Astron. Astrophys. 363, 289-294 (2000)

5. Discussion and conclusion

The height dependencies of the model correlations will change if we use another method to calculate mean correlation coefficients: we may collect data belonging to each horizontal level over a whole time series of our models and then calculate correlations between these data. But the height dependencies of the new correlations differ from those discussed in this paper not more than by 20-30 km and this does not change significantly our main conclusions.

From our simulated spectral observations we do not detect the sharp drop of correlation between spatial fluctuations of vertical velocities in the low and in the middle photosphere as it was reported by Karpinsky (1990). An explanation can be that our models are obviously more laminar than the real solar photosphere. However, this conclusion of Karpinsky was not confirmed later.

Our results with high correlation between horizontal velocities in the model photosphere are also in disagreement with a study of Nesis et al. (1988), where they found, based on observations, that horizontal velocities are coherent only in the low photosphere. We do not exclude that our model result could be influenced by the 2-D cartesian approach, close to laminar treatment of the medium, and spatially limited computational domain.

In spite of this the correlations derived from observations and from simulations of several Fe I and Fe II lines are in good agreement. Therefore, summarizing our results we may conclude that:

• a) almost in the whole photosphere above granules the matter predominantly ascends.

• b) clouds of cold and dense matter are located above bright granules beginning from heights about 150-170 km;

• c) above 250 km the columnar structure of the photosphere is broken down.

Schematically, the structure of the model photosphere can be presented as the following:

• a) low photosphere: it seems that the top of the thermal convection zone lays at 20-50 km above the visible surface;

• b) then we have convection overshoot into stable layers up to 150-170 km;

• c) between 170 and 250-300 km we detect a transition layer where the convective columnar structure still exists (due to the influence of convective pressure variations) and where the inversion of temperature fluctuations is largest;

• d) above 300 km this columnar structure is broken on the average and the photospheric medium is controlled mostly by oscillations.

This scheme agrees with the picture from spectral observations (Nesis et al. 1988, Karpinsky 1990).

Finally we point out that multidimensional selfconsistent model atmospheres can be successfully used to test possible criteria in estimation of "line formation depths". Although our results show that the scale seems to be better suited for diagnostic purpose than the line formation depths found with the depression contribution functions we suggest to test also physically more adequate methods using the conception of the response functions.

© European Southern Observatory (ESO) 2000

Online publication: December 5, 2000
helpdesk.link@springer.de