6 Conclusions

Our simulations confirm that conventional mean-field dynamo models can reproduce magnetic field structures that are similar to those observed in barred galaxies. Even though we have used a velocity field model for a generic barred galaxy (Athanassoula 1992), our magnetic configurations show an overall agreement with observations of the prototypical barred galaxy NGC 1097. Our models are most successful when they include enhanced turbulence in regions of strong shear, most notably near the shock fronts offset from the bar major axis. The best overall agreement with observations seems to be provided by models with $\widetilde\alpha \propto \omega$ and $f_\eta\approx3$, although the overall appearance of the field in the outer regions is not very sensitive to the choice of parameters.

We argue that regular magnetic fields can be dynamically important in large regions within the corotation radius. The energy density of the magnetic field in our models and, plausibly, in real barred galaxies can exceed that of the interstellar turbulence and is largely controlled by the local shear in the regular (noncircular) velocity. This is in a striking contrast to the situation in normal spiral galaxies where regular magnetic fields and turbulence are close to energy equipartition.

Our models can successfully reproduce some salient features of the observed field structures, notably the unexpectedly strong field upstream of the shock, the gentle bending of the field across the shock, and the overall structure of the field near the galactic centre. There are well developed trailing spiral arms present outside of the corotation radius, which begin approximately at the ends of the bar.

The intensity of interstellar turbulence must be enhanced in the dust lanes and circumnuclear region in order to obtain a radial contrast in the regular magnetic field similar to that observed in NGC 1097. We argue that this enhancement can arise naturally due to shear flow instabilities. The enhancement of turbulence may have important implications for star formation and gas dynamical models in barred galaxies. Enhancements in turbulent velocity implied by CO observations for the central parts of NGC 1097 (Gerin et al. 1988) and NGC 3504 (Kenney et al. 1993) are consistent with those used in our models.

The initially surprising bending of magnetic lines upstream of the shock can be explained by (i) advection of magnetic field from other regions, and (ii) by the alignment of $\vec B$ with the principal axis of the rate of strain tensor rather than the velocity itself. Note however that advection in an approximately uniform (non-sheared) flow does not cause alignment of $\vec u$ and $\vec B$, even in the perfect conductivity limit.

In conclusion, we emphasize three important findings of this paper.

1.

Barred galaxies are different magnetically to "normal'' spiral galaxies, in that the field strengths are significantly higher, and the alignment between magnetic and velocity vectors is, on average, closer.

2.

The magnetic field may be dynamically important: Alfvén speeds can be comparable with noncircular velocities (although generally a little lower). Such strong fields require relatively, but not unrealistically, strong dynamo action ( for our models with $\eta_0=10^{26}$ cm²s^-1).

3.

In order to model satisfactorily the magnetic fields it is necessary for the turbulence to be locally enhanced in the vicinity of the dust lanes and the circumnuclear ring, in a manner consistent with other evidence. This will affect other aspects of modelling of these galaxies.

Acknowledgements

We are indebted to E. Athanassoula for providing us with the velocity and density data from her model. We also thank D. Elstner, P. Englmaier and M. Urbanik for useful discussions and comments. We acknowledge support from PPARC (grant PPA/G/S/2000/00528), NATO (grant PST.CLG 974737) and the RFBR (grant 01-02-16158). The generous assistance of W. Dobler is gratefully acknowledged.