Conclusions

We have described in this dissertation the main effects of the local atmospheric environment on the performance of an astronomical telescope and presented experimental databases, theoretical analysis and parameterizations which should contribute to establish the engineering of telescope enclosures on a more solid knowledge base.

Three main topics were studied:

• The characterization of the wind turbulence responsible for high frequency guiding errors
• The turbulent pressure fluctuations on the primary mirror
• The local seeing effects caused by the mirror, inside the enclosure and in the atmospheric surface layer

• The sharp edges of the enclosure slit are the main cause of the turbulence on the upper part of the telescope and of the induced guiding errors. The length scale of this turbulence is related to the slit width with a factor varying generally between 0.2 and 1, which brings the largest part of the turbulent energy in the frequency range between 1 and 10 Hz. In conventional oversized domes the large distance of the top ring from the slit edge and the absence of through-flow prevent effects on the telescope. On the more recent enclosures designed for through-flow to eliminate dome seeing and tightly fitted on the telescope, the only practical possibility of corrective action is given by windscreens with a permeability of 20%.

• The surface figure of the relatively thin primary mirrors of the new 8-m generation are quite sensitive to the turbulent pressure fluctuations caused by wind buffeting. The turbulent pressure field on the mirror surface has been measured and analyzed with respect to its effects on dynamic deformations and consequent optical aberrations. It was found that, over a wide range of enclosure arrangements, the spatial correlation of the pressure field is mainly determined by the mirror disk shape and that the resulting optical aberration is predominantly astigmatic.

  As a consequence a simple parameterisation can be found between the average rms of pressure variations on the mirror surface and the total optical aberration. Simple relationships can also be found between local speed values and pressure variations on the mirror, but they will depend on the enclosure arrangement. In the particular case of the VLT cylindrical enclosure, a parameterisation independent of the azimuth angle is found between the average flow speed on the mirror surface and the rms of pressure variations. These expressions are particularly useful for a parametric analysis of the overall wind+seeing effects of the primary mirror.

• The various seeing effects generated in the local atmospheric environment surrounding the telescope and its enclosure have been described and quantified with the help of experimental measurements and similarity analysis. We have analyzed in particular the seeing caused by free convection flow inside the enclosure, the mirror seeing, the effect of heat dissipation at the secondary mirror, and the influence of the atmospheric surface layer outside the telescope enclosure.

• Mirror seeing is the predominant component of what was previously called dome seeing. The mechanism of the seeing production on the mirror was elucidated. The seeing effect is generated in a thin region just above the viscous-conductive layer where the temperature fluctuations are largest and most intermittent. If its cause could be visualized, seeing would appear to come from a thin but very turbulent layer "floating" a few millimeters above the surface.

  The hypothesis that mirror seeing depends essentially in the surface heat transfer leads to extend to this case the similarity expressions used for the atmospheric surface layer. The agreement of computations of seeing by means of this model with the experimental data indicate that this model does reflect the physics of the phenomenon. The analysis of different experiments allows also to validate two simple parameterizations for mirror seeing with respect to the surface-air temperature difference, with and without external ventilation.

• The similarity criteria for a wind tunnel simulation of the seeing effects of the atmospheric surface layer have been determined. The positive outcome of some pilot tests in the LASEN wind tunnels open the possibility to apply reduced scale tests to the site and layout studies for astronomical observatories.

However, the new knowledge elements brought here, as well as the ones that may come from a deeper insight of the various aspects of the telescope/environment interaction, will not by themselves make the design process of astronomical observatory more straightforward. Rather, they quantify more accurately a set of requirements that remain contradictory (e.g. the wind-versus-seeing dilemma mentioned in the introduction and through chapter ). As these knowledge gaps are overcome, the engineering of telescope enclosures shall also tend to a better structured process of concurrent engineering, in which there is a large scope for the system analysis and the optimization of the overall performance of the telescope+enclosure combination.

In absolute terms, the optimal enclosure without local seeing can only be made for an outstanding telescope, possibly still better than allowed by the present state of the art, at least when large 8-m telescopes are concerned. Essentially this better telescope should have a primary mirror designed for a turbulent wind flow of 2-3 m/s and a guiding accuracy better than 0.05 arcsec with the top ring in open air. Then the best enclosure will be represented by the retractable dome illustrated in fig. .

Yet, even within the limits of the present state of the art, there is another approach to an optimized observatory, which consists in including in the engineering phase also all the operational aspects of astronomical observations. We have mentioned in chapter the random nature of the influence of local atmospheric turbulence and how these effects interact with the variability of the natural seeing and with a apriori also random distribution of telescope orientations. We have used these considerations to propose a statistical approach for the performance assessment. In fact one may envisage a few steps further in that direction.

Not all types of observations require the best image quality but some can only be made if this is outstanding. This consideration has already prompted the idea of a flexible scheduling of observations as a function of the natural seeing conditions. We can envisage to extend the scope of flexible scheduling to take into account also all the local parameters contributing to the image quality, such as the actual wind loading on the telescope observing any given sky object and mirror seeing predicted by processing the information on the respective temperature evolutions for mirror and ambient air. When all the environment dependent error sources are reliable parameterized, the overall image quality can be predicted for any option in the observation program, which can then be scheduled appropriately for optimized observations.

For example, recalling from chapter that the mechanical turbulence on the telescope will be stronger when the slit azimuth is between 15 and 50 from the average wind direction, there will be only two ranges in azimuth, for a total of 70 on 360 which will generally be critical for guiding errors. Therefore it is certainly possible to avoid the critical range simply by choosing an appropriate observation timing. Similar reasoning can be applied to all other effects and, with the help of the parameterizations provided by the present work, lead to develop the criteria for an optimized scheduling of the observations concurrently to the general design of an astronomical observatory.