Non-imaging data analysis

Abstract

In an ideal world, astronomers would be able to regard a VLBI array as a blackbox imaging device: they would schedule their observations, wait for the correlation to be completed, pass their data through a standard analysis package, and out would come an image. But this ideal is far from being reached with current VLBI observations, and producing a standard image is not always the best way to analyze VLBI data. In this chapter I want to try to convince you that there is a lot to be learned by inspection of the data in the visibility domain or (u, v) plane. This is, after all, the domain in which the measurements are made, and where errors in the data are easiest to recognize: the Fourier transform involved in imaging spreads errors that are localized in the (u, v) plane throughout the image. There are some types of observation in which it is difficult or impossible to make an image, and in these cases it is necessary to interpret the observed visibility data directly. I shall also show that some quantitative astronomical questions can be addressed better in the visibility domain than in the image. For example, in comparing two images made at different times it can be difficult to determine whether apparent changes are due to real changes in the source, or just to differences in the (u, v) plane sampling and the imaging parameters. It is much more straightforward to compare the measured visibilities directly. I will restrict myself to continuum data (single frequency channel, single polarization), although the techniques are readily extended to more complex data sets. I will also ignore astrometric and geodetic applications, which are discussed in chapters 19 and 18.