This is the message sent on the VLBI exploder on Aug. 10, 2005. VLBA Correlator EOP Errors A significant issue has been found with the Earth Orientation Parameters (EOP) used in VLBA correlator job scripts since May 2003. The problem adversely affects projects that depend on an accurate correlator model. Such projects include any that use phase referencing or that try to solve for the atmosphere using sources scattered around the sky. Projects that depend primarily on self calibration on the target source or that use the total delays from the correlator are not affected. A bug was found in the correlator job generator that caused it to use predicted values of EOP rather than measured rapid service values. The magnitude of the errors, mostly in UT1-UTC, is roughly equivalent to having source position errors of typically about 10 mas before June 2004 and about 20 mas since then. The worst cases were about 4 times these values. The effect on phase referenced observations depends on details of the offsets and geometry. The phase error on a source depends on position on the sky and on time. To first order, corrections based on calibrator observations will fix the phases on a target. But, as with any other geometry error, the residual error on the target after calibration will be approximately the calibrator phase error reduced by the calibrator to target separation in radians (typically well under 0.1). This phase residual can shift measured positions. But it not completely equivalent to a position shift so it can also degrade phase reference image dynamic range and, in extreme cases, can prevent detection of the target. A more detailed description of the problem will be published eventually in a VLBA Test Memo which will be available at http://www.vlba.nrao.edu/memos/test/. The job generator was fixed on August 8, 2005. There is currently no AIPS program that can correct the affected data sets. But soon the capability will be added as an option in CLCOR. Note that even without this bug, VLBA correlation is based on rapid service EOP values from close to the date of observation, not on the finals that are only available much later. Users wanting high accuracy may wish to make corrections for any observations. Thank you to Mark Reid for calling to our attention that we were using EOP values with significant errors. For further information, contact me or any of the other Socorro VLBA staff. Craig Walker --------------------------------------------------------------------- R. Craig Walker Array Operations Center cwalker@nrao.edu National Radio Astronomy Observatory Phone 505 835 7247 P. O. Box O Fax 505 835 7027 Socorro NM 87801 USA ---------------------------------------------------------------------