JSTOR

1. INTRODUCTION

Halfway between ground and space conditions, the Antarctic plateau has been considered for a long time as one of the best sites on Earth for astronomy. The dry, cold, and clear air is particularly well adapted for astronomical observations, increasing the sensitivity in the mid-infrared and opening many new windows in the submillimeter range (Burton 1996). The possibility of observing the Sun (Grec et al. 1980) or the stars without discontinuity during several months is another strong point of the Antarctic Plateau.

Since 1994, the LUAN (Laboratoire Universitaire d’Astrophysique de Nice) has been involved in site testing to study the seeing and the optical turbulence above the Antarctic Plateau. R. D. Marks made the first measurements of the optical turbulence in Antarctica, during the 1994 winter, between April 26 and August 10 at the South Pole using microthermal sensors on a mast at various altitudes (Marks et al. 1996). He performed a second campaign during the 1995 winter, between June 20 and August 18, using balloon-borne microthermal sensors to sense the whole atmosphere above the South Pole (Marks et al. 1999). The mean seeing was measured to be 1.86″ and the free atmosphere component was 0.37″. The boundary layer is highly turbulent and is associated with a strong temperature inversion and wind shear in the first 200 m above the ice level. As the boundary layer winds are of katabatic origin, exceptional seeings were expected at other positions in Antarctica, like Dome Argus (Dome A), the highest point of the plateau, or Dome Concordia (Dome C), a local maximum, where the wind speed and thus the optical turbulence are likely to be weak.

Pushed by the European program of glaciology EPICA, an infrastructure has been built progressively at Dome C and our laboratory worked on the site qualification since 1995. As expected, with a median value of 2.9 ms^-1 between 1984 and 2003, the wind speed at the ground level is one of the lowest ever recorded at an existing observatory (Aristidi et al. 2005b). During the 3-month summer campaigns in 2003 and 2004, exceptional conditions for a daytime observations were recorded by a differential image motion monitor (DIMM) at 8 m above the ice: a median value of 0.54″ and a seeing better than 0.4″ 25% of the time and large median isoplanatic angle (6.8″) have been observed (Aristidi et al. 2005a).

After the construction of the French-Italian permanent station at Dome C, (75°06′ South, 123°23′ East and an altitude of 3233 m), the first winterover took place in 2005 with the first sunset observed on February 12. Measurements of the optical turbulence and of the meteorological parameters using balloon-borne experiments started on March 15. According to the first results provided by Agabi et al. (2006), the optical turbulence in the atmosphere could be divided into two regions: a 36 m high surface layer responsible for 87% of the turbulence, and a very stable free atmosphere above with a median seeing of 0.36″. The situation appears to be similar to what was observed at the South Pole with a much thinner turbulent boundary layer, mainly due to the stronger thermal gradient occurring in winter. Another result is the progressive increase with time of the seeing recorded by the DIMM, at 8 m above the ice, reaching a median value of 1.6″ in August.

The aim of this paper is to complete these results with the complete data set provided by balloon-borne experiments and to fully qualify the optical turbulence structure above Dome C during the full winter, 2005 between February and November. We present a statistical study of the measurements and special attention is focused on the thickness of the surface layer and its evolution during the winter. The optimal elevation of a telescope above the ice, mainly limited by the free atmosphere and the upper part of the boundary layer, is given. Finally, the parameters used to design adaptive optics systems (seeing, coherence time, isoplanatic angle) and a four-layer decomposition are computed in order to establish the main specifications for future adaptive optics systems (AO) to be used on telescopes at Dome C.

