RESULTS

The spectra observed at positions (-11'', -78'') and (+33'', +105'') are shown in Figures 1 and 2, respectively. Selected sequences of vibrational-rotational lines are indicated. Table 1 lists all lines that can be identified together with their flux densities. These were determined by Gaussian fitting the spectra with the FWHM set to the width of the instrumental profile and adjusting the amplitude and wavelength as free parameters, using a least-squares fitting algorithm. In many cases lines were blended, in which case multiple Gaussians were fitted, keeping both line width and separations fixed.

However, the accuracy of this procedure can be severely affected by unequal attenuation of the line source at the specific line frequency compared to the atmospheric attenuation of the continuum source, averaged over a resolution element (see Howe, 1992). The atmospheric transmission spectrum from 1-2.5m contains a multitude of unresolved absorption lines, many of which are nearly opaque. Thus, even though the continuum transmission is high, when observed at moderate resolution an individual H spectral line coincident in wavelength with a narrow atmospheric absorption line could be attenuated severely, resulting in an underestimate of the H line flux after calibration by the standard. Likewise, the strength of an unresolved line may be overestimated if the continuum transmission is low. To reduce such occurrences we modeled the atmospheric transmission across the pass bands as observed at Mauna Kea using a model developed by Grossman (1989). The transmission at specific Doppler-shifted wavelengths of H line emission from NGC 2023 (which was redshifted by 19km and 42km on Nov 28-Dec 1 and Jan 19, respectively, using a velocity of +10km with respect to the local standard of rest for the source) was determined using the model and compared to the mean transmission over the spectral resolution of the observation. The reliability of the flux calibration of a particular H line could then be estimated from its proximity to a telluric feature. In practice, uncertainties in the widths of the telluric features made the correction of the line flux of an affected H line very uncertain, so line fluxes were only determined if the attenuation was less than 10%.

Over 100 lines were observed and identified. However, of these nearly half were too severely blended to allow a reliable determination of individual amplitudes. Of the remainder, thirteen were at wavelengths where atmospheric absorption features made the calibration unreliable. Thus roughly one third of the lines identified could have their specific intensities accurately determined. These are listed in Table 2, converted to erg sr. Also listed is their level column density, , calculated by
 
where I is the specific intensity of a transition from level (v,J), with radiative decay rate A (taken from Turner, Kirby-Docken & Dalgarno, 1977) emitting a photon of wavelength . Note that no attempt has been made to correct for extinction, though this is likely to be A magnitudes (McCartney, 1998).

CALIBRATION ERROR IN 8000Å DATA

We note here, for completeness, that there is a calibration error in the 8000Å H data presented by Burton et al. (1992) for the (-11'', -78'') position in NGC 2023. The intensities quoted in that paper should be divided by 23 to yield surface brightnesses in erg arcsec.

© Copyright Astronomical Society of Australia 1997