Preventing Future Problems

This whole episode raises the question of how we can prevent anything similar from happening in the future. It is easy to keep tabs on the EOP in the future, but surely the next problem will be something else so a scheme is needed to detect anything that degrades the VLBA's astrometric capabilities. The difficulty is that anything like the EOP problem is subtle and easily lost in other large, but normal, error sources such as the atmosphere. The raw correlator output from any test involving new observations cannot be directly compared with a standard run because of variations in clocks, atmosphere, station positions, EOP etc. A standard observation could be recorrelated periodically to be sure the correlator had not changed, but that would miss something like the EOP error because it would not be sensitive to failure to keep up with changing parameters.

A possible test, perhaps to be added to the weekly ``MT'' test observations, would be to do something like an hour of phase referencing between 2 or 3 very bright calibrators with well known position. Such calibrators should be much brighter than those used in the obserations reported here. They would need to be processed in AIPS, probably with the pipeline script, and relative positions and image RMS's should be measured. If there is any significant change, that would be cause for a staff scientist to investigate.

Another thing to do is just to keep an eye on the raw phases coming off the correlator. This is easiest for observations that are a long track on a bright source. If those sources seem to be subject to excessive phase winding, flags should be raised. At frequencies such as 4 and 2 cm, the number of turns of phase across a day should only be a few -- we supposedly know all of the geometry except the atmosphere and clocks to better than a wavelength.