The dirty map F resulting from a normal, single-field interferometric
observation can be described by the equation:
Due to the coupling between the receiver horn and the primary mirror of the antennas (see lecture by A. Greve), the primary beam B is, to a good approximation, a gaussian. Its FWHM (proportional to the ratio of the wavelength to the antenna diameter ) can therefore be used to define a ``field of view''. Table 15.1 gives the resulting values for the Plateau de Bure interferometer, for different frequencies. To map regions more extended than the primary beam width, it is necessary to observe a mosaic of several adjacent fields. Clearly, due to the gaussian-shape of the primary beam attenuation, these fields have to strongly overlap to ensure a roughly uniform sensitivity over the whole mapped region.
A further complication arises from the lack of the short-spacings information in the interferometer data set. Due to their diameter, the antennas cannot be put too close to each other, which results in a minimal measured baseline (24 m at the Plateau de Bure). Even if projection effects reduce the effective baselines, a central ``hole'' in the data distribution in the uv plane cannot be avoided. As a consequence, the extended structures (whose visibilities are confined in a small region in the uv plane) are filtered out. The largest structure it is possible to map with a single-field interferometric observation is thus even smaller than the field of view, and can be very roughly estimated by the ratio of the wavelength to the minimal baseline (Table 15.1). In the framework of mosaic observations, the short-spacings problem has however to be thought in slightly different terms, because it introduces now artifacts on an intermediate scale (see Sec. 15.5) but also because it can, at least in theory, be partially solved (Sec. 15.2).