©D. Andre Erasmus, 1987, 1997
FOREWORD: Even with the advent of space telescopes, astronomers recognize that ground-based observatories will continue to be used for many years. For this reason, modern telescope projects have placed high priority on site quality. To find good quality sites, astronomers have cultivated a willingness to examine locations as potential telescope sites, previously thought to be too remote or inaccessible. In this quest for the best observing sites, factors traditionally considered important, such as proximity to the parent institution, have become of secondary importance.
1. Introduction
Experience has taught astronomers that a fairly well defined set of atmospheric conditions make a particular site good for astronomical observations: Cloud cover should be a minimum. The frequency of occurrence of bad weather events such as cyclonic storms and thunderstorms should be as low as possible. The air above the observatory sites should be dry. The drop in air temperature during the night should be minimal. Atmospheric transparency should be high. The magnitude and persistence of high frequency temperature fluctuations associated with atmospheric turbulence (the effects usually designated by the term "seeing") should be low. .
It is generally the case that long term records of these parameters are not available for potential and even some existing observatory sites. However, there is a good probability that a comprehensive data set of meteorological parameters is available for nearby locations. These data could be used directly or indirectly to evaluate the observing quality at existing and potential telescope sites - directly, by analysis of temperature, humidity, wind, atmospheric pressure and cloud cover data, indirectly by using an understanding of the relationship between observing quality and meteorological conditions to interpret the data. Clearly, the use of archived meteorological data alone to evaluate site quality is not sufficient. There will always be a need for onsite measurements of site quality. But, available meteorological knowledge and data bases are not being employed to their full potential in the evaluation of telescope sites. Appropriate use of this resource could streamline and reduce the costs of telescope site selection. This article discusses the fundamental meteorological principles that relate to astronomical observing quality.
2. The Large-Scale Meteorological Situation
A noteworthy connection between atmospheric conditions and telescope site selection can be gleaned from an examination of the location of modern large telescopes (Fig.1). With few exceptions, these telescopes are located in the latitude belt 15o-35o on the western side of continents or in the eastern oceans. This observation is significant from a meteorological point of view since these areas have similar climatological characteristics.
These areas, which are located within the subtropical high pressure belt, are associated with semi-permanent anticyclones or cellular-shaped high pressure systems. These high pressure systems develop as a consequence of a thermally driven circulation between equatorial and subtropical latitudes, called the Hadley circulation (Fig.2), and the effects of the earth's rotation. The vertical component of the Hadley circulation consists of ascending motion near the equator and descending motion in subtropical latitudes. The broadscale subsidence in the high pressure cells produces adiabatic compression and, therefore, warming, drying and stabilization of the atmosphere in these areas. Disturbed weather and cloud formation is, therefore, discouraged. So the weather conditions that prevail in these areas are generally favorable for astronomical observations. In contrast, equatorial, temperate and polar latitudes are characterized by unstable air masses and disturbed weather, which factors produce generally unfavorable observing conditions.
On a global scale, therefore, telescope sites now known to be superior are located within the region that would be identified as the preferred region for siting telescopes from meteorological and climatological considerations. Within this generally favorable region, however, significant differences in observing quality occur. A number of meteorological factors are responsible for these differences. These will now be discussed.
3. Cloud Cover and Atmospheric Water Vapor
High levels of humidity and cloud formation are generally associated with disturbed weather events such as cyclones and thunderstorms. Disturbed weather does occur occasionally in the sub-tropical high pressure regions but the weather conditions in the area are dominated by anticyclones which are typically fair weather systems.
Within these high pressure systems the distribution of cloud cover varies according to the vertical structure of the atmosphere within these systems. Subsidence in the upper and middle troposphere (the layer of the atmosphere between the surface and an altitude of about 12km in which nearly 90% of the mass of the atmosphere is found), stabilizes the air at these levels while surface heating and mechanical mixing destabilizes the air near the ground. This produces a well mixed layer with the likelihood of cloud formation near the ground (Fig.3). At the boundary between the mixed layer and the stable subsiding air, a temperature inversion exists. The temperature inversion itself is a stable layer that effectively suppresses vertical motion through it.