2. EXPERIMENTS

During the 2005 winterover, two experiments were used to estimate the effects of the optical turbulence in the atmosphere as seen by a telescope. The DIMM, first developed by Sarazin & Roddier (1990), has been specially designed to work in polar conditions. It uses a C-11 telescope with an entrance pupil made of two circular subapertures recording the seeing at 8 m above the ice every 2 minutes. The seeing ϵ₀ is the full width at half maximum of the long exposure image of an unresolved star, at the focal plane of a telescope. Usually expressed in arc second (″), the seeing represents the angular resolution of a telescope for a given atmospheric condition. The atmospheric diffraction-limited equivalent pupil r₀, as introduced by Fried (1966), is related to the seeing at a given wavelength λ by the following expression:

Other parameters are defined to describe the spatial and temporal properties of the turbulent wave front and are used to design adaptive optics systems. The isoplanatic angle θ₀, as defined by Roddier et al. (1982), is the angular domain in which phase fluctuations induced by the optical turbulence are still correlated. It is the field of view in which adaptive optics remains efficient. The coherence time τ₀ is the time during which phase fluctuations are correlated. τ₀ sets the upper limit of an adaptive optic bandwidth. In 2005 May and June, the DIMM telescope was modified and used to monitor θ₀ from the scintillation of an unresolved star through a 10 cm pupil as stated by Loos & Hogge (1979). To simplify we will call it DIMM-θ₀ monitor hereafter. Thus simultaneous measurements of seeing and θ₀ are not possible.

The second instrument, used to qualify the site, consists of microthermal sensors raised by meteorological balloons. Developed by Azouit & Vernin (2005), these sensors have been intensively used to qualify other observatories like Cerro Paranal, Cerro Pachón, San Pedro Mártir, South Pole, and La Palma. This instrument gives an in situ snapshot of the local optical turbulence strength and of the meteorological parameters (pressure, temperature, humidity, and wind speed components) from the ground up to 20 km, with a vertical resolution between 5 and 10 m. This technique has proved to be reliable and gives quantitative optical turbulence measurements that have been successfully compared during many astronomical site testing campaigns with DIMM and Scidar optical devices (Azouit & Vernin 2005 and references therein). For a telescope at a given elevation h₀ above the ground, and from a microthermal profile defined from h₀ up to h_max, the seeing, the isoplanatic angle, and the coherence time for a wavelength λ are respectively estimated by

During the 2005 winter campaign, 34 balloons were successfully launched between March and October. Details are given in Table 1. All the flights start from the ground to a variable maximum altitude h_max. Most of the time the balloons exploded at the altitude h_max where the temperature decreases below -70°C. At a midlatitude site, a meteorological balloon can easily reach an altitude greater than 20 km. ϵ₀, θ₀, and τ₀ are computed from the profiles at two elevations above the ground (Table 2). Figure 1 shows the monthly median DIMM seeing and the balloon seeing at the same level (h₀ = 8 m). If one considers the seeing as sensed by the DIMM at the same time as the seeing estimated by the balloons (Table 1), a particularly good agreement is found (Fig. 2) considering that the balloons are not launched exactly at the same place as the DIMM and sensed a different part of the atmosphere. If the measurements in the first kilometers are significant to estimate ϵ₀ and τ₀, measurements above 20 km are essential to estimate θ₀ with less than 10% error. At Dome C, with the difficulties of protecting the balloon from the dry and cold air, the maximum altitude reached decreases progressively during the winter. For θ₀, only the balloon measurements during autumn are reliable and are in very good agreement with the DIMM-θ₀ monitor data set (Table 2). The other estimates of θ₀ from the balloon profiles are overestimated and give us an upper limit.

(17 KB)

Fig. 1.— DIMM and instrumented balloon seeing at Dome C for year 2005. The DIMM seeing is the median of seeing measurements over one month. The balloon seeing is computed at 8 m above ground level like the DIMM.

Open New Window

(14 KB)

Fig. 2.— Seeing sensed by the DIMM at 8 m vs. the seeing estimated from the balloon profiles at the same height and around the same time (Table 1).

Open New Window

TABLE 1 Instrumented balloons launched during the 2005 winterover at Dome C. Flight number (LUAN database), date, time (UT), and sun elevation H_⊙ at the ground level are given. Computed surface layer height h_sl (eq. [5]) and maximum altitude (above sea level) reached by the balloon h_max. Seeing as sensed by the DIMM ϵ₀(DIMM) and retrieved from the balloon profiles at 8 m and 33 m above the ice. ϵ₀(DIMM) is the median value of the seeing as sensed by the DIMM for 2 hr after the balloon launching.

