Astron. Astrophys. 336, 613-625 (1998)

2. Observations and reductions

A sample of five early M dwarfs, shown in Table 1, was selected from the combined photometric lists of Leggett (1992) and Stauffer & Hartmann (1986). The selection criteria were: , , , and . The criterion limits the MK spectral type to the range of K8 to M4. This constraint was placed on the sample because the extensive line blanketed chromospheric modelling of Short & Doyle (1997, henceforth SD) is limited to models with equal to 3700K, which corresponds to a spectral type of M0 or M1 (Lang 1992, Mihalas & Binney 1981). The sample was chosen to cover a range of chromospheric activity level from low (Gl 212, Gl 382), through intermediate (Gl 900) to high activity level (Gl 388 (AD Leo), Gl 494).

Table 1. Program objects

Table 2 contains a journal of the observations. Echellograms centered at were taken with the Utrecht Echelle Spectrograph (UES) of the William Herchel Telescope (WHT) at the Observatorio Roque de los Muchachos in La Palma on the night of December 23/24, 1996 (JDN=2450441). We used a 79 l mm^-1 grating that gives , and a TEK 10241024 CCD detector. The H line falls near the centre of the detector in order 34 and the Na I D doublet falls near the centre in order 38. Multiple exposures with a maximum exposure time of 1800 seconds were taken of most objects and were co-added to increase . The total values are shown in Table 3.

Table 2. Journal of Observations

Table 3. Signal-to-noise values in yellow and red, and and FWHM (emission) of H in

All reductions were performed with the NOAO Image Reduction and Analysis Facility (IRAF) The stellar frames were corrected by subtraction of the mean of twelve bias frames taken throughout the night, and division by a mean of 32 bias subtracted W lamp flat field frames that were corrected for scattered light (see below) and normalized to a signal level of approximately unity. Cosmic rays were removed from the bias and flat field frames during the calculation of the average frame by clipping. Stellar exposure were bracketed by exposures of a Th-Ar arc lamp for wavelength calibration.

Both the mean flat field frame and the stellar frames were found to have significant levels of undispersed background light visible between the echelle orders as a result of internal scattering in the spectrograph. The mean background level varies from exposure to exposure, but was typically 50 to 100 analogue-to-digital conversion units (ADUs) in the stellar frames. A function was fit to the background light with a sixth order cubic spline perpendicular to the dispersion and a twelfth order cubic spline parallel to the dispersion. These high order fits were required to approximately fit structure of high spatial frequency in the background scattered light. The fitting was performed interactively and the fits to each row and column were visually inspected to insure that there were no spurious peaks in the fitting function. The 79 l mm^-1 grating was used precisely because its wider order separation allows a closer fitting of the background light. Typical RMS deviations from the fit were six ADUs for the fit perpendicular to the dispersion and two ADUs for the fit parallel to the dispersion.

The spectra were extracted using variance weighted extraction (Horne 1986) with a model point spread function (PSF) fit to the strongest stellar exposure of the night. Cosmic rays were removed from the stellar exposures during extraction by clipping.

Bright Na I D sky lines were visible in the stellar frames in the inter-order region between order 38 (the Na I D order) and order 37, but, surprisingly, not in the region between orders 38 and 39. Therefore, we also extracted a variance weighted sky spectrum from the region between orders 37 and 38. These sky spectra were scaled to compensate for the different aperture sizes used in the stellar and sky extractions and then subtracted from the stellar spectra. Because the star was centered in the slit, the lack of detectable sky lines in the region between orders 38 and 39 is difficult to understand. However, in the context of this work, the important consideration is to be able to extract the profile of the sky lines, which then may be scaled, and subtracted from the stellar spectrum, which we were able to do.

Rectification is complicated by the complete over-blanketing of M star spectra in the yellow and red. We obtained a rectification function by fitting a third order Legendre polynomial to the highest peaks of the pseudo-continuum. However, we have probably overestimated the continuum level. We calibrated the wavelength scale of the entire echellogram in each stellar frame by fitting a dispersion function to the positions of 100 lines in the mean of the Th-Ar exposures accompanying the stellar exposure. The RMS deviation is .

© European Southern Observatory (ESO) 1998

Online publication: July 20, 1998
helpdesk.link@springer.de