Dedicated site-testing instruments have documented the 1.25-14 µm sky brightness (Chamberlain et al. 1999, Phillips et al. 1999). Typical values of sky brightness during winter at the South Pole are 70-180 µJy/square arcsecond at 2.4 µm, 0.12-0.19 Jy/square arcsecond at 3.6 µm, 0.8-1.3 Jy/square arcsecond at 4.8 µm and -30 Jy/square arcsecond at 9 and 11 µm. The incidence of clear, dark, skies suitable for astronomical observations is of the order of 50%. (See table below for more values and comparison to other sites.)
In the near and mid-infrared, the darkest atmospheric window lies between 2.3 and 2.45 µm. (See figures below.) The sky brightness can fall below 70 µJy/square arcsecond, to values comparable to those measured from balloon altitudes (Ashley et al. 1996, Nguyen et al. 1996, Phillips et al. 1999). The agreement between the South Pole and balloon data strongly suggests that the residual emission in this window during the coldest and darkest observing conditions comes predominantly from airglow at altitudes above 38 km. No correlation is observed between 2.4 µm sky brightness and auroral activity.
Practical benefits may be even greater at the longer wavelengths of the L and M bands, since at warmer sites the infrared arrays which have been developed for astronomical applications are rapidly saturated by the high thermal backgrounds. Furthermore, the benefits of reduced thermal flux are available over the full bandwidths of the atmospheric windows. The largest gains may come at 4.6-5.5 µm, where the observed fluxes appear to be more than 50 times smaller than at other sites and the extremely low values of atmospheric water vapor can help open the entire window for routine observations.
In the N band (8-13 µm), the sky brightness is lower by a factor of 3-10. The low sky fluxes and the homogeneity of the atmosphere also contribute to lower sky noise (Smith and Harper 1998). In the most transparent parts of the window, the fact that the strong temperature inversion at the South Pole makes the telescope colder than the mean atmospheric temperature is also significant, since it makes it easier to reach the limit set by the atmospheric emission. In this waveband, the gains realized for a specific type of measurement will depend in detail on the relative importance of sky brightness, telescope emissivity, temperature, and sky noise.
There are also modest gains at wavelengths shorter than 2.3 µm, where the sky brightness is dominated by OH airglow rather than thermal emission. The airglow is fainter by about a factor of two and the atmospheric windows are wider and more transparent.
Scattered sunlight exceeds intrinsic sky emission at wavelengths less than 2.5 µm when the Sun is higher than 10 degrees below the horizon. Scattered moonlight is unimportant at wavelengths longer than the J band (1.25 µm). At wavelengths longer than 3 µm, the quality of the observing conditions depends principally on temperature, so good observing conditions extend well beyond the boundaries of the polar night.
Click image for a larger, readable version.
Histograms of zenith intensity measured during wintertime at the South Pole, 1995, through the broadband J (1.25 µm), H (1.65 µm), K(2.2 µm), Kdark (2.3 µm), and M (4.8 µm) filters, and a 1% CVF filter at 2.43, 3.10, 3.60, and 4.00 µm. For the L and M-band fluxes, the signal contains a significant instrumental component. Zenith intensities are plotted against the fraction of dark-sky data in the flux bin. For more information, see Phillips et al. 1999.
Click each image for a larger, readable version.
In these plots, we present a comparison of the measured sky brightness at zenith between the South Pole and Siding Spring (Australia) from 2-4.3 µm. In the first figure (2-3 µm), the South Pole spectra was collected on 31 May 1994 when the ambient temperature was -62 °C, while in the second figure (3-5 µm) the South Pole spectra was collected on 2 June 1994 when the ambient temperature was -66 C. In both plots, the Siding Spring data was collected on 9 December 1993 at +10 C. The units on the vertical axis of the left-hand plot are micro-Jy arcsec. Note the large dip in the South Pole sky brightness between 2.25-2.5 µm compared to the temperate latitude site. The South Pole sky is 20-100 times darker at these wavelengths than any other sight on earth which is accessible for large telescopes. CARA specifically designed the so-called K_dark (2.29-2.43 µm) filter to exploit this atmospheric window. The units on the vertical axis of the right-hand plot are milli-Jy arcsec. Note that between 2.9-4 µm, the South Pole sky is more than 10 times darker than at a temperate latitude site. For more information, please see Ashley et al. 1996.