Astronomical Techniques - Observatories and Sites

Telescope Enclosures

Telescopes must generally be enclosed for protection against the elements when not in use, to block the wind during observation, and to help keep them free of dust and other contaminants. The traditional form of enclosure has been a hemispherical dome, with a rotating top in which a vertical slit may be opened and positioned in front of the telescope. A wind screen is sometimes used to keep wind through the slit from buffeting the telescope. For some purposes, as when rapid movement is required, the enclosure may be a building whose roof slides completely off during observations, with a wall or temporary blockage to control wind.

It is important to make sure that the enclosure design does not introduce any additional turbulence and attendant degradation of the image (so-called dome seeing). This may result from non-laminar airflow around the structure or at the slit, and more insidiously from thermal effects - a large mass of metal takes a long time to come to temperature equilibrium with its surroundings (important for the telescope structure as well). One of the important lessons from the minimalist enclosure of the Multiple Mirror Telescope has been how much dome seeing was produced by large domes, even the ones designed to rise above ground effects. Current designs minimize the mass and flow intrusion of the enclosure, using wind tunnels to test for smooth airflow and in many cases allowing wind flow through louvered panels. It can be helpful to use passive and active measures to control daytime heating - reflective or white external paint, restricting airflow at certain times and speeding it with fans at others, even refrigerating the dome interior. Many installations have found it helpful to introduce large openings into the enclosure to allow unrestricted airflow around the telescope, which helps flush warmer air from the vicinity. For large telescopes, a mirror 1 º C warmer than the surrounding air will degrade the seeing to 1 arcsecond, though being slightly cooler is fine, so active temperature control systems are catching on.


Some atmospheric hindrances can be minimized by locating telescopes in appropriate places. An obvious one is cloud cover - all other things being equal, a telescope is more valuable if it has more clear nights. The usual suspects among observing locations appear in long-term mean satellite cloud maps. All other things, however, are seldom equal. There may be seasonal patterns related to climate, such that (for example) nights at one time of year are more likely to be useful than another. Remember that at temperate latitudes winter nights are nearly twice as long as midsummer nights.

Climatic patterns dominate some regions. The otherwise excellent climate of coastal California has pronounced wet and dry seasons, such that from April to October clear weather is virtually assured, while the wet season may have long periods of unbroken cloud and precipitation. This was a factor in locating the national observatory at Kitt Peak, Arizona - southern Arizona has its own pronounced climate pattern out of phase with California's, so that at least one set of large telescopes would likely have good weather at a given time. In this case, the driving force is monsoon circulation, where moisture from the Gulf of Mexico is pulled across the Sierra Madre and northward into Arizona, generating widespread orogenic thunderstorms.

Other desiderata for observatory sites are (1) seeing, (2) altitude, (3) lack of light pollution, (4) local transparency, and (5) accessibility. The line in the text about sites appearing at their best during testing is really not a joke. If many sites of comparable quality are investigated, the best one will be having a 3 σ good year.

  • (1) Seeing: this requires at minimum a lack of extra turbulence at all atmospheric levels, and seems to be best satisfied in the convergence zones just outside the tropics, at latitudes about ± 30 º. Also, minimal local turbulence is often associated with mountain peaks that reach into the otherwise undisturbed oceanic airflow, as on islands or coastal ranges (and given the direction of the planet's rotation, this generally favors western coast ranges). It has often been thought that the tip of a peak that is roughly triangular as seen from the prevailing wind direction has smoother airflow, but it appears from recent experience that the construction of the dome has a stronger influence. A small amount of ground cover may help stabilize flow near the ground (as opposed to pure desert or trees).
  • (2) Altitude. Beyond its effects on seeing (as in rising above some of the turbulent layers), altitude has advantages in increasing the atmospheric transparency. Atmospheric absorption varies widely with frequency, with useful transmission confined to so-called atmospheric windows. The best-known are in the optical and centimeter regimes, but there are several partially transparent ranges in the infrared; these ranges largely define the standard IR passbands (JHKLMN). These are bounded mostly by H2O bands, so the more water vapor below the site, the better. Thus there is strong emphasis on sites such as Mauna Kea (at nearly 4000 meters) for IR work; it is simply not possible close to sea level. There is a minor gain at the other end of the spectrum, where aerosol scattering and ozone absorption compete around 3200 Å. However, the ozone layer is so high that its importance is reduced only at 30 km and higher. The continuous absorption is given in the table from section 56 of Astrophysical Quantities (3rd edition, revised beyond recognizability in the 4th):

    Wavelength (μm) τ (molecular scattering) τ (ozone) τ (dust) Net transmission
    0.20 7.36 2.4 0.24 0.00
    0.22 4.76 17 0.21 0.00
    0.24 3.21 65 0.19 0.00
    0.26 2.25 88 0.17 0.00
    0.28 1.63 34 0.157 0.00
    0.30 1.21 3.2 0.143 0.011
    0.32 0.92 0.24 0.132 0.27
    0.34 0.71 0.02 0.122 0.43
    0.36 0.56 0.0 0.113 0.51
    0.38 0.45 0.0 0.106 0.58
    0.40 0.36 0.0 0.099 0.63
    0.45 0.22 0.001 0.084 0.73
    0.50 0.14 0.012 0.074 0.79
    0.55 0.098 0.031 0.065 0.82
    0.60 0.068 0.0044 0.058 0.84
    0.65 0.0495 0.023 0.053 0.88
    0.70 0.0366 0.008 0.048 0.91
    0.80 0.0215 0.001 0.035 0.94
    0.90 0.0133 0.0 0.035 0.95
    1.0 0.0087 0.0 0.030 0.96
    1.2 0.0042 0.0 0.024 0.97
    1.4 0.0022 0.0 0.019 0.98
    1.6 0.0013 0.0 0.016 0.98
    1.8 0.0008 0.0 0.014 0.99
    2.0 0.0005 0.0 0.012 0.99

