10.5 Combining single-dish and interferometer data

We add this section to this chapter with some trepidation since the combination of single-dish data into interferometric imaging is still an area more suited to research than to production. In principle, the problem is fairly simple. You begin by observing a region of sky with a single-dish telescope rather larger than the individual telescopes of the interferometer. From these observations, you make an image which you correct if necessary (e.g., by removing spectral baselines). Then you deconvolve the image removing the convolution of the sky with the beam of the large single-dish telescope. The “sky” observed with the interferometer is the product of the real sky (estimated by your deconvolved image) and the beam of the individual telescopes of the interferometer. Therefore, you multiply your deconvolved image with an image of the single-dish beam and Fourier transform the result. Adjusting the flux scales (usually of the single-dish data), you append or “feather in” the “visibilities” produced by the Fourier transform.

This is a lot of steps and contains several dangers, namely pointing, image alignment, the deconvolution, and the flux re-calibration. AIPS can provide you with some help. The imaging and image correction software is described earlier in this chapter. The deconvolution is tricky. Try DCONV first. It attempts an iterative solution of the deconvolution problem in the image plane. If that is not acceptable, try CONVL with OPCODE ’DCON (in 15JAN96 and later releases). This is a brute force deconvolution that will be very noisy at high spatial frequencies, but these frequencies will be tapered or truncated away later. A third approach is to use PATGN (OPCODE ’GAUS’) to make an image of the single-dish beam of the large telescope. APCLN (5.3.7) can then be persuaded to do a Clark image-based Clean; use a small restoring beam. Remember that this image will be tapered in the uv plane. It does not have to be beautiful in detail in the image plane.

The next step is to make an image of the interferometer single-dish beam on the same cell size and center as your deconvolved image. Use PATGN with OPCODE ’BEAM’ for this. Then multiply the result by the deconvolved image with COMB using OPCODE ’MULT(7.1). If this produces an image with any blanked pixels, run REMAG to convert the blanks to zeros. Then start trying IM2UV to produce a uv data set. Use UVTAPER to weight down longer spacings, FLUX to scale the visibilities, and UVRANGE to omit the outer spacings. (The first two options appear only in 15JAN96 and later releases.) You should use PRTUV, UVPLT, and even UVFLG on the output of IM2UV to make sure that the visibility phases and amplitudes of your single-dish and interferometer data are in reasonable agreement. Finally, combine the two data sets with DBCON and have fun with IMAGR (Chapter 5).