The next step in planning your observations involves thinking about , the largest angular scale of structure that you must sample well to produce a useful final image. will be the angular diameter of the most extended structure that you must image accurately. This will usually be the diameter of the most extended component of your target that is of astrophysical interest.
Do not confuse with , the required field of view, which is discussed below. When observing a target 10'' in extent in the presence of a point confusing source 1' away, you would set = 10'', not = 1'.
determines the shortest baseline length that you need in your data for the images to be useful. The visibility of a circular Gaussian source with FWHM falls to half of its peak amplitude at a normalized baseline length of 91,000/ wavelengths, if is in arc seconds. You will not be able to image a two-dimensional structure of size well unless projected baselines at least this short are present in the data. Depending on the density of tracks in the inner u-v plane, and on the sensitivity per baseline, more conservative criteria may be needed--halving the above estimate gives a baseline on which the visibility falls to about 85% of its peak. Note also that the detailed array geometry, target declination and hour-angle range determine how baselines are foreshortened by projection. The most reliable approach is therefore to look at u-v plots for your expected antenna layout, declination, and hour-angle range to ensure that your planned short-baseline coverage meets the requirements imposed by . (This can be done for any antenna array using the program UVSIM in the NRAO's Astronomical Image Processing System.)
The factors affecting the choice of resolution cannot be estimated reliably in advance if the source structure is poorly known. If you are not sure what your source may look like, its is safer to guess on the side of low resolution when designing an initial observation. A ``diagnostic" low resolution image may tell you the source's total angular extent, and warn you about surrounding emission. This information will let you optimize the parameters for a more time-consuming high resolution study. It is also easier to justify reobserving a detected emission region at high resolution than to justify reobserving at low resolution what previously appeared to be empty sky!