Theoretical models of atmospheric opacity at submillimeter wavelengths are unreliable predictors of the opacity at a particular site, because they depend upon semi-empirical terms that may not be correct for the site in question. Few measurements have been made covering a broad frequency range, and above 1 THz, present atmospheric models are essentially unconstrained by experimental data. This is an important problem in the frequency range from 1 THz to 2 THz, where several windows exist that may be usable for ground-based astronomy from a sufficiently high, dry site.

We have deployed a Fourier Transform Spectrometer (FTS) at 5000 metres altitude at a site near Cerro Chajnantor in northern Chile, located within the unusually dry climate of the Atacama. The FTS measures the sky brightness in order to determine the zenith atmospheric opacity over the frequency range 350 GHz to 3 THz, with 3 GHz resolution. It operates autonomously, recording spectra at 10-minute intervals. The primary goal of this project is to measure the opacity in the transmission windows above 1 THz, and to assess the prospects for using these windows for ground-based astronomical observations.

The FTS has recorded a nearly continuous record of opacity covering two months of the 1998 austral winter. We find that significant transmission is frequently observed from the Chajnantor site in windows at 1.03 THz, 1.3 THz, and 1.5 THz. Our intention is to continue operation of FTS to continue gathering statistics on the opacity. Here we present our measurements and data reduction procedures, and discuss implications for astronomy and atmospheric modeling.