3. Observations and data reduction
The observations were taken on March 19, 1997, at the William Herschel Telescope using the TAURUS II Fabry-Perot interferometer at the Cassegrain focus, with the f/2 camera and the 500µ etalon. The detector used was a TEK CCD. With this configuration the pixel size is 0.56 arcsec. We rebinned 2x2 the CCD reading, resulting in a pixel size of 1.12 arcsec (which for an adopted distance to NGC 5668 of 22.6 Mpc results in 123 pc/pixel). The observing conditions were not photometric (therefore we were not able to make photometric calibrations and we will use arbitrary units for intensity throughout this paper). The measured seeing in the final images was 3 arcsec which is still a very good spatial resolution when compared with the previously mentioned kinematical data available for this galaxy. We used the line to trace the distribution and kinematics of the ionized gas. According to the observed velocity of the galaxy (see Table 1) the wavelength for the redshifted line is 6597.5 . Therefore we used the 6601/15 filter for order sorting. We scanned the free spectral range (FSR) in 55 steps with an exposure time of 120 seconds per frame. This gave a spectral resolution of 3.61 km/s/pixel. Calibration datacubes illuminating the instrument with a CuNe lamp were taken at the beginning and at the end of the night. Moreover a ring calibration frame was taken before and after the datacube exposure to make the RVT correction. The instrumental width (measured after phase correcting the calibration datacube) was 7.7 km/s. The main properties of the observational setup are summarized in Table 2.
Table 2. Observational parameters
After phase and wavelength calibration the datacube was bias subtracted. The continuum level was calculated by fitting a first order polynomial to the channels where there is no line emission from the galaxy, and the continuum map was then subtracted from the datacube. From the continuum free datacube we calculated the intensity, velocity and velocity dispersion maps. We calculated the maps by two different methods: a moments procedure and a gaussian fitting. The intensity and velocity maps are very similar in both cases and we use the ones calculated by the gaussian fitting procedure. The velocity dispersion maps are substantially different and we use the one calculated by the fitting procedure as well, in order to avoid systematic bias as pointed out by van der Kruit & Shostak (1982). The calculated maps are very noisy. To clean these maps we imposed certain conditions to flag out bad data points. We therefore kept only those pixels which had a velocity dispersion between 7.7 km/s (the instrumental width) and 50 km/s (data points above this level are clearly not valid). We also required the points to have a peak intensity at least 2 times larger than the noise dispersion. Finally some clearly bad data points were excluded interactively by inspection of the resulting maps. As we are interested in the non-thermal velocity dispersion, we corrected the calculated dispersion for the natural width of the line (3 km/s), the insrtumental width (7.7 km/s) and the thermal width (9.1 km/s assuming a temperature of K for the ionized gas). The corrected average velocity dispersion calculated for this galaxy is around 16.5 km/s independent of radius. This value is somewhat higher than that measured for other galaxies, which is usually around 10 km/s (see for example Jiménez-Vicente et al. 1999), showing that the ISM of NGC 5668 is more turbulent . The intensity map for NGC 5668 is shown in Fig. 2.
© European Southern Observatory (ESO) 2000
Online publication: June 20, 2000