4. Discussion and conclusion
From Sect. 3 it is clear that all three types of models can explain the observed asymmetry to some extent. However, the first two models require unrealistic amounts of HI above the plane. The third model has a much more realistic vertical distribution, with less than 20% of the gas in the thick component, and it requires that the gas above the plane rotates more slowly than the gas in the disk, by about 25 km . This model also appears to reproduce the observed position-velocity diagram better than the other models, in particular near the center, where the HI has a narrow and peaked distribution, and at large radii, where low level wings are visible.
Note that in the observed position-velocity diagram (Fig. 1) the wings near the center extend almost to the systemic velocity. Clearly, such features cannot be reproduced in models with corotation, and not even in those with a velocity decrease as high as 35 km . To explain these extended wings a much larger velocity decrease is needed, probably at least 50 km .
What produces such a thick HI layer and what causes its slower rotation? Galactic fountain models (Bregman 1980; Spitzer 1990) may provide the framework for an answer. The fountain is formed by hot gas rising from the disk, its energy derived from stellar winds and supernova explosions. In the halo region the gas cools and condenses into clouds that fall back onto the disk. As the gas moves up, the centrally directed gravitational force decreases, and the gas moves outwards. Due to conservation of angular momentum the azimuthal velocity decreases. This decrease will be most pronounced in the central parts of the galaxy, where a fixed radial displacement will result in a larger velocity decrease than in the outer regions. Using thermohydrodynamic models, Struck & Smith (1999) have recently shown that a reduced circular velocity above and below the plane is expected to be present in turbulent disks as a result of radial motions driven by star formation activity.
NGC 2403 appears to have sufficient star formation activity to drive a galactic fountain. This is for instance indicated by its large number of HII regions (Sivan et al. 1990). Four of these are exceptionally bright, comparable to the most massive starburst region in the Local Group, the 30 Doradus complex (Drissen et al. 1999). Furthermore, Thilker et al. (1998) have found that the surface of NGC 2403 is covered by shells and fragmentary structures which are likely to have formed as a result of star formation activity. They have also found that these structures are part of a diffuse component of neutral hydrogen extending at least 400 pc from the plane.
The overall picture of NGC 2403, as suggested by the modelling, is reminiscent of that of NGC 891. For this galaxy it was found (Swaters et al. 1997) that the thick component has a FWHM of about 4 kpc, and this gas appears to rotate more slowly than the gas in the disk by about 25 km . In the central parts the velocity decrease was found to be much larger, perhaps up to 100 km .
In conclusion, we have presented evidence that the thin hydrogen disk of NGC 2403 is surrounded by a vertically extended layer of HI, which rotates more slowly than the disk. The observational picture is very similar to that found for NGC 891. These results suggest that a vertically extended, slowly rotating HI layer may be common among spiral galaxies, at least among those with high levels of star formation. Such extended HI layers have, however, very low surface densities and therefore very sensitive observations are needed to detect them.
© European Southern Observatory (ESO) 2000
Online publication: April 10, 2000