2.2 Complete Rotation Curves
A first requirement to discuss rotation curve shapes is to have them defined over as large a range in radius as possible. This means that HI rotation curves need usually to be supplemented with optical data in the inner parts, to assure a better definition of the curve there. This can be done by e.g. long slit spectra of optical emission lines - usually H and [NII] - or, preferably, with two-dimensional Fabry-Pérot data. Alternatively, CO observations in the millimeter wavelength range can be used, either from single dish data, or from interferometry. Sofue (1997) published a number of combined rotation curves using available data in HI, CO and H.
Begeman (1989) developed a method to correct HI rotation curves for the effects of beam smoothing, using high sensitivity data for the galaxy NGC 3198. He fits multiple gaussians to the line profiles, and retains the principal peak. His results have been compared with long slit optical emission line data by Bottema (1988) and Hunter et al. (1986), and with Fabry-Pérot data by Corradi et al. (1991). The agreement between all these data is excellent.
The necessity of good data in the inner parts cannot be overstressed, since in particular for the multi-component mass models, the outcome of the decomposition depends on a correct determination of the rotation curve at all radii. This is particularly important for dwarf galaxies, where the disc rotation curve is often compared with observational data consisting of only a couple of data points.