The purpose of the post-real time calibration is to estimate as precisely as possible all the various terms of Equations 10 and 11. The calibration receipes described in this document are based entirely on their relative importance and behaviour (time and frequency dependence).
The basic principle is to measure the visibilities of sources of known (spatial and frequency) structure, in practice continuum point sources of known flux F for which Wbijk = F. By simple division, the product Ai.A*j.Sij.S*ijis determined at some time intervals, then interpolated at other times and applied to the sources to be studied.
The signal to noise ratio for these measurements may be the limiting factor, especially for the Sik terms since the bandwidths of the spectral channels are narrow. It is then necessary to take full advantage of the low time or low frequency dependence of all factors to optimize the signal to noise ratio on all determinations.
Another important limitation concerns the validity of the interpolations used to predict the values of the calibration factors on the studied sources. It is desirable to estimate the quality of these predictions, and to ignore data which for any reason may deviate substantially from the predictions (instrumental problems, strong atmospheric variations, etc...). This part of the calibration is often called ``editing''.