The air above the inversion, therefore, tends to be dry, stable and cloud free while the layer below the inversion is moist, unstable and cloudy. The importance of locating telescopes above the level of the inversion (typically 500 m - 1500 m above the ground) is therefore evident. The height and strength of the inversion, and hence the relative importance of subsidence versus mixing varies diurnally, seasonally and geographically. The inversion undergoes a daily cycle as a consequence of surface heating during the day and cooling at night. At night radiative cooling of the surface inhibits mixing near the ground, lowers the inversion level and enhances stability. These are ideal conditions for astronomical observations. During the day, surface heating promotes mixing near the ground, lifting the inversion and weakening it to some extent. If surface heating is not too strong, then the higher mountain peaks will remain above the inversion even in the daytime, allowing for good daytime observing conditions also. Seasonal variations are dependent on latitude. At the equatorward side of the subtropical high pressure areas, the inversion lifts and weakens in summer and drops and strengthens in winter, because these latitudes are more strongly influenced by equatorial disturbances which dominate in the summer months. In more temperate latitudes within these areas of high pressure, the inversion lifts and weakens in winter and drops and strengthens in summer, because these latitudes are more strongly influenced by temperate disturbances which dominate in winter.
The height and strength of the inversion also varies meridionally, depending on the relative importance of oceanic versus continental effects. Over the continents, the daily cycle is more pronounced than over the oceans since the ground heats up and cools down more than water, and the atmosphere's response is therefore more extreme. In the summer, surface heating can be so intense during the day that convective processes overcome the effects of subsidence. The mixed layer becomes very deep, the inversion is elevated, becomes very weak, or may even disappear. Under these conditions, dust and other particulates are carried above observatory sites. If a source of low level moisture is present under these circumstances, vertical mixing is invigorated by the latent heat released when this moisture is carried upwards in convection currents and condenses. This leads to the development of thunderstorms which inject vast quantities of moisture into the upper troposphere producing the anvil-shaped cloud typical of these storms. As these storms dissipate, the high-level clouds spread horizontally, forming a thick deck of cirrus that can linger well into the early morning hours. Thus, both daytime and nighttime observing conditions are adversely affected. This is consistent with the observed increase in thunderstorm frequency and cirrus cloud cover near the eastern periphery of the subtropical high pressure systems that extend over the continents (Fig. 4). For example, summer thunderstorms are a well known phenomenon in Texas, New Mexico and eastern Arizona where some telescopes routinely close during the peak of the thunderstorm season.
Over the oceans, because a large fraction of the incident solar energy is used for evaporation and water's high heat capacity, the diurnal cycle of surface heating and cooling is moderated. The consequence is that over the oceans the depth of the mixed layer and height and strength of the inversion do not undergo large diurnal and seasonal fluctuations. The inversion height varies from about 500 m to 2000 m depending on the ocean temperature (the colder the sea, the lower the inversion) and, of course, latitude. At a particular location the inversion height varies about 500 m from day to night. Ocean sites, therefore, have the distinctive advantage of a semi-permanent inversion layer that undergoes small seasonal and diurnal variations.
Another factor that causes the inversion to break down and promotes cloud formation in the
subtropical high pressure region is the passage of cyclones that move from west to east in middle
latitudes. The frequency of occurrence of cyclone events shows considerable variability both
latitudinally and meridionally. Reitan (1974) determined the distribution of cyclonic events for
North America and adjacent oceans. The area investigated was divided into squares measuring
740 km on a side and the mean monthly frequencies of cyclonic events for the 20-year period
1951-1970 were determined. His results are shown in Fig.5 a-d for the mid-season months of
January, April, July and October. The maps show that, within the subtropical region, there is an
increase in cyclonic activity over the continental United States especially in the lee of the
Rockies which is a preferred region of cyclogenesis. The factors influencing observing quality
are therefore not only latitude dependent but also longitude dependent. At a particular longitude
the relative frequency of equatorial and temperate disturbances and their influence on inversion
height and strength make the selection of the ideal latitude critical. On the other hand, for a
given latitude, observing quality varies meridionally because variations in the quantity and type
of cloud cover that result from relative importance of oceanic and continental effects.
4. Atmospheric Transparency
While several factors determine atmospheric transparency, spatial and temporal variations in this
quantity are related almost entirely to differences in the concentration of suspended particulates.
