OBSERVING CAMPAIGN FOR THE 2011 PERIASTRON PASSAGE OF DELTA SCORPII
Spectroscopic observations and recommendations how to obtain radial velocity
The periastron passage in delta Sco may lead to exciting interactions between the two stars and between the stars and the stars and the circumstellar matter (the latter manifests itself through emission lines such as in H_alpha). Since the orbital period of delta Sco is of the order of 10 years, and similarly eccentric systems with an early B-type primary are not common, the 2011 event will be a rare opportunity to gain deep insights into the physical responses to such an event.
Through observations permitting sufficiently accurate radial-velocity measurements, it will be possible to substantially improve on the uncertainties of the predicted date of the periastron. Such an improvement will enable us to use scarce resources at some of the world's premiere observatories much more effectively by scheduling the observations for exactly the moment when the main action is expected to happen.
If you find this an interesting project to support, you will below find some recommendations on how to best organize such observations and their reduction.
Any contribution of the kind described will be a great help, and we will be very grateful for everyone of them.
Many thanks for your interest and kind regards,
Dietrich Baade, Thomas Rivinius, and Stan Stefl
(European Organisation for Astronomical Research in the Southern Hemisphere)
on behalf of a much larger worldwide community
================================================================================
GENERAL
The latest forecast for the periastron passage in the delta Sco binary is early July 2011. However, the uncertainty can be as large as a few days while it is important that specific observations aimed at hitting the event must be timed much more accurately.
The best empirical criterion to predict the date of the periastron passage is the radial velocity (RV) of the primary star. Because of the high eccentricity, the RV stays roughly constant for the most part of the orbital cycle and begins to change significantly only about one month before the closest encounter.
In the case of delta Sco, "significant" means about 10 km/s for the first week, which is unambiguously detectable from carefully calibrated spectra. The maximum variation of the radial velocity is expected to reach about 40 km/s.
Note that is not
necessary to achieve high precision in the absolute
RV. Calibration errors of a spectrograph/telescope and specific spectral lines
do not matter as long as they are constant with time so that variations
can still be inferred with high confidence. Therefore,
o Select spectral lines without too obvious emission components.
o Check your equipment/procedures (including data reduction!) around March/April
when there should be no major variability yet of the RV.
o Once you are satisfied with the tests, do not change any part of the procedure
and follow it rigorously.
For a single observer, it would be necessary to observe delta Sco almost daily, starting around mid-April. But with a large community, it will be enough if everyone observes delta Sco roughly once a week and submits the RV measurements soon after the observations.
Once the measured RV has dropped by 30 km/s or more, the observations can be terminated unless they regularly achieve a S/N of 100 or more and have a spectral resolving power, R, greater than 15,000. In this case, see also the last paragraph.
OBSERVATIONS
One set of observations should consist of the following sequence:
- bias (average of 10 exposures)
- flat field (average of 3-5 exposures)
- wavelength comparison spectrum (lamp in your spectrograph)
- a comparison star (F-type to early M-type) or a RV standard star (see
a list of recommended stars)
- delta Sco
- a comparison star (F-type to early M-type) or a RV standard star
- wavelength comparison spectrum
If your time permits, take two such sets of observations in fairly direct succession. If the RVs of delta Sco derived from these observations differ by more than 5 km/s, neither of them should be reported, because the uncertainties are too large.
If you can, adjust exposure times such that the signal level is between 30% and 60% of the saturation limit of the detector. If in doubt, stay closer to the lower one.
For the wavelength comparison spectra, saturation of some strong lines may not be avoidable. In that case, set the exposure time such that you have 10-20 unsaturated lines with known wavelengths (i.e., no blends of two or more lines) distributed over the wavelength range of interest.
For the comparison and RV standard star, use the
following selection criteria:
o They should be bright.
o They should have a similar declination as delta Sco has.
o They should be observed at a similar hour angle as delta Sco. Since delta
Sco and these stars should be observed close in time, this means that the right
ascensions should be similar.
o A comparison star with a spectral type earlier than F has not enough spectral
lines for getting the RV, while a M-type star has too many lines which can be
blended.
You do not need to select both a comparison star and an RV standard star. One star can serve both purposes.
The times of the observations of delta Sco must be given in UT and accurate to 5 minutes. They should refer to the beginning of the exposure, and the exposure time should also be provided.
DATA REDUCTION
+ Subtract the bias as a single number unless there is significant large-scale structure in the bias observation.
+ Divide the bias-corrected stellar spectra by the bias-corrected flat field on a 2-D basis. Extract the 1-D spectra only after this 2-D operation.
+ For the subsequent wavelength calibration, make sure that your algorithm always finds and utilizes the same comparison lines. If you have enough lines to choose from, select the narrowest lines. If a second-order polynomial gives a good fit (better than 0.05 of a pixel), use it. Do not use polynomial orders larger than 3 even if the error apparently decreases (in most cases, this is really just apparent).
+ Resample the spectra to a step size about half as wide (in wavelength) as a detector pixel.
+ Check whether dividing the reference-star spectrum (your F-to-M type or radial-velocity standard star) by a second or at most third-order polynomial fitted to the continuum yields a clearly improved rectification of the continuum. If it does, divide the spectrum of delta Sco by the same polynomial function (i.e. do not fit the continuum of delta Sco) unless this division make the continuum of delta Sco more strongly bent than before.
+ Measure the RV in the so processed spectrum of delta Sco. (See instructions how to measure RV here)
+ Measure the RV of the reference star in the same way. If the two measurements differ by more than 5 km/s, do not submit the measurements of delta Sco. They would be too uncertain.
+ Reduce one stellar spectrum with both observations of the wavelength calibration lamp. If the two results differ by more than 5 km/s, do not submit the measurements of delta Sco. They would be too uncertain.
+ Otherwise, submit the raw RV; do not correct for the Earth's rotation or orbital motion.
USAGE OF THE RV DATA
The RV data will be used to establish a more precise forecast of the periastron date. It is not intended to derive a new orbital solution but everyone will, of course, be free to do so anyway.
ADDENDUM FOR OBSERVATIONS WITH S/N >100 AND
R > 15,000
From such observations, a so-called dynamical spectrum could be constructed: Plot all spectra in one frame. Use vertical offsets between them, which correspond to the time elapsed between the observations. If the wavelength coverage of the spectra is large, the wavelength (horizontal axis) should be broken up into several regions, each plotted separately so that the line profiles and possible variations are well visible.
Such observations can usefully be obtained throughout the full 2011 observing season for delta Sco. However, in order to be of practical value, at least 10 observations within 3 consecutive nights need to be obtained. There is no upper limit unless there are no trustworthy variations. Initially, it will be sufficient to submit dynamical spectra in .jpg format. But later FITS spectra would be desirable, especially in the case of major variations.
SUBMISSION OF DATA
Send all RV measurements and observing times to Anatoly Miroshnenko (a_mirosh@uncg.edu). The earlier the submission, the better.
Last updated: 11-Feb-2011