5. Conclusions

From our numerical simulations we draw the following important conclusions:

1. The presence of velocity fields affects the level populations via the increase of the downward radiation. Nevertheless, for velocities that do not exceed the thermal velocity of plasma, one can use the static populations for the formal solution of radiative transfer equation including the velocity to reproduce the observed line asymmetries.
2. The line asymmetry is caused by the changes of the optical depth scale for different line frequencies. The type of asymmetry depends on values of the source function at heights corresponding to .
3. A monotonous velocity field affects only one wing of the line profile, the red one for downward motions, the blue one for upward motions.
4. Application of the bisector method would lead in some cases to reverse velocity, in others to underestimation of the velocity. It includes, besides the part of the profile directly affected by the moving material, also a ``static" part of the profile. To use the bisector in terms of Doppler shift the static part should be somehow eliminated.

To deduce velocities from asymmetric line profiles using these conclusions, we need to guess how the static profile would look like. For this purpose one could use a symmetric profile from a close observing time or position, but then it seems to be necessary to perform NLTE calculations for each particular case.

The determination of the velocity height structure from a single line is inaccurate as an emergent line intensity is affected only in heights where , which cover only a small part of the atmosphere for a single line. We therefore need simultaneous spectral observations of more lines, preferably those that originate in different heights, to reconstruct the velocity structure in a flare atmosphere. The more lines will be included the more precise the determination of the height structure of the velocity will be.

Acknowledgements

The author is indebted to Dr. Petr Heinzel for helpful discussions and comments. This work was supported by the Grants No. 205/94/1577 and No. 1199 of the Grant Agency of Czech Republic and by the Grant No. K1-003-601 of the Academy of Science of Czech Republic.