What happens when the pipeline runs? (part III)

Imaging

The final stage of the pipeline is to produce images from the calibrated visibility data. Because synthesis arrays sample the u-v plane at discrete locations, the observed intensity distribution can be thought of as the true intensity convolved with the sampling function (or beam). The main process involved in imaging is deconvolution, which tries to solve for the true intensity, given the observed data and the beam. Deconvolution is most commonly done using the CLEAN algorithm and is often referred to as CLEANing. For information about how CLEANing works, see the Miriad Users Guide section on deconvolution.

When creating the image, each visibility sample is given a weight. There are several different weighting schemes available, to account for things such as differences in the density of sampling in different parts of the u-v plane, or to improve sensitivity to extended objects. In the pipeline, uniform weighting is used, which gives each visibility a weight This gives a weight inversely proportional to the sampling density function. This minimises the sidelobe level, however the noise level is worse than in other weighting schemes.