stars of known brightness in the standard system, it is easist to solve equations 2a and 2b to find the primed coefficients k1' and k2'. In order to convert the raw, instrumental magnitudes onto the standard system, however, one must solve for coefficients k1 and k2.

In general, it is true that the magnitude of the primed and unprimed coefficients are approximately equal. I wanted to find out if the TASS natural system -- that is, the response of the camera lenses, filters, optical window and CCD chip -- came close to matching the Johnson-Cousins system (which is defined formally by two reflections off aluminum, a filter, and a photodiode). The closer the match, the easier it will be to transform the TASS magnitudes to the standard scale, and the more accurate will be the transformations.

I will compare the measured and standard magnitudes for stars detected in several the Landolt fields, and solve equations 2a and 2b for the primed color terms. The closer the coefficients are to zero, the better.

For each field, I measured aperture magnitudes in the images, and compared them with tabulated magnitudes from the Landolt papers. The simplest comparison states that the two sets of magnitudes are have no color-dependent terms (i.e. k1' = k2 = 0.0), and seeks a single offset that will convert the measured to the standard magnitudes.

             V  =  v + zp 
             I  =  i + zp
I split the stars into bright and faint subsamples for this task, and calculated the required offset for the entire sample, plus each subsample: Here are the results:
   Table 3.  Photometric solutions in Landolt fields -- no color terms

                         all stars     "bright" stars     "faint" stars
field      images      zp   sig   N      zp   sig   N       zp   sig   N
--------------------------------------------------------------------------
SA 113       V        4.51  0.07  16    4.50  0.03  6      4.51  0.09  10
           East I     4.99  0.15  10    5.03  0.06  6      4.93  0.24   4
           West I     4.38  0.08  14    4.33  0.02  5      4.40  0.10   9

SA 114       V        4.60  0.04  11    4.60  0.05  8      4.61  0.02   3
           West I     4.34  0.05   7    4.32  0.02  3      4.36  0.06   4

SA 115       V        4.57  0.03   8    4.56  0.03  6      4.59  0.01   2
           East I     5.03  0.12   9    4.96  0.03  5      5.11  0.14   4
           West I     4.37  0.02   3    4.37  0.02  3
--------------------------------------------------------------------------
Notes: "bright" stars are V <= 12.0, I <= 10.0

The next level of approximation allows the existence of a color terms -- that is, the value of the coefficients k1' and k2' in equations 2a and 2b are allowed to be non-zero. For this calculation, I used all stars, given them equal weight (which is not the proper thing to do).

   Table 4.  Photometric solutions in Landolt fields -- with color terms

                         all stars       
field      images       zp  coeff  sig   N       95% conf of coeff
--------------------------------------------------------------------------
SA 113       V        4.41   0.11  0.07  16         +/- 0.17
           East I     4.85   0.17  0.16  10         +/- 0.45
           West I     4.49  -0.14  0.08  14         +/- 0.19
 
SA 114       V        4.57   0.03  0.08  11         +/- 0.08
           West I     4.31   0.02  0.05   7         +/- 0.10

SA 115       V        4.56   0.01  0.03   8         +/- 0.11
           East I     5.26  -0.27  0.11   9         +/- 0.37
           West I     4.37  -0.16  0.02   3         +/- 3.20
--------------------------------------------------------------------------
Note that in no case is a color coefficient significantly different from zero. This is good news -- it means that the TASS and Johnson-Cousins photometric systems are not very different, and that it should be possible to convert between the two with reasonable accuracy. However, I do not believe that I have enough information to make a true photometric solution; that would require two or three times as many stars as I measured for this night.

Shown below are plots of the differences between instrumental and standard magnitude, as a function of standard (V-I) color, for each of the fields listed above.




Comparison of East and West I-band photometry

My triplet has I-band filters in the cameras pointed to the East and West of the meridian. Stars pass first through the East I-band camera, and two hours later, through the West I-band camera. The East camera points over my parents' house, and is therefore likely to be contaminated by scattered light; as Table 2 shows, the "sky" brightness is considerably higher in the Eastern camera. Moreover, the East camera was less well focussed than the West camera during this run.

Nevertheless, one can compare the raw, instrumental magnitudes of stars detected by both cameras. This should give some indication of the internal accuracy of measurements, and provide an estimate of the smallest variations in light which might be detected over many nights.

Since the Eastern I-band image of SA 114 was contaminated by house lights, there are only two fields for this comparison. I show below the differences between measured aperture magnitudes of Landolt standard stars; there are, of course, many more stars in the field which could be used in such a comparison.

   Table 5.  Comparison of raw East and West I-band photometry

                         all stars     "bright" stars     "faint" stars
field     difference   zp   sig   N      zp   sig   N       zp   sig   N
--------------------------------------------------------------------------
SA 113  East - West   0.69  0.05   8    0.69  0.05  8                     

SA 115  East - West   0.57  0.01   3    0.57  0.01  3                     
--------------------------------------------------------------------------
Notes: "bright" stars are I <= 10.0
There are certainly too few stars from which to draw any firm conclusion, but it seems reasonable that one might be able to detect variations of less than 0.1 magnitude over the course of several hours by comparing photometry from the East and West cameras of a triplet.