Getting Started

Documentation

Glish

Learn More

Programming

Contact Us

Version 1.9 Build 1367

News	FAQ
Search	Home

An approach to cross-calibration

Our preferred approach to calibration uses the reference table or MS concept to keep both the original data together in one MS and to provide calibrated views of that data via reference tables.

• The MeasurementSet M is subject to some selection, S_cal, to produce a logical subset MS M(S_cal), the calibrator MS. The selection used is of interest in two ways: first to ensure that the correction is applied to the relevant data (e.g. 6cm to 6cm observations), and second as information to the user (e.g. ``wind-velocity $\leq$ 3 m/s'').
• For each such calibrator MS, a SkyModel is used to predict the corresponding coherences.
• For each such calibrator MS, one will want to perform a solution for one or more MeasurementComponents (e.g. GJones and DJones).
• Each MeasurementComponent is then stored as a Table in the calibrator MS.
• The MeasurementSet M is subject to some selection, S_source, to produce a logical subset MS M(S_source), the source MS.
• MeasurementComponents are then explicitly copied from calibrator to source MS, either by the user or by some simple program.
• After copying, a MeasurementComponent can be subject to a number of operations:
  • Re-interpolation in some coordinate such as time or some derived quantity such as antenna elevation or airmass,
  • Editing by hand or by algorithm,
  • Boot-strapping of some form (e.g. determination of the flux of some sources from a primary calibrater),
  • Recasting to another form (e.g. antenna-based phases to position-based phase screen).
  All of these are probably best thought of as construction of a new MeasurementComponent that replaces the initial one.

The net result is that one main MS and a number of reference MSes pointing to it and containing the relevant MeasurementComponents. This scheme has a number of advantages:

1.

(Relative) simplicity. We consider this to be very important,

2.

The MeasurementComponents in the source MSes exist independently of the calibrator MSes and can be edited or munged as one chooses,

3.

Self-calibration is easily accomodated by simply updating the MeasurementComponents in the source MSes.

An alternate but more complicated approach is to defer the connection between sources and calibrators until the correction of data is actually performed. In this scheme, each reference MS is assigned a unique integer, CG, and an MS column ``CALIBRATION_GROUP'' is set to that integer. The links between calibrator and source are then made via this index. The data for which these corrections are to be applied are selected via a selection S<sub>source</sub> to form a target MS, M(S<sub>source</sub>). The column CALIBRATION_GROUP of this MS, is set to CG. We discarded this type of deferred selection of MeasurementComponents as being overly complex, and not allowing the user to munge the MeasurementComponents ``by hand''.