To make sensitive continuum images over wide fields of view at high resolution we need mosaicing and multi-frequency synthesis software which can handle large numbers of channels and make continuum and polarization maps over the entire primary beam. We must be able to look at the data; some sort of automatic flagging, preferably fairly flexible, is probably also essential. We must be able to find and reasonably analyze point sources in the midst of diffuse backgrounds at many frequencies; we must be able to do multi-resolution CLEANs or the equivalent routinely. We must reach dynamic ranges of :1 routinely, and several million to one with some effort. The latter target requires a concerted study of the problems which limit us now.
Real-time imaging is important for rapid imaging of variable sources to make good use of the new high-sensitivity systems.
Improvements in computing should accompany the enhancement of the VLA, as projects of all types will benefit. These should include streamlined procedures for combining data from the E configuration with data from the larger VLA configurations, autocorrelations, and single dishes. Techniques going beyond the current robust weighting and multi-resolution CLEAN may be needed to image structures on all available scales. Astrometry software for using VLA images fully in multi-wavelength high-resolution studies will also be important.
With more spectral-line channels, data sets will be larger, requiring files larger than the current 2 Gbyte limit and very large disks, perhaps with the ability to write extra-large files across disks. True joint-deconvolution mosaicing will be in demand. Both will bring about demand for faster processing.
Visualization and display of data sets with many channels will require more sophisticated displays and high-bandwidth transfer from recording media to such displays.
Finally, the on-line computing system must be upgraded to handle the additional antennas in Phase 2 of this plan.