Since observatory sites are usually in dry isolated areas, dust is the most common particulate
affecting transparency. The quantity of dust in the air is related to the altitude of the observing
site and its location in relation to a dust source (eg. deserts) and prevailing wind currents. Even if
an observatory site lies well above the inversion layer, dust can still have an important effect on
atmospheric transparency. This is so because particulates that are carried into the middle and
upper troposphere by convection currents and thunderstorms become trapped in stable layers and
are transported horizontally by the upper-level winds.
For example, in the Canary Islands when easterly winds occur above the inversion layer, Saharan dust is carried over the islands from the desert interior of North Africa, degrading atmospheric transparency. Fortunately, in this region westerlies prevail above the inversion level. Most of the time, therefore, the air reaching the observatory sites has traversed thousands of kilometers of featureless ocean and will accordingly be free of suspended particles. Isolated oceanic sites therefore have a decided advantage over continental sites in regard to atmospheric transparency provided that the sites lie above the inversion layer. The Hawaiian Islands are probably the best example of an isolated oceanic site since they are not only 2500 miles from the nearest landmass, but also have mountain peaks that extend several hundreds of meters above the subsidence inversion.
5. Temperature Change at Night
The nocturnal temperature change is an important indicator of site quality. Telescope performance is closely related to the temperature difference between the ambient air and air inside the telescope chamber. Most telescope components have high thermal inertia so that they and the air directly around them usually cool at a rate slower than the ambient air. Detrimental effects of this thermal imbalance are minimized where the ambient temperature change during the night is small. Cooling of the air directly above the ground at night is a consequence largely of being in contact with the radiatively cooling surface. The colder the ground and the longer the contact with the ground, the greater the actual cooling of the ambient air during the night will be. To the extent possible, therefore, the air passing over an observatory site should be that from the free atmosphere which is not affected significantly by ground cooling.
The probability that the air stream will be in contact with the cooling ground at night increases with distance inland. Even if a telescope is located on a mountain site that protrudes above the inversion layer, the temperature change at night will still be larger at an inland site than at a similar coastal or island site. This is the case since some fraction of the land area upwind of the site will also lie above the inversion level. Especially in winter, when the inversion level drops very low over the continent, would this be the case. So, the air traversing the site will be cooled by contact with the ground area upwind of the observatory that protrudes above the inversion level. Limited available data (Table 1) shows that this relationship between distance from the coast and nocturnal temperature change is observed.
6. Atmospheric Turbulence and Seeing Quality
Another factor determining observing quality concerns the effect of high frequency temperature fluctuations associated with atmospheric turbulence on image quality. This effect is commonly called "seeing". Turbulent mixing of air parcels of different temperatures (and hence, densities) produces temporal and spatial variations in atmospheric refractive index. This distorts the wavefront of light reaching the telescope mirror, causing motion, blurring and scintillation of the focused image. In selecting a telescope site, therefore, it is desirable for the air above the site to be microthermally quiet. There are three main regions of the atmosphere where turbulence occurs (Fig. 6).
i) Near-surface effects: The ground layer extends from a few centimeters to several tens of meters (typically 20m-100m) above the ground and is characterized by turbulent flow. Frictional effects result in an increase in wind speed with height within the ground layer. This produces a wind shear within the layer that generates and supports turbulent mixing. Horizontal variations in layer depth and magnitude of microthermal fluctuations occur in response to differences in topography, surface roughness and surface material that affect the dynamical and thermal properties of the air near the ground. A significant difference in ground layer characteristics exists between daytime and nightime conditions. Surface heating in the daytime enhances mixing and deepens the ground layer while surface cooling at night supresses mixing causing the ground layer to be much thinner. By virtue of the physical processes involved, spatial variations in the magnitude and extent of microthermal activity within the ground layer occurs at the micro-scale to sub-mesoscale (100 m to 10 km). Consequently, ground layer effects are important in determining local scale variations in site quality following the selection of an observatory site.
ii) Mid-tropospheric effects: This layer extends from the top of the ground layer to between 300 m and 1000 m above the observatory site. It only exists as a distinct layer if the observatory site is above the inversion level that marks the top of the boundary layer. The upper boundary of the perturbed layer is determined largely by the effects of local and upwind topography. If the site is located on a mountain range oriented perpendicularly to the prevailing wind the air is forced to flow over the mountain barrier. The air stream is compressed and therefore accelerates, creating a wind shear that may be strong enough to support turbulent mixing. If air being mixed in this manner has different temperatures enhanced microthermal activity will occur.
