While strictly outside of the aperture-size restrictions of this workshop, we present a variable star survey undertaken with the USNO 0.2m FASTT telescope as an example of one project that most small telescopes can perform.
The FASTT telescope is a completely automated transit instrument equipped with a 2048x2048 CCD and a 50arcmin field of view. Each pixel is 1.5arcsec in size; the telescope is defocussed so that the image FWHM is typically 3arcsec, thereby removing the undersampling and improving the astrometry and photometry. The FASTT and scanning techniques are discussed in detail in Stone et al. (1991, AJ, 111, 1721).
The original motivation for the 0.2m survey of 16 regions arranged along the celestial equator was to set up astrometric calibration regions for the SDSS survey discussed by Gunn and Knapp (1993 ASP Conf. Series 43, p267). Each region is approximately 3.2 x 7.5 degrees in size, for a total area of 24 square degrees per region or 384 square degrees for the survey. Each region was scanned in 50arcmin strips, with 202 seconds of integration per object within the strip. Strips were overlapped center-to-edge, and block adjustment reductions used to tie scans into a common reference system. Mean positions for the 679,866 stars included in the survey regions are known to better than 50mas; mean photometry for the non-variable objects is accurate to ±0.02 mag.
The methods used to compute the magnitudes needs further discussion. Namely, standards from Landolt and the Harvard E-fields were observed with the FASTT in order to set up a grid of secondary standards with accuracies of ±0.008 mag. These standards were observed each night and used in the usual reductions to compute magnitude zero-point and primary extinction values that, in turn, were used to reduce all aperture readings to a FASTT instrumental system, hereby designed by R8, which closely matches the SDSS r-passband and is similar to the Cousin's R-passband. If (V - R) colors should become available, then the FASTT R8 magnitudes can be converted to true R magnitudes using the transformation given in Stone and Pier (1996, BAAS 27, 1389).
Except for obvious blends, the number of stars observed in each SDSS field are reasonably complete down to an apparent magnitude of R 17. The four fields E, F, M, and N are exceptions. These fields are at low Galactic latitude and contain very high densities of stars. In order to reduce the processing, these fields are only complete to R 14 mag; however, fainter stars were observed in subareas of these zones down to the limiting magnitude of the FASTT. These subareas comprise 17% of the total area observed in these zones.
Efforts were made to observe SDSS fields only on nights of good photometric quality, in that nights with high levels of extinction were eliminated. Many of the nights were photometric or mostly photometric during the observing time. The photometric nights were identified by comparing individual nights, where differences in magnitudes would be very small if both of the nights were photometric. Mean magnitudes and standard errors were then formed for all stars observed on photometric nights. In general, these means were computed from 5.5 to 8.6 measures obtained on different nights. The formal accuracy of these magnitudes is ±0.015 mag, and their rms-scatter is ±0.03 mag, except for the faintest stars where the scatter is ±0.12 mag.
Three criteria were used to identify variables: photometry, astrometry, and image profiles. If the rms-scatter was larger than 3 X the expected error in the magnitude, then the star was identified as a possible variable. This initial cut yielded 4000 potential variables. However, it was found that many of these stars were actually close pairs which appeared as single stars on nights of good seeing and blends on nights of poorer seeing, causing the observed magnitudes to falsely vary. Fortunately, there is good astrometry for all of the SDSS calibration stars, and these false variables can be often identified as those stars with large astrometric errors. Possible variables with 5- or greater X the expected errors in their positions were removed from sample as likely misidentifications. The astrometric error criteria reduced the number of potential variables to 2060. A final cut was made by removing those possible variables that had only one discrepant photometric measure, and for whom the ratio of FWHM in x and y for that measure was more than 2- from the mean for the star. This usually indicated that some instrumental problem occurred on that night, or that a blend occurred that did not strongly affect the image centroid. This criteria rejected an additional 223 candidates. We then used non-photometric nights and differential techniques to extend the amount of photometry for each variable, resulting in 8-10 total measures per variable in the survey.
A total of 1837 variable stars have been identified. Outside of fields E,F,M and N, where we do not do complete sampling, there are 55 previously known variables and 770 new variables in the remaining 12 fields. Of the known variables, we identify 35. Of the remaining 20, 15 are either too bright or too faint for detection, and 5 are missed for some other reason. This means that 96 percent of the detected variables in this survey are previously unknown. The variables constitute 0.3 percent on average of all the stars identified in this survey. If this ratio of new/old variables is held, and since our survey only covers 1percent of the sky and avoids most of the galactic plane, then such a telescope (or pair of telescopes) surveying the entire sky would detect many tens of thousands of new variables in our galaxy.
More information about the survey, or a list of variables within a particular category, can be obtained from the authors. While we intend to investigate small subsets of these variables, the majority are available to interested observers. Note that these stars are true variables, have accurate coordinates, an estimate of the variable type, light curve and period, and have excellent finding charts.