Interferometer Phase Correction Using Millimetre Wave Atmospheric Water Vapour Sounding

A Project Proposal

Peter J. Hall, 16 October 1995.

Updated 10 January 1996.


Water vapour is a poorly-mixed constituent of the troposphere and, at short cm- and mm-wavelengths, contributes significantly to the effective delay (phase) seen by cosmic radio signals propagating through the Earth's atmosphere. WV inhomogeneities ('blobs') have scale sizes ranging from centimetres to kilometres, making it likely that individual elements in a synthesis array will view an astronomical source through a different WV column. Of course, the WV distributions vary as winds aloft carry an evolving WV pattern past the array. The differential, time variable, phase distortion produces image degradation in images from the telescope. The effect is most severe at short wavelengths and during seasonal conditions which produce severe atmospheric path fluctuations (e.g. summer days).

In some experiments it may be possible to correct the image distortion off-line using techniques such as self-calibration. However, at short wavelengths sources are generally weaker and the integration times required to reach S/N ratios which permit such techniques to be effective often exceed the coherence times imposed by atmospheric phase fluctuations. We therefore seek a method by which the coherent integration time of high frequency interferometers can be increased.

Phase Correction Using Water Vapour Sounding

In principle, it ought to be possible to arrive at atmospheric phase corrections by measuring continuously the amount of water vapour in the beam of individual array elements. Experiments beginning in the 1970s sought to use the results of 22 GHz water vapour radiometry (i.e., measurement of atmospheric WV emission brightness) to arrive at corrected connected-element and VLBI phase. While the work produced valuable theoretical and practical outcomes (including inversion algorithms to relate water emission brightness to electrical length and phase), the 22GHz emission brightness turns out to be relatively insensitive to water compared with millimetre-wave transitions. In addition, it is more sensitive to other atmospheric constituents than the higher frequency measurements. Finally, despite intensive development efforts, most water vapour radiometers (WVRs) lacked sufficient sensitivity and stability to verify the physics of the proposed phase correction technique.

Interest in the WVR technique waned somewhat until a few years ago when the pressure to develop a viable mm-wave phase correction scheme for the BIMA (or Hat Creek) Array, the Australia Telescope Compact Array, the VLA, the IRAM Plateau de Bure Interferometer, and proposed sub-mm interferometers, forced a re-appraisal. In particular, the initial BIMA work (based on powers measured in the 3 mm continuum) yielded some data showing a high correlation between interferometer phase and antenna total power differences.

AT Involvement and Results So Far

Despite limited resources being directed to the phase correction project thus far, two significant results have been obtained:

  • A loose collaboration between Lazareff, Bremer and other IRAM workers, and Hall (ATNF), formulated some SIS receiver design guidelines and PdBI experiments; experimental results illustrating the value of 230 GHz sounding in correcting 3mm (and, recently, 230 GHz) phase are shown in the attached plots and IRAM Newsletter extract. Notice that the technique has turned unusable time on the PdBI into useful observing time.

  • A collaboration between Hall (ATNF) and Abbott (Ph.D student, SU EE) has produced a precision 225 GHz WVR with the required sub-1K sensitivity and a stability exceeding 1 part in 104 . Table 1 is a summary of the prototype instrument specifications and measured performance. The important point is that the 0.16 K stability achieved should permit ATCA 3mm phase correction to better than 10 deg. rms. For other ATCA bands, the rms corrected phase fluctuation should not exceed 0.1 deg per GHz sky frequency.

Earlier presentations have shown that, on the basis of scaled 9 GHz interferometric phase measurements, the expected 100 GHz ATCA phase fluctuations will not preclude 3 mm observations, at least during winter nights. Incorporation of a WVR-based phase correction scheme should extend the usable mm-wave time (by factors of 2 or more) as well as improving the instrument performance at short cm-wavelengths significantly.

The Proposal

The ATCA will never be equipped with 230 GHz astronomy receivers and, as future systems evolve on the PdBI, simultaneous 230 GHz sounding will be difficult. Both instruments, as well as future arrays, will therefore benefit from the development of independent WVRs. Given the success of the prototype Australian radiometer, and the desirability of verifying the viability of sounding using a sensing beam independent of the astronomy antenna optics, I propose an ATNF project to:

  • Construct a second 225 GHz WVR.

  • Use both WVRs in conjunction with two elements of the ATCA to establish the benefits of phase correction at short cm- and mm-wavelengths.

  • Liaise with IRAM, ESO, NRAO and other players with the intention of implementing similar sounding schemes on the PdBI and proposed large millimetre arrays.

Resource Estimates

The estimates below are based on experience gained in producing the prototype WVR and allow for the fact that much of the work involves replication of the existing design. WVR #1 has also proved to be slightly over-designed in a few areas and some savings will be possible in WVR #2. An important practical consideration is that the instrument could be built with only minimal call on the Receiver Group if this is deemed desirable.

Financial investment: $40 k
Electronics tech/eng time to construct WVR #2 0.8 p-y
Workshop time to construct WVR #2 0.1 p-y
Engineering firmware development (WVR #1 & #2) 0.3 p-y
ATCA system software development 0.2 p-y
Miscellaneous telescope installation assistance etc. 0.1 p-y
Total human resource investment 1.5 p-y


For a modest investment the ATNF can maintain a leading position in a niche program important to short-wavelength interferometry. We will be well-placed to maximise the scientific returns from our own mm-wave programs should MNRF funding permit the upgrades, to significantly improve the ATCA performance in the soon-to-be-available 1 and 2-cm bands, and to make an important fundamental contribution to international projects involving next-generation mm and sub-mm arrays.

Table 1 - ATNF Radiometer Specifications & Prototype Performance

Operating frequency 225 GHz
Radiometer type Switched (quasi-optical chopper)
*switching mode, algorithm etc programmable
Optical system Classical cassegrain (0.5 m primary)
Front-end Schottky mixer; 2nd harmonic mixer; InP Gunn LO
Backend DSP system for data acquisition and control
(TMS 320C25 mP, Xilinx FPGA hardware controller, multi-channel digital, analog I/O)
Communication Fibre optic serial link
System temperature 1600 K
Half-power beamwidth 11 arcmin.
Minor lobes < 10 dB
Beam efficiency 50 %
Sensitivity 0.08 K (1 sec. integration)
Stability Better than 0.08 K in 1 s; better than 0.16 K thereafter

Figure 1 and Figure 2: Caption

Plateau de Bure Interferometer data recorded 18/4/95 using two antennas separated by 160.5m. Total power differences at 230 GHz are compared with 86 GHz phase fluctuations evident on 3C279. The raw phase data show large excursions over short timescales, resulting in essentially useless astronomical data. By applying phase corrections based on atmospheric sounding, the corrected phase is constrained to perhaps 10-15 deg. rms over timescales of minutes, making useful astronomy possible.

Note that this performance was achieved using a sounding system which is a factor of two poorer in stability than the ATNF WVR.

(Courtesy of Michael Bremer, IRAM Grenoble).


Extract from latest IRAM newsletter.

(available at