Presentation #353.09 in the session “Computational Augmentation to Observations”.
Before a radio image can be reconstructed from interferometric data, it must undergo calibration. As part of this process, calibration of the data is done using a known radio source, chosen by the observer. They must balance the angular distance to the source, the strength of the source, and the shape of the source. If use an analytical system to assess potential calibrators, we can remove this source of human error. As well, we can move toward automating the calibrator selection and assessment processes. Rapid assessment will become more important as we move toward the next generation of radio telescopes, whose increased resolutions and sensitivities will render many current calibrators no longer useful.
To do so, we introduce a scoring system for the quality of a calibrator based on its divergence from an ideal point source. This divergence is calculated as the distance in a coordinate space of made of average calculated errors in the visibilities of a source, giving a value for how point-like the source is. To further refine this into a calibrator score, this is combined with a system of weights and cut-offs to enhance the differentiation between known good calibrators and poor calibrators. The system tested analyzes potential calibrators in the VLA S-band, many of which are known calibrators in the VLA L-band and C-band. Using the known calibrator classifications to predict the likely classification in S-band, we demonstrate that this scoring system can, with refinement, effectively classify calibrators through direct analysis of observations of a source.
This research was conducted as part of the summer internship program at the National Radio Astronomy Observatory. The National Radio Astronomy Observatory is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc.