M. Steffen: Radiative hydrodynamics (RHD) models of stellar convectionSteffen presented models of stellar surface convection which implement 3D hydrodynamics, thermodynamics (equation of state) and 3D radiative transfer (including realistic opacities). The code used to calculate the numerical solution is CO5BOLD, described in detail in its

on-line User Manual. Some important points are that no diffusion or Eddington approximations are made in the radiative transfer, the opacities are based on ATLAS or MARCS ODFs using the multi-group opproach (5 bins). The calculations are done in strict LTE.

A Box-in-a-star setup was used (cartesian simulation box with periodic side boundaries), which covers a tiny fraction of the solar convection zone. A model is characterized by Teff, logg, and abundances with no additional free parameters and is intended to be used for direct comparison with real stars.

An example for such a comparison was presented, using observations from the

Swedish 1-meter Solar Telescope - a 15"x15" field at a wavelength of 4364Å compared with a simulation using 400x400x165 cells (see Steffen 2004). Another example was a favorable comparison of simulations to

Dutch Open Telescope observations (Leenaarts & Wedemeyer 2005 A&A 431, 687).

A presentation of the flow topology of deeper layers in the numerical simulation (not accessible to observation) showed that the granulation pattern is a very shallow surface phenomenon.

Next came a strictly differential comparison of 3D RHD simulations with 1D Mixing Length Theory calculations for the Sun: The vertical velocity in 1D is too small throughout the simulated region, and the temperature structure cannot be matched in 1D with a single mixing length parameter.

The next part was on surface convection in A type stars, which possess two distinct convection zones (H/HeI + HeII). The simulations show that the buffer zone between convection zones is fully mixed and the upper boundary of photosphere is dynamic (see Freytag & Steffen 2004). The surface convection cells of A type stars are larger than in the Sun. The simulations can be used to calculate synthetic spectra from A type stars, which are presented in a poster by Oleg Kochukhov (No. 1 in Terrace I).

A list of applications was followed by a discussion of the solar oxygen abundance, which has a history of downward revisions. It was redetermined with a 3D CO5BOLD model atmosphere (completely independent from Asplund et al.) using 9 OI spectral lines. A fit to Neckel and Kurucz fluxes gives a value of 8.70+-0.05 (this is work in progress). The 3D effects are not more than -0.04 dex.

Conclusions: There are quantitative and qualitative differences between 1D and 3D atmospheres. The photospheric temperature is reduced in metal poor stars (metallicity of -2 dex, not mentioned in the summary above). The numerical simulations can be used to calibrate MLT for stellar evolution calculations. CO5BOLD 3D atmospheres can be used for detailed studies of spectral line formation in the presence of photospheric temperature inhomogeneities.

J. Trujillo Bueno: Radiative transfer modeling of the Hanle effect in convective atmospheresThis talk was about the Zeeman effect versus the Hanle effect.

Scattering experiments in the absence and in the presence of a magnetic field show that the

Hanle effect reduces the line scattering polarization amplitude. This can be used to measure small magnetic fields, whereas the

Zeeman effect is blind to small entangled magnetic fields. The Zeeman effect can only detect 1% of a resolution element.

For the Sun, voids in circularization patterns are detected by the Zeeman effect. Are the voids field-free?

As an example, modelling of the Sr I 4607 line profile and a fit to observations (Stenflo et al.) was presented. The Sr I line is a triplet, where the magnetic sublevels are unequally populated due to optical pumping by directional light. This is important if there is anisotropic illumination, which is the case at the stellar surface (limb darkening). To model this anisotropy accurately, it is essential to use 3D models. To fit the Sr I line, a mean magnetic field strength on the order of 100 Gauss is required to take into account the Hanle effect (no distinction between upflows and down flows).

Another example are the molecular lines of C2 (Swan system R1,R2,R3). Here, the Hanle effect requires a mean magnetic field strength on the order of 10 Gauss. Where does the discrepancy come from? The solution (presented in Trujillo Bueno 2003, 2004) is an anisotropy between downflows and upflows. The observed scattering polarization in (weak) molecular lines comes mainly from upflowing regions, whereas the scattering polarization strong Sr I lines comes mainly from downflowing regions.