The simulation consists of two steps, setting up the measurement set (basically computing the uvws), and then inverting a model.
- CASA6 Python libraries (https://casa.nrao.edu/casadocs/casa-5.6.0/introduction/casa6-installation-and-usage)
- Sky model as FITS image (https://astronomers.skatelescope.org/ska-science-data-challenge-1/)
- wsclean (for inversion; https://sourceforge.net/projects/wsclean/)
For the first step I used casa for that (see simobs.py) and download an image from the SKA data challenge (https://astronomers.skatelescope.org/ska-science-data-challenge-1/). Also see http://library.nrao.edu/public/memos/ngvla/NGVLA_55.pdf. The cfg file used is the uvw coordinates of the array on the tangent plane set in simobs.py (this turned out to be the easiest way for me to get things working without a deep understanding of how UTM works in CASA).
./simobs.py <image.fits>
For the second step I used wsclean which handles wide-field effects and is numerically more stable. I did have to make sure that the model fits file have unit Jy/px (e.g. https://casaguides.nrao.edu/index.php/Simulation_Recipes#Flux_Density_Scaling), and then changed the fits header so that the sources are above horizon.
To predict I did something like
wsclean -predict -size 1001 1001 -weight natural -scale 0.604asec -name SKAMid_B1_100h_v3_scaled_cropped_fun DSA_2000_ant_60x15s_mma_SKAMid.ms
where the model image has to have the name SKAMid_B1_100h_v3_scaled_cropped_fun.fits.
To image
wsclean -no-update-model-required -niter 0 -size 1200 1200 -scale 0.5asec -weight briggs 0 -datacolumn MODEL_DATA -make-psf -name SKAMid_60x15s_bright DSA_2000_ant_60x15s_mma_SKAMid.ms
One can skip the -make-psf option if they don't need the PSF.
TODO: Running subbands in parallel would make things faster.