CASA synthetic observations

To evaluate imaging robustness & quality, it is necessary to do some sort of synthetic observation. These synthetic observations can be done on real images, e.g. Herschel data shifted to greater distances, or on simulated data.

Some examples:

General process:

  1. Import file into CASA .image form (e.g., importfits)
  2. Load the .ms file and replace visibilities w/fourier transform of image
  3. Clean.

The hard part is setting up the files to be imported.

For example, you can import a FITS file, which to me seems to be the most straightforward approach:

           # 18" = 1.22 lambda/D

The beam probably matters for Jy->K, or Jy/beam->Jy conversion. The defaultaxesvalues should correspond to the pointing center of the interferometer pointing or mosaic.

Next is bringing this image - which we now pray is in the correct units - into UV space.:

success = sm.predict(perseus_casa_image)
assert success
# TODO: get these from ASDM_CALWVR and WEATHER
success2 = sm.setnoise(mode='tsys-atm', relhum=60.0, pwv='2mm', tatmos=265.0, )
success3 = sm.corrupt()

This snippet loads the .ms file (the visibilities), then overwrites them with the fourier transform of the image using sm.predict. It's not clear whether setvp is necessary. setnoise + corrupt just adds "appropriate" atmospheric noise to the visibilities.

From there, you should just use the same clean parameters as for the original data, or try to optimize clean here and use these parameters on the real data.

The result will be the 'perfect' (no phase error, no amplitude error) interferometric image.