[daip] New Staff Reply - [!IDT-632492]: Data weights in Aips

[Eric Greisen]

New Staff Reply: Data weights in Aips

Data weights are (except for EVLA data) presumed to be in the
form of 1/sigma^2 where sigma is the uncertainty in the same
units as the visibilities.  Thus, when amplitude calibration
is applied, the sigmas are scaled by that gain amplitude.  The
weights are then multiplied by 1/Gain^2.  UVSUB when division
occurs also needs to divide sigma by the model.  This raises
the sigmas a lot when the model has a low value in the outer
portions of the UV plane.  This can have a large effect on the
gain solutions which is why we offer WEIGHTIT to alter the
weights to be less drastic in CALIB and similar tasks.

For EVLA the weights are only an integration time unless REWAY or
TYAPL are applied.  Then they become weights like those discussed
above.

Eric Greisen


Ticket Details
===================
Ticket ID: IDT-632492
Department: AIPS Data Processing
Priority: Default
Status: Closed
Link:  https://help.nrao.edu/staff/index.php?_m=tickets&_a=viewticket&ticketid=513