Open New Window

TABLE 2 Statistics of the adaptive optics parameters and the surface layer height at Dome C during the winter 2005. Surface layer height h_sl as defined by eq. (5). Adaptive optics parameters (ϵ₀, θ₀, τ₀) are estimated for a wavelength of 0.5 μm. Two different elevations h₀ above the ground are considered. Median values, quartiles (), and number of data used n are given.

Open New Window

3. VERTICAL STRUCTURE OF THE OPTICAL TURBULENCE ABOVE DOME C

As analyzed by Abahamid et al. (2004), more than 60% of the global optical turbulence contribution lies within the first kilometer above the ground and is defined as the boundary layer. The residual optical turbulence in the free atmosphere is distributed in thin layers and is mainly concentrated at tropopause level. Here balloon profiles were resampled to a 10 m vertical resolution in the boundary layer, and 100 m in the free atmosphere. The median profile, the mean profiles of the potential temperature θ, horizontal wind speed V_h, and wind direction are computed for the whole campaign (Fig. 3) and for each considered season: the autumn profiles (Fig. 4) from March 20 to June 20, the winter profiles (Fig. 5) from June 21 to September 21 and the spring profiles (Fig. 6) after September 22. Unlike what have been observed at midlatitude sites, the optical turbulence is mainly restricted to the boundary layer (89% of the global optical turbulence) and the residual (11%) is in the free atmosphere.

(33 KB)

Fig. 3.— Median , mean potential temperature 〈θ〉, mean horizontal wind speed 〈V_h〉, and mean wind direction 〈Dir〉 profiles (dots), computed from the 34 balloon profiles launched during 2005 at Dome C. Standard deviations (error bars) are shown for the (in logarithmic scale), 〈θ〉, and 〈V_h〉. N is the number of valid flights at the corresponding altitude. The dashed curve is the median profile for midlatitude sites. In each main plot starting at 3250 m over sea level, the mean value is computed over a range of 100 m every 100 m. In each subplot, starting at 5 m over the ground, the mean value is computed over a range of 10 m every 10 m.

Open New Window

(32 KB)

Fig. 4.— Same as Fig. 3, considering only the balloons launched during the autumn 2005 at Dome C.

Open New Window

(30 KB)

Fig. 5.— Same as Fig. 3, considering only the balloons launched during the winter 2005 at Dome C.

Open New Window

(29 KB)

Fig. 6.— Same as Fig. 3, considering only the balloons launched during the spring 2005 at Dome C.

Open New Window

3.1. Boundary Layer

With 3.10^-14 m^-2/3 near the ground, the median value decreases rapidly to reach 10^-17 m^-2/3 at 100 m height. The mean wind speed is less than 3 m s^-1 in the first few meters and increases up to 8 m s^-1 at 100 m. Masciadri et al. (ARENA Site testing Workshop1), retrieved from PNRA radio soundings a median wind speed of 13 m s^-1 at 100 m and 8.6 m s^-1 at 10 m in the central part of the winter period (2006 July). Measurements in 2005 appear therefore more optimistic. Compared with the median profile of midlatitude sites (Fig. 3), an excess of optical turbulence is observed in the first 50 m. Here we define the thickness of this surface layer h_sl as the part of the boundary layer containing 90% of the total boundary layer optical turbulence: The first point is fixed at 8 m, to be at the same level as the DIMM elevation and to discard the turbulence very near the ground. During the 2005 campaign, h_sl fluctuates between 10 and 517 m (Fig. 7, Table 1, Table 2). Figure 7 shows the evolution of h_sl. The median value is 33 m and the 75th percentile is 42 m.

(23 KB)

Fig. 7.— Dome C surface layer thickness (left), as defined by eq. (5), during the campaign 2005. Each flight number is printed near the marker. The cumulative distribution (right) is the probability to be above the surface layer at a given elevation. The upper vertical scale is fixed to 100 m for convenience. Flight 569 is out of scale (517 m).

