[evlatests] WIDAR bronze dataset - C band dual pol.

[Michael Rupen]

> