Grading the Image Plane models

Final goal is to construct a data-base of image plane models which is readily available to the users. An objective grading of these models would allow observers to make quick decisions/choices without worrying about the details of the data-processing done to obtain the model for a given calibrator. We propose that the image plane models be classified into I+ and I- categories. Former corresponds to the image plane models that were obtained via calibrating the visibility data using some standard VLA calibrator. However, this may not always be possible, especially at the low frequencies. For example, for $J1829+487$ there is no nearby (not within $20^\circ$) compact P-band calibrator. In such cases, bootstrapping is the only way and consistency checks have to come from the analysis itself! As mentioned in the last section, one way to ascertain that bootstrapping has worked or not is to check the number of closure relations that are satisfied within the noise level. The image plane models so obtained can be graded as I-.

Another simpler way to check whether the bootstrapping worked or not is to look at the gains of the antennas common in two successive iterations of the bootstrapping process. E.g., lets say that the gains for say $M$ antennas have been determined correctly in an iteration. Any significant change in the gains of these antennas in the next iteration would mean that at least one of the following two things is happening: (1) one of the new antennas included in the current iteration is really bad - i.e. gives high closure errors and hence corrupts the earlier solutions, or (2) the process is pushing the source structure based information into the antenna based gains and that is a serious problem. For $J1829+487$, we checked for such a consistency at each iteration. Satisfied with this, the image plane model for $J1829+487$ is marked as I- and can be used as an input model for solving for antenna gains in the future observations.