With the installation of the two new dual frequency receivers, it became possible to study the correlation between the total power fluctuations at 230 GHz and the phase noise induced by atmospheric water vapor. This is an important step to extend the total amount of available observing time, because the performance of the interferometer in general and especially at longer baselines and higher frequencies depends critically on the variations of water vapor along the line of sight, which is one of the least stable constituents of the atmosphere (with time scales down to a few seconds). Due to its spectral lines in the infrared, the water molecule changes its optical properties when moving from the visible range to mm wavelengths, notably the refractive index. That means that a ``clear sky'' in human terms is in fact filled with refracting bubbles with sizes from some kilometers down to less than a meter for the PdB Interferometer. During summer, the available observing time is mainly limited by the abundance of the turbulence cells above the telescopes.
Various phase correction schemes are under development at other interferometers (ATNF, Nobeyama, VLA, BIMA, and others) that partly rely on dedicated radiometers for remote atmospheric sounding because of the stringent stability requirements for the receivers. At IRAM, the previous mixer generation was not sufficiently stable to isolate the atmospheric component from the total power fluctuations, but the new receivers showed promise that the necessary requirements could be met. The dual frequency capabilities would allow to use the spectral region around 230 GHz for the water vapor estimation which shows a sensitivity about six times higher (Bremer 1994) than the 22.2 GHz line monitored in conventional radiometers (Moran and Rosen 1981, Elgered, Rönnäng and Askne 1980) and to correct the phase at both frequencies in real time.
In the evening of April 18, antennas 1 and 4 observed the quasar 3C279 (a strong source regularly used as phase calibrator) at 86 and 230 GHz simultaneously on a baseline of 160.5 m. Adjustments of the hardware were still in progress, so the data is not calibrated. Figure 4 shows the variations of amplitude and phase at 86 GHz with time, the ``raw'' difference in the total powers TP of antennas 1 and 4 at 230 GHz, and a fit of the form
to the phase, which allows for a linear drift in time. After
the subtraction of a
linear slope, the uncorrected phase shows a mean deviation of
, whereas the corrected phase shows
a mean deviation of 
. Approximating the atmospheric phase
shifts with a Gaussian random variable of mean deviation 
(Thompson 1986),
one can describe their influence on the visibilities V as
With the reduction of the noise, the sensitivity of the interferometer
has therefore been improved by a factor of 
.
The true improvement is however even greater in our case, since
phase calibration would be only marginally possible with a phase rms of
, while it should be easy with a rms of 
.
Although these results are encouraging, it will be necessary to do more testing under different atmospheric conditions before real time phase correction can be offered on a regular basis. Especially in the presence of clouds, the calibration factor between phase and total power fluctuations is supposed to change dramatically which could cause a counterproductive correction.
Further observations will also test the validity of the phase correction when the interferometer is switched between nearby phase calibrators.
Figure: Observation of 3C279 on April 18 with antennas 1 and 4 on a
baseline of 160.5 m. Fit parameters were 
(cf. equ. 7).
References:
Bremer M. 1994, The Phase Project: First Results, IRAM internal report
Moran J.M., Rosen B.R. 1981, Radio Science Vol. 16 No. 2, pp. 235-244
Elgered G., Rönnäng B.O., Askne J.I.H. 1980, Research Report No. 141, Chalmers University of Technology, Gothenburg, Sweden
Thompson A.R. 1986, Interferometry and Synthesis in Radio Astronomy, Wiley & Sons
Michael BREMER, Stéphane GUILLOTEAU,
Robert LUCAS
