On the critical importance of sky noise

Sky noise refers to fluctuations in total power or phase of a detector caused by variations in atmospheric emissivity and path length on timescales of order one second. Sky noise causes systematic errors in the measurement of astronomical sources. In an instrument that is well-designed, meaning that it has no intrinsic systematic errors, the sky noise will determine the minimum flux that can be observed, the flux below which the instrument will no longer "integrate down". This flux limit is proportional to the power in the sky noise spectral energy distribution at the switching frequency of the observing equipment. Sky noise causes observational techniques to fail: fluctuations in a component of the data due to sky noise integrates down more slowly than t^-1/2 and will come to dominate the error during long observations. For deep background experiments, it is important to choose the best possible site. Sky noise is a source of systematic noise which is not within the control of the instrument designer, and the limits to measurement imposed by sky noise are frequently reached in ground-based submillimeter-wave instrumentation.

The success or failure of various submillimeter-wave observational techniques depends critically on sky noise. Just as it makes no sense to carry out visual-wavelength photometry in cloudy weather, there is an atmospheric opacity and sky noise above which any particular submillimeter-wave observational technique will fail to give usable results. This threshold depends on the details of the particular technique and its sensitivity to the spectrum of atmospheric noise.

Sky noise measurements at the Pole

Sky noise at the Pole is considerably smaller than at other sites at the same opacity. As shown here, the PWV at the Pole is often so low that the opacity is dominated by the "dry air" component; the "dry air" emissivity and phase error do not vary as strongly or rapidly as the emissivity and phase error due to water vapor. 

PWV vs. Opacity at Pole. Opacity at various frequencies measured by skydips plotted against precipitable water vapor. Fits to the data show a positive intercept at zero PWV, a measure of "dry air" opacity. The `"dry air" component of opacity is less variable than the water vapor component and therefore generates less sky noise. Figure courtesy of R. Chamberlin.

Sky noise at the South Pole is low because the precipitable water vapor is low.

The spectral energy density of sky noise is determined by turbulence in the atmosphere and has a roughly similar spectral shape at all sites. Measurement of spectral noise at one frequency can therefore be extrapolated to other frequencies. 

Noise and Opacity Measurements at 350µm from Three Sites.  These plots show data from identical NRAO-CMU 350µm broadband tippers located at Mauna Kea, the ALMA site at Chajnantor, and South Pole during 1998. The upper plot of each pair shows the rms deviation in the opacity during a one-hour period---a measure of sky noise on large scales; the lower plot of each pair shows the broadband 350µm opacity. The first 100 days of 1998 on Mauna Kea were exceptionally good for that site. During the best weather at the Pole the rms deviation in the opacity was dominated by detector noise rather than sky noise.

This figure gives a direct indication of sky noise at submillimeter wavelengths at the largest scales. The upper plot for each site shows the root-mean-square deviation of opacity measurements made within an hour's time. As a measure of sky noise, this value has two defects:

1. During the best weather it is limited by detector noise within the NRAO-CMU tipper (which uses room-temperature bolometers).
2. The ~10^-3 Hz fluctuations it measures are far from the switching frequencies used for astronomical observations. 

It is, however, an indication that the power in sky noise at Pole is usually several times less than at Mauna Kea or Chajnantor. 

Other instruments are sensitive to sky noise at frequencies near 10  Hz and can be used to give quantitative results over more limited periods of time. Sky noise at the Pole has been measured in conjunction with cosmic microwave background experiments at the Pole. Python, with a 2.75^o throw, had  1 mK Hz^-1/2 sky noise on a median summer day, and White Dish, which had a 0.5^o throw, was much less affected by sky noise. Extrapolating to 218 GHz and a 0.2^o throw, the median sky noise is estimated to be 150µK Hz^-1/2 even in the Austral summer, lower by a factor of ten than the sky noise observed during Sunyeav-Zel'dovich (S-Z) effect observations on Mauna Kea by Holzapfel and collaborators.  We therefore expect that kinetic S-Z effect experiments which have barely worked during the best conditions at Mauna Kea will work well from the Pole.

Lay and Halverson have compared the Python experiment at Pole with the Site Testing Interferometer at Chajnantor. These are very different instruments, but the differences can be bridged by fitting to a parametric model. Lay and Halverson have developed an atmospheric model for sky noise with a Kolmogorov power law with both three- and two-dimensional regimes, and have applied it to data from Python and the Chajnantor Testing Interferometer. They find that the amplitude of the sky noise at the Pole is 10 to 50 × less than that at Chajnantor.  

Sky noise at the South Pole is significantly less than at other sites. This will translate into an order-of-magnitude improvement in the flux limits that can be observed.

Send mail to help@cfa.harvard.edu with questions or comments about this web site.
Last modified: November 05, 1999