Under calm to light wind conditions a drainage flow develops on the slopes of mountains at night. For a mountain range the flow is directed perpendicular to the crest. In the case of an isolated mountain the down-slope wind is directed radially outwards from the summit. Mass continuity demands that air being removed from the summit be replaced. This is accomplished by enhanced subsidence. Enhanced subsidence directly above the mountaintop can lead to the formation of a layer of greater stability or even an inversion layer. This gives rise to adjacent layers with different potential temperatures. If mixing occurs under these circumstances, microthermal activity can increase in the perturbed layer.
Finally, if a mountain range which lies upwind of the observatory site is oriented perpendicular to the wind, the flow in the wake of the barrier is often perturbed. The mountain barrier does not perturb the flow only at the altitude of the mountain top, but also to a height of a 1000 m or more above the summit. Most often these perturbations take the form of lee-waves which are readily identifiable by their attendant lenticular (lens-shaped) clouds. While the flow through these waves is usually laminar, changes in the thermal stability or wind speed can cause it to break down into turbulent eddies, called rotors, at any time. The effect of a mountain range on the air stream transversing it can extend for 100 km or more in the downwind direction. Wh en considering the quality of continental sites, therefore, the presence of an upwind mountain range is an important factor.
iii) Upper-tropospheric effects: The third major region of microthermal activity is the layer directly below the tropopause (10-12 km) at the level of the jet stream. The jet stream is attended by strong wind shears and temperature gradients which promote turbulent mixing of thermally different layers. At a particular location the maximum wind speed, and hence greatest potential for turbulence generation, occurs at approximately the 200 millibar pressure level. The 200 mb wind speed can therefore be used as an indicator of the contribution of the upper troposphere to the "seeing".
Table 2 shows the 200 mb wind speed at selected locations nearby major observatories for the
mid-winter and mid-summer months of January and July. The continental (inland) sites show a
greater seasonal variation. There does not, however, appear to be any significant difference in the
annual average wind speeds at these sites. This suggests that differences in site quality within
the areas best suited for astronomical observations are probably not caused by upper-tropospheric
effects. However, this is an assertion that requires further investigation for verification.
7. Conclusion
Differences in observing quality at telescope sites are a consequence of temporal and spatial variations in atmospheric conditions. The atmospheric conditions that favor astronomical observations are associated with particular meteorological processes that operate at various scales. At the broad-scale, the subtropical high pressure systems are attended by weather conditions that
cause the western side of continents and the eastern oceans in the latitude belt between 15o and 35o to be ideally suited for astronomical observations. Within these areas, smaller scale variations in observing conditions are associated with differences in inversion layer characteristics, thunderstorm development, the frequency of cyclonic events, the relative importance of continental versus oceanic effects, upper-tropospheric wind speed, topography, surface roughness
and other less significant factors. An understanding of the physical relationship between
meteorological phenomena and the atmospheric conditions that determine astronomical
observing quality, provides a background with which important deductions about observing
quality at existing and potential telescope sites can be made. An improved understanding of this
relationship could potentially lead to the use of meteorological data to quantify site quality and to
forecast observing conditions.
Table 1
Nocturnal temperature change at some existing and possible telescope sites and the distance of these sites from the coast.
| Site | T (oF) | Distance from the coast (miles) |
| Mauna Kea | 2.0 | 24 |
| Haleakala | 2.2 | 18 |
| Cerro Tololo | 2.7 | 40 |
| Junipero Sera | 3.2 | 10 |
| Kitt Peak | 5.2 | 300 |
Table 2
Mean 200 mb (10-12 km) wind speed in knots for January and July at selected locations near existing and potential astronomy observatory sites.
| Location | January | July |
| Hilo, Hawaii | 48 | 30 |
| San Francisco | 42 | 34 |
| San Diego | 45 | 24 |
| Tucson | 55 | 9 |
| Tenerife, Canary Isl | 43 | 23 |
| Gilbraltar | 36 | 36 |
| Antofagasta, Chile | 35 | 54 |
| Quintero | 55 | 43 |
| Windhoek, Namibia | 19 | 51 |
| Cape Town, So. Africa | 50 | 47 |
| Perth, Australia | 47 | 66 |
| Sydney, Australia | 34 | 54 |