Open New Window

3.3. A Four-Layer Decomposition

The vertical structure of optical turbulence in the atmosphere is a succession of thin turbulent layers, between 10 and 100 m of thickness. The vertical distribution is given by the median vertical profiles. To simulate the effect of the turbulence on a wave front propagation, a more convenient way is to consider a set of equivalent layers. As stated by Roddier et al. (1982) the equivalent turbulent layers, for adaptive optics, are defined by the equivalent altitude h_∗ and the equivalent wind speed v_∗:

where [h₁,h₂] defines the considered part of the atmosphere. In Table 3 the equivalent layers in a four-layer decomposition for the three considered seasons are given. Mean values and standard deviations are estimated from the profiles.

TABLE 3 Four-layer model consistent with isoplanatic and coherence time. The statistics are realized for each considered seasons sampled by the 34 instrumented balloons during the 2005 campaign at Dome C. For each layer, the equivalent altitude h_∗, the equivalent wind speed v_∗, and the mean wind direction Dir are computed. Quartiles () and standard deviations ( ± σ) are given when the number of profiles is greater than 3. When the season or the layer is undersampled, values are computed from the median profiles. Altitudes are given above the ground level [m_agl] at Dome C (ground level at 3233 m above sea level).

Open New Window

4. DISCUSSION AND CONCLUSIONS

Compared with other sites and particularly with Paranal (0.7″), La Palma (0.7″), or Mauna Kea (0.6″) as published by Racine (2005), Dome C (1.4″) has a very poor seeing at 8 m above the ice. The major part of the turbulence is confined in the surface layer, which is found to be particularly strong and thin. To compare with midlatitude sites, using the fitted median profile in the boundary layer (Abahamid et al. 2004), the equivalent surface layer height is found to be greater than 200 m. To avoid this surface layer, telescopes at Dome C have to be elevated at least at 33 m above the ground to have a seeing better than 0.4″ half of the time. As already published by Agabi et al. (2006), the seeing at 8 m increases from 1.2″ in autumn to 1.6″ in winter. Figure 1 gives the evolution of the seeing at 8 m during the winter. The seeing is found to decrease in spring with a mean value of 0.8″.

Thanks to the low wind speed up to 14 km, the coherence time is particularly large as compared with other observatories (Lawrence et al. 2004). As for midlatitude sites the median free atmosphere seeing is close to 0.4″ and the vertical distribution of is quite similar. The amount of optical turbulence in the free atmosphere is not as weak as expected: the median isoplanatic angle found in this study (2.7″) and by Lawrence et al. (2004) (5.7″) for the same period of the year (autumn) are quite different one year apart. Deeper investigations are therefore necessary to better identify the causes of such a variability of turbulence in the free atmosphere in the same period of the year, as suggested by Masciadri et al. (2007), expecting the influence of the wind acceleration in winter, trigging the turbulence above 15 km.

Two solutions may be considered to improve significantly the seeing conditions. The first one could be to elevate the telescope up to 40 m where 75% of the time the seeing should be better than 0.4″. A classical adaptive optics system could then be designed to compensate the upper turbulent layers. At this level, thanks to the long coherence time and with such an exceptional seeing, the adaptive optics system should be particularly efficient and sensitive to faint stars. Another strategy could be to develop a specific ground layer adaptive optic (GLAO) for a telescope at the ice level. Using such a GLAO one might expect to reach a 0.4″ angular resolution over an estimated field of view of 5′ at 0.5 μm.2 Adding a classical single layer adaptive optics system, the angular resolution could then be improved down to the diffraction limit of the telescope within a 3″ field of view anywhere in the 5′ field. Such a medium angular resolution over a large field of view combined with continuous observations gives access to many astrophysical programs such as microlensing detection (Ragazzoni et al. 2004), asteroseismology (Mosser & Aristidi 2007), and pulsating stars.