10.3 Fourier Transform Spectroscopy for site testing

10.3.1 FTS measurements at Mauna Kea

Broadband submillimeter atmospheric measurements with a Fourier-transform spectrometer have been performed in recent years at the Caltech Submillimeter Observatory (CSO) with the primary goal of accurately measuring the atmospheric transmission in the submillimeter-wave domain and using the data to build a state-of-the-art transmission model.

The instrument used for this work has been described in Serabyn & Weisstein [1995]. Recently, new 1.1 THz and 1.6 THz low-pass filters have been installed in front of the $^{3}$He cooled bolometer detector in order to efficiently cover both the full subTHz domain and the supraterahertz windows that are predicted by theory. The first successful measurements on Mauna Kea using the 1.1 THz filter were obtained on the night of 1998 April 1$^{st}$ under extremely dry ``El Niño'' conditions. It was estimated that the H$_{2}$O column was under 0.2 mm, so it is clear that these data offered the best opportunity for a determination of the dry and wet longwave nonresonant terms [$\kappa$ $_{c,dry}(\nu)$, $\kappa$ $_{c,H_{2}O}(\nu)$]. However, to separate one from the other, an independent measurement under the same P/T conditions was needed. This occurred in a second run in July 1999, greatly simplifying our analysis (see Figure 10.2). From these 2 datasets, our goals have been to:

I) Extract the dry continuum from our measurements and determine its origin.

II) Determine the H$_{2}$O excess absorption in the submillimeter domain in low humidity conditions (when only foreign-gas collisions have to be taken into account), and compare it to proposed formulations.

The separation of these two terms using the data from the upper panel of Figure 10.2 is described in detail in Pardo et al [2001a] and leads to the formulation given in section 10.2.5. This is the first time that such a separation is done successfully in the submillimeter.

Our results indicate that the H$_{2}$O and dry continuum-like terms of existing models are not accurate in the submillimeter range and that the models should be updated accordingly. This has been done in Pardo et al [2001b]. The lower panel of Figure 10.5 shows the April 98 data and the separate opacity contributions that add up to fit the observed opacity.

Figure 10.2: Upper panel: High resolution fully calibrated FTS atmospheric zenith spectra ($\Delta \nu$=199 MHz) obtained with the FTS experiment at Mauna Kea (Hawaii). Bottom Panel: Fit of the above April 98 zenith spectrum, shown in terms of opacity on a logarithmic scale. The opacity contributions from the different species and the continua leading to the final fit are plotted separately (assuming a VVW line shape for water). The middle panel shows the fit residual when using a VVW lineshape. The bottom panel is the difference between the fit considering VVW lineshapes and the one considering Zhevakin-Naumov (ZN) profiles.

10.3.2 FTS measurements at Atacama (future ALMA site)

Two other FTS experiments are reported at sites in the Atacama desert, undergoing site testing measurements in areas selected for the Atacama Large Millimeter Array (ALMA) construction.

The first of such experiments has been operated by Nobeyama Observatory at Pampa la Bola, 4800 m above sea level in northern Chile on September 1997 and June 1998. The instrument is a Martin-Puplett type FTS with an InSb bolometer as detector. The frequency range covered is 150 - 1600 GHz (or 2 mm to slightly under 200 $\mu$m wavelength). Further details on the instrument can be found in Matsushita et al [1999].

On the morning of 1998 June 17, the best atmospheric transmission spectrum of the experiment was recorded. (Fig. 10.3, top). However, due to the limited sensitivity around 1500 GHz and offset errors in the phase correction of the Fourier-transformed spectra, there are some systematic errors in the transmission spectra around the 1450 - 1600 GHz window (it could be up to 10% in transmission). During the observing run, side-by-side measurements with the second FTS experiment in the area (at Chajnantor [Paine et al 2000]) were performed. The measured transmission spectra showed very good correspondence to each other within an accuracy of $\le1$% in the 650 GHz and 850 GHz windows.

Again, radiative-transfer calculations using the model ATM were performed to fit the data. The best fit of the June 17$^{th}$ spectrum is shown in Fig. 10.3. The model, that took advantage of what we learned from the Mauna Kea data, fits very well the observed spectrum with only one free parameter (the precipitable water vapor column [ $N_{H_{2}O}$] above the site) except for frequencies higher than 1350 GHz, where the measurement suffers from systematic errors. The fit results in a water vapor column of 0.284 mm.

Figure 10.3: Atmospheric transmission spectrum obtained at Pampa La Bola on 1998 June 17 (courtesy of Satoki Matsushita). The absorption line and continuum components from the ATM fit of the data, along with the fit residual are also shown.