    In the IR, more importance attaches to molecular absorption bands, largely water vapor with an occasional supporting role by CO2 (that's why these are the major greenhouse gases). This starts with strong narrow bands at 6867 and 7609 Å ; the so-called B and A bands respectively. Farther to the red, it is easier to speak of transmission windows than absorption regions: the best-known are centered near 1.22, 1.63, 2.19, 3.45, and 4.75 μ m, with more difficult work still possible at 10.4 and around 20 μ m. In the so-called thermal infrared (5 μ m and longer) the atmosphere's thermal emission (as opposed to line emission from airglow) is the dominant background source, so there is some advantage to working where the atmosphere is colder. In fact, higher sensitivity is often possible on cold winter nights even at the K band (2.2 μ m). This explains the interest in IR astronomy from the South Pole, which otherwise might be easier than Earth orbit - or then again, maybe not. The Gemini Observatory staff has put a set of transmission plots up for planning and instruction, showing the IR transmission regions from 1-28 microns.

  • (3) Dark skies. Unwanted light, natural or artificial, degrades observations of faint objects, sometimes pushing them beyond the realm of the practical. Natural sources (airglow) can only be broadly minimized by staying outside the major airglow bands driven by the geomagnetic field (which coincide uncomfortably with the convergence zones of atmospheric circulation). Natural sources can be cities, mining operations, etc. Thus good sites will generally be far from cities, and nearby light sources may be subject to legal controls if the observatory exercises sufficient clout in the local economy. Garstang has done numerous models of light pollution from cities of various sizes. Light pollution from most cities is dominated by scattered light from emisison-line sources, though high-pressure sodium lamps contaminate a broad spectral region. There have been pitched political battles in California and Arizona over these issues. Detailed maps from satellite data are available from and the DMSP satellite composite
  • (4) Transparency. It doesn't help to have a great site if you are so close to a desert that the air is full of blowing dust, or a nearby copper mine pollutes the air to high altitudes. La Palma sometimes sees a cover of windblown Saharan dust, and otherwise promising Chilean sites are ruled out by mining activities.
  • (5) Accessibility. Again, all other things being equal, the closer a site is to civilization the better support will be available. This need is less important now that high-speed data links are carried via satellite relay, so that astronomers need not always go to the observatory.

    With all of these requirements, it may not be surprising that there are rather few excellent sites available. We need (1) a large fraction of clear nights, (2) a mountain top, 2000-4000 meters tall, not too far from a west coast, (3) ideal latitude ± 30 º or so for best chances of good seeing and reasonable sky coverage, (4) far from large cities, and (5) no compromising sources of light or scattering (meaning a semi-arid rather than arid region). A first screening of sites may use long-term satellite cloud maps, rainfall and river records, vegetation, even light `aircraft flyovers to assess turbulence. There is no substitute for on-site testing, since seeing depends on many poorly-understood variables. Seeing can be measured by actual observations with a small telescope (aperture 40 cm or less), or in a more controlled way by measuring the image motion in star trails or using a two-mirror interferometer. Acoustic sounding and LIDAR allow a more direct measurement of atmospheric turbulence, which has recently been linked to the astronomical measurements. All these give the site seeing, which one must strive not to degrade with additional effects from the dome or telescope.

    Known good areas include coastal California (fighting light pollution), southern Arizona and Baja California, the coastal Andes of central and northern Chile, the Canary Islands, and of course Mauna Kea. There have been claims of certain excellent but remote sites in Central Asia. "Dome C" in Antarctica is probably the best site on the planet for mid- and far-infrared astronomy being at a high enough elevation to avoid very nasty ground layers found at the pole itself). Australia has what low mountains there are near the wrong coast, and testing in southern Africa never progressed very far, though the mountains do not compare with the Andes. Very high peaks are usually ruled out as they create their own local weather, not only bad enough to rule out observation but bad enough to destroy the equipment. Some otherwise promising peaks in Nevada are unfortunately within line-of-sight viewing of downtown Las Vegas (even the one called Telescope Peak).

    Some instruments impose local topographic requirements. Interferometers (radio and optical) may need a large flat region, and even optical interferometers may overfill some existing sites. Millimeter-range instruments require a high site because of atmospheric absorption, while in the centimeter regime lack of interference and wind loading are more crucial. Perfectly acceptable CO maps at a few millimeters have been done from downtown New York.


    Some kinds of observations (repeated brightness monitoring of variable stars, for example) require very long series of observations or an utterly routine nature. There are now robotic telescopes capable of doing such measurements to high precision. A typical operating sequence might be as follows:

  • (1) Is it raining? Interrogate a precipitation gauge.
  • (2) If not, is it cloudy? Interrogate a near-IR sensor comparing sky and ground brightness.
  • (3) If not, open the roof and find a bright star.
  • (4) Now find each program and comparison star, repeating sets of measures with recentering using a photomultiplier or CCD.
  • (5) At the end of the night, have your PC send the data to you. Observations may be perused with the morning paper.

    A set of automated telescopes of 20-75 cm aperture has been in use on Mt. Hopkins for some years. Strides have been made toward fully automatic imaging and spectroscopic instruments, perhaps useful test beds for lunar or antarctic instruments.

    « Telescope mountings | Spaceborne astronomy »

    Course home page | Bill Keel's Home Page | Image Usage and Copyright Info | UA Astronomy
    Last changes: 2/2011	  © 2000-2011