5 Conclusions

The GSM mission to Paranal was highly successful and for the first time provided statistical data on the outer scale at this site. Our main results consist in the following:

• The distribution of the outer scale ${\cal {L}}_0$at Paranal is close to log-normal, with a median value of 22.0 m. It means that because ${\cal {L}}_0$is comparable to the telescope aperture size the effects of the finite outer scale on the operation of VLT are significant and result in reduction of image motion;

• The correlation of ${\cal {L}}_0$with local meteorological conditions was noted. For typical conditions of NE wind, stable nightly median values of ${\cal {L}}_0$around 16 m are found, whereas for the southern wind (associated with the wind shear in the boundary layer) they are over 45 m;

• Excellent agreement of the seeing measured by the GSM and by the ESO DIMM is found. Median seeing for our campaign was 0.88''for GSM and 0.87'' for DIMM;

• The median value of the isoplanatic angle $\theta _0$ was 1.90''during our mission. This appears as an intermediate value in comparison with our estimations during previous measurement campaigns at other observatories (1.26'' at La Silla, 2.48''at Maidanak and 2.73'' at Cerro Pachon). Thus, the high-altitude turbulence over Paranal appears to be not substantially different from other observatories;

• The strong local turbulence in the first few meters of the surface layer exists at Paranal for some periods of time, as evidenced by DIMM-GSM comparison and micro-thermal mast measurements;

  Turbulent energy in the SL contributes an average amount of only 12.5% to the total turbulent energy detected by GSM along the line of sight. This corresponds to an average seeing degradation around 8% if turbulence effects in the SL are added to the turbulence above 21 m. While unit telescopes of the VLT are not affected by these layers, the situation will not be the same for auxiliary telescopes (ATs) of the VLTI which have been designed closer to the ground (4.5-6.5 m). Hence, continuous micro-thermal monitoring at Paranal appears to be desirable to study the local effects in detail and to evaluate their importance for the site seeing and particularly for the ATs;

• The temporal evolution of atmospheric parameters at Paranal seems to be rather specific (prolonged periods of stable conditions; dependence on the wind speed and its fluctuations; repeatable increases of surface layer turbulence) compared to other sites, and warrants further investigation. A better understanding of the mechanisms of Paranal seeing will eventually lead to forecast observing conditions.

The analysis of the spatio-temporal cross-correlations of the AA fluctuations can lead to the detection of the atmospheric layers (Avila et al. [1997]). From the AA cross-correlations one can estimate the number of predominant turbulent layers, their respective strengths, the velocity and direction of the wind which carries the turbulent eddies. The principle of the wind measurement with GSM is given in Conan et al. ([1998]). The full wind analysis and cross-correlation interpretation must include simultaneous comparisons with other parameters, ${\cal {L}}_0$, $\varepsilon$ and $\theta _0$. The methods of such complete analysis are still under development.

Acknowledgements

The success of this observing campaign is a result of a long and tedious preparatory work done at the Département d'Astrophysique of the Nice University. We acknowledge the efforts of A. Robini, J.-F. Manigault and M. Azouit in the development of GSM and of the mast equipment, and the help of the staff of the "Observatoire de la Côte d'Azur''. On-site preparation of GSM campaign at Paranal was organized and implemented by ESO under the supervision of M. Sarazin. We are indebted to Francisco Gomez for helping with observations during the second half of the mission.

Financial support came essentially from INSU, CNRS and MENRT for development of GSM experiment, and from ESO for this measurement campaign.