1.4 Young Stars
2.1.3 CTIOPI observing
My involvement in this process has mostly been as an observer and user of a functional pipeline. CTIOPI (the Cerro Tololo Inter-american Observatory Parallax Investigation) started in 1999 as an NOAO Surveys program, and since 2003 has been under the auspices of the SMARTS Consortium. Originally, it consisted of two parallax programs, one on the CTIO 1.5m and one on the CTIO 0.9m, but the 1.5m program (Costa et al. 2005, 2006) was shut down in 2003.
CTIOPI observing takes place over (usually) whole weeks of time on the CTIO 0.9m Boller & Chivens Cassegrain reflector, with its dedicated Tek 2048 #3 CCD. Within that week, we observe (ideally) around 140 parallax targets. The rough rule of thumb is that a parallax target is to be observed five times (or for 30 total minutes of observing time, whichever is less) in a particular filter. Filters, out of CTIO’s #2 Tektronix V RI filter set, are selected to maximize the number of available bright and spatially nearby reference stars – a star might be brightest inV, but if theR orI filter yields more reference stars of suitable brightness, we use that filter. Our centroids on stars, and consequently our astrometric solutions, improve with longer exposure times up to several minutes as the atmospheric distortions blur to randomness. Thus, bright red stars are often observed in V where the exposures take longer. Faint stars are often observed in I so the star is sufficiently well- exposed in 10 minutes, our adopted maximum for a single exposure.
The observations that are most important for high quality parallaxes are at the ends of the RA extensions of the parallax ellipse – found when the star transits at dawn or dusk.
Thus, the first and last hour of any given night will contain the most important observations4.
RECONS nevertheless observes stars whenever they are visible, as it improves our ability to see astrometric perturbations caused by (typically) unseen companions. For proper motions, timespan is the critical factor: the longer the spread between first and last epochs, the more well-determined the proper motion is – hence the importance of the plates from the circa-1900 Astrographic Catalog (Urban et al. 1998) to modern (e.g. TYCHO-2) audiences. RECONS typically requires at least 2 years of observations and at least 30 frames in the morning and evening halves of the eclipse before publishing (with the hope that at least 20 of each are used in the final reduction), and is reaching 13 years of coverage for some long-term targets (e.g. GJ 1207).
Because RECONS uses wide-band V RI filters, the effective centers of the bandpasses are very different for an M star and an A star. These differences lead to Differential Color Refraction (DCR) effects, which affect the positions of stars in images, and must be corrected. To minimize DCR corrections, RECONS only observes stars as they transit the meridian – their minimum zenith distance, plus or minus 2 hours (±30 minutes is preferred)5. These
restrictions are not always helpful, as stars at +30 DEC (the northern limit) or −90 DEC will always have zenith distances of at least 60 degrees. RECONS has been coordinating the CTIO 0.9m for SMARTS since 2003, which has left us in a position to maintain the optical path by changing the telescope, filters and camera as little as possible6. This minimizes
4For a star near an ecliptic pole, midnight is just as important, as the DEC extension of the parallax
ellipse is quite large.
5Another method to reduce DCR corrections is to use filters with narrower bandpasses, but they would
also require longer integration times.
other optical distortion effects (or at least keeps them consistent).
The selection of reference stars is a key element in measuring an accurate parallax. Good reference stars are bright in the chosen filter, close in angular separation to the target, and as a group surround it on all sides. In many cases CTIOPI does not center the target in the field to bring in other reference stars that more fully surround the target. Parallax reductions use between 5 and 12 reference stars. RECONS arrived at this concept by informal experimen- tation, but is more formally known and mathematically defined as the geometric multiplier, and as “dependencies” (van de Kamp 1981). In practice, we must use the reference stars available, and they often fall short of the ideal.
Trigonometric parallaxes currently have limited reach. HIPPARCOS, with a typical 0.7 mas error, is only accurate to 10% within 150 pc. CTIOPI is typically only that accurate out to 66 pc; we usually drop stars with preliminary parallaxes beyond 100 pc. (The most accurate measurements are from long-baseline radio interferometers with sub-milliarcsecond parallaxes; no long-baseline optical interferometers currently do parallax work.) Aside from these concerns, the time required (12 visits of 30 minutes each spread out over at least two years) makes obtaining large numbers of parallaxes difficult. HIPPARCOS, with nearly 120,000 stars including allstars brighter than V=7.3, obtained parallaxes to 14 times more systems than all ground-based observatories before it combined. The biggest change to come is ESA’s Gaia mission, a follow-up to HIPPARCOS, which promises to measure parallaxes
ity of the CTIO 0.9m (constructed 1965) and its CCD (in continuous use since at least 1994, http://www.ctio.noao.edu/ccd info/ccd news.html#6 retrieved 15 MAY 2012) and the maintenance staff at CTIO, are in no small part responsible for our results being competitive with parallaxes from newer, more powerful instruments (Dupuy & Liu 2012; Faherty et al. 2012).
for 107 stars from V=6 to V=20 at the 10-100 microarcsecond level, which will extend the
penetration of accurate (10%) parallaxes out to kiloparsec scales (Lindegren 2007).