[daip] Re: automatic data flagging

[Eric Greisen]

We do not do web pages describing things.  The task itself is called
FLAGR and has been released and so is current at sites that run the
31DEC04 MNJ.  A task FINDR is useful as support to determine "normal"
values.  The band-edge issue is interesting - in principle with the
VLA (except at 74 MHz) the S/N is set by the front end with a bandpass
very much larger than that observed.  As a consequence all channels
should be equally good except perhaps the outer 1-2 due to aliasing
affects.  This is not seen to be the case however and BPASS shows an
increase in closure error at the edges that mimics the observed noise
levels.  It is not clear why this happens since the slope of the
bandpass does not seem to matter as it should in closure problems
arising from a change across the channel.

Noise should not increase with uv distance other than in imaging where
there are fewer long spacings than short ones.  To delete those is to
throw away the resolution the user wants.

Eric Greisen