4. Veiling derivation from low resolution spectra
The need of high to very high spectral resolution for veiling derivations, claimed by various authors, comes naturally within the framework of the use of individual absorption lines . Based on this standard approach, Gullbring et al. (1998) have calculated veilings in the visible domain from the lowest spectral resolution used so far, of about 2000. The problem we would like to study here is: can we derive veilings of CTTS from much lower spectral resolution, of a few hundreds only? The answer is clear: we cannot within the current framework for the following reasons. At very low spectral resolution, it is no longer possible: i) to properly isolate individual absorption lines and, ii) to use bandpasses of a few tens of Angstroms only, because spectral mismatches together with the small number of independent spectral points, small spectral contrasts and [see Eq. (4)] and small q's, will lead to unsuperable biases.
To overcome these difficulties, we must change the framework. A very simple approach consists to work on large scale structures of high contrast, extending over spectral widths of a few hundreds Angstroms, and to make a point to point energy balance between the CTTS and the template. The prominent and deep structure seen from 4950 to 5280 Å in K and M spectral types is ideal for this purpose, although it is not the only one which can be used. In this context, we propose to approximate the p function by a second degree polynom and to solve Eq. (1) for a single veiling value R which, given that the veiling is slowly varying with wavelength, would represent its mean value within the band.
In order to test experimentally these ideas, we used two low resolution spectra (400) of BP Tau observed with the Faint Object Spectroscopic Camera (Zickgraf et al. 1997) at the 2.1m telescope of the Guillermo Haro Observatory at Cananea (Sonora, México). They were obtained during the night of November, 1995, with 5 minutes integration each, one spectrum 9 hours before (JD=2450049.95) and the other spectrum 7.5 hours after (JD=2450050.63) the BP Tau high resolution spectrum presented in the previous section. Unfortunately, HD 201092 has not been observed at Cananea, hence we used the ELODIE high resolution spectrum of this star properly reconstructed from the spectral orders. The BP Tau spectra from Cananea (with hydrogen emission lines removed) and the HD 201092 spectrum from ELODIE, both smoothed by gaussian filtering to the same spectral resolution of approximatively 300, are shown in Fig. 3. The structure selected for veiling derivation between 4950 and 5280 Å is indicated by the two vertical dotted lines and corresponds to q values of about 7 and larger than 100 for BP Tau and the template (spectral contrasts of 7% and 15%), respectively. The region between 5160 and 5200 Å, which is highly contaminated by emission or partially filled lines, has been excluded from the fit. More generally, strong emission lines must be excluded. Weak emission lines have no appreciable effect on the derived veiling value because their energy content is small. They will be interpreted by the algorithm as slight spectral mismatches, often comparable to systematic errors, and will only produce slightly larger veiling errors.
The results of the fits are shown in Fig. 4 and the calculated veilings, for JD=49.95 and for JD=50.63, are reported in Fig. 2 (black points) and separated along the x-axis for better clarity. The agreement between the veiling derivations from the quasi-simultaneous (within a few hours) high (10000) and low (300) resolution spectra of BP Tau is excellent. These results validate the proposed algorithm and demonstrate that the veiling can be extracted from low resolution spectra of CTTS by making a point to point energy balance between the object and the template over large scale and deep spectral structures.
© European Southern Observatory (ESO) 1999
Online publication: September 13, 1999