Dome seeing, defined here as excluding the mirror contribution
which is treated
separately, is due to free or weakly mixed convection inside the
entire volume of the telescope enclosure.
The applicable similarity relationships (equation
(
)) show
that the occurrence of dome seeing requires significant surface heat
fluxes, of the order of 10 W/m
for 
arcsec.
In a conventional dome where no ventilation occurs or is desired, the best technical solution to prevent dome seeing is the provision of a chilled floor. This will create a stable thermal inversion which will also "damp" any free convection air motions that may arise from localized heat sources. As a "second best" solution, probably less expensive, one can provide an internal insulation of the floor and of the walls, which will reduce the heat exchange between the building structure, with its large thermal inertia, and the air volume, the temperature of which varies more rapidly.
In a ventilated enclosure such as the NTT, dome seeing is eliminated by the wind flushing through the telescope. Therefore no particular thermal control of the enclosure interior during observations needs to be planned. Suitable operating procedures should be established for the case when the enclosure is operated at a more shielded configuration (windshield up, louvers closed, side walls facing the wind). Then an occasional occurrence of dome seeing can be corrected by the operator "flushing" the enclosure during an interruption of the observations.
A particular issue is set by heat generation at the secondary mirror
unit. Seeing from this source was quite feared during the
development of most recent telescopes and therefore heat extraction
circuits were implemented. These circuits can be quite complex,
also because they must reach the M2 unit through the spiders.
In fact we
have shown that the seeing effect (see equation ())
is smaller than it was feared, even in low wind conditions.
If the top ring of the telescope is reasonably exposed to the wind, the
designer should consider trading-off the complications of a heat
extracting circuit against a minor and occasional loss of
image quality.