Forum Springer Astron. Astrophys.
Forum Whats New Search Orders

Astron. Astrophys. 351, 644-656 (1999)

Previous Section Next Section Title Page Table of Contents

2. Observations and data reduction

Time series of infrared spectra have been obtained for a sample of 15 short period SRVs with the Coudé Feed Telescope at Kitt Peak National Observatory using the NICMASS camera as detector. This spectrograph-detector combination is described in Joyce et al. (1998). The setup was chosen to achieve a resolution of 44000. At this resolution the wavelength range covered with NICMASS is rather small, only about 40 Å. Observations were done in two wavelength ranges with central wavelength at 16412 Å (wavelength region 1) and 16461 Å (wavelength region 2), respectively. In these wavelength ranges vibration-rotational lines of CO 7-4 dominate the spectrum, including the band head at 1.64 µm (wavelength region 1). Fig. 1 illustrates the typical spectral features observed in the two wavelength regions. Lines of similar transition are found in both ranges, therefore we do not expect a significant difference in the velocity behavior between them. The selection of the wavelength range was further limited by the response curve of the blocking filter necessary to limit the background radiation of the spectrograph.

[FIGURE] Fig. 1. Sample spectra of g Her for wavelength region 1 (upper panel) and wavelength region 2, respectively. Several spectral features of interest are marked. `A' denotes telluric features. The units on the Y-axis are arbitrary.

Observations were obtained in March (four runs), May (one run) and June 1995 (one run) and in March 1996 (two runs). Each run consisted typically out of 2 to 3 observing nights with a switch in central wavelength from one range to the other between them. Telluric lines from the spectra of A-type stars were used for wavelength calibration as the chosen wavelength range did not cover enough comparison lines from the calibrations lamps available. The stars in our sample were selected both on their infrared brightness and to sample the period range between 30 and 150 days. Only oxygen rich stars of spectral type M were included. The stars of our sample are listed in Table 1 with some of their fundamental properties. Typically, the stars were observed once each night (see last column of Table 1).


Table 1. Properties of the sample stars. Data are from the General Catalogue of Variable Stars (Kholopov et al. 1985-88, GCVS4). Column 7 lists the SRV subtype according to the new classification scheme by Kerschbaum & Hron (1992, 1994). The last column list the number of nights each star was observed for this investigation.

From the results found for the velocity variations from miras and long period SRVs we expected variations with only a few km s-1 amplitude (Hinkle et al. 1997). Therefore a high accuracy in the determination of the velocity was necessary. As most of the lines in the selected spectral ranges are still blended at the used resolution, we used a correlation method to determine the velocity variations between the observing nights. For the correlation we used a program written by Schultheis (1998) which is based on an earlier code by Stift (1986). Fig. 2 gives a typical resulting correlation function. The minimum is well expressed and calculated by fitting a parabola through the three points centered on the minimum. A possible influence of telluric lines has been checked and can be excluded. After correction of a detected velocity shift introduced by an instability of the mounting of the NICMASS dewar we achieved a final accuracy of approximately 350 m s-1 for the shift between two nights for most of the nights (for details on this correction see Lebzelter 1999a). The two wavelength regions were analyzed completely independently such that a similarity in the velocity variations could be used as a check for the reliability of the results obtained with the adopted correction method. As will be shown later, the agreement is satisfying and it can be assumed that the required accuracy could be reached. A further check of the reduction applied was possible by the comparison of velocity variations from spectra of R Leo, additionally obtained in each night, with published results for this star (Hinkle 1978, Hinkle et al. 1984). A comparison of the results can be found in Fig. 3. The FTS results both for [FORMULA]v[FORMULA]3 and for high excitation [FORMULA]v[FORMULA]2 CO lines are plotted with small symbols, while our Coudé Feed/NICMASS velocities are plotted with large filled or open boxes for wavelength region 1 and 2, respectively. The absolute velocity for wavelength region 1 has been derived as described in Sect. 3.2. As for wavelength region 2 no absolute velocity calibration was available, we calculated them with the assumption that R Leo had the same velocity in wavelength region 2 at JD 2450174 as in wavelength region 1 at JD 2450175. It has to be stressed that, as this assumption is rather arbitrary, the absolute velocities of region 2 shown in Fig. 3 are no measured results (only the velocity shifts were measured). They were only calculated to allow an easy comparison with region 1 and the FTS velocities. Typical error bars have been used for the Coudé Feed data. Note, that the FTS velocities have been obtained about 20 years (i.e. more than 20 light cycles of R Leo) earlier than the NICMASS velocities. Taking into account the well known cycle to cycle variations in the velocitiy curves of miras (e.g. Hinkle et al. 1997) the agreement is satisfying.

[FIGURE] Fig. 2. Example of a typical correlation function used to determine the velocity shift.

[FIGURE] Fig. 3. Comparison of velocities obtained with the Coudé Feed/NICMASS with FTS velocities from Hinkle (1978) and Hinkle et al. (1984). Large symbols mark Coudé Feed velocities, small symbols FTS velocities from [FORMULA]v[FORMULA]3 and high excitation [FORMULA]v[FORMULA]2 CO lines. See text for more details. In the case of line doubling (FTS) only the smaller velocity component has been plotted. The error bars in phase of some of the FTS measurements are due to the fact that Hinkle (1978) gave only velocities averaged from several spectra obtained within a few days.

It was originally planned to derive not only velocity shifts but absolute velocities for each observation. However, the stars we wanted to use as velocity references (the irregular variables UW Lyn, µ Gem and CF Boo) turned out to be not stable enough in velocity either. K-giants could not be used as their spectra were not similar enough to the M-stars of our sample for a satisfying use of the correlation method. Therefore, in Sect. 3.1 we will present only velocity shifts relative to the last two observing nights of our data set (March 1996). Sect. 3.2 will report on an estimate of the absolute velocities with the help of a contemporaneous FTS observation of three stars in our sample.

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 1999

Online publication: November 3, 1999