Changes/gpcal

gpcal
Item	Complexity	Research	Milestone	Time
XYPhase	easy	some	6	2d
Wide band calibration	hard	moderate	10	20d

Things to Change

• potential XYPhase issue
• wide band calibration option - add option to split wide frequency band in a number of sub-bands for gain and leakage calibration and derive spline fits across the frequency range instead of scalars.

Research:

• check whether gpcal reads xyphase from input uv data. At present this is in a variable?
  • No, xyphase from RPFITS syscal records can be applied in atlod and is stored in a uv variable (xyphase), but is not read by the calibration tasks. This eliminates this item from the table above.
• Study how miriad stores calibration tables and how we can extend these to store additional variation in frequency for complex antenna gains and leakages.
  • see MiriadCalibrationTables for some reverse engineered formats and possible extensions.
• It's unclear to what extent the wide band at high freq (eg 20GHz) is an analogue of the 20cm situation, vis-a-vis calibration (particularly phase winding across band).
  • Phase winding is directly proportional to the relative bandwidth, so across a 2 GHz band at 20 GHz the phase winding due to atmospheric delay would be 10% of the phase variation. For conditions where phase winding causes decorrelation of the vector averaged visibility over the band, we are likely to need selfcalibration on very short timescales. Correcting just the average phase would leave time variable phase slopes across the band in the data. Thus a delay correction should be applied instead. Mfcal can already solve for a time variable delay term on a calibrator, there is no task to do this on a general field yet.

Changes Home
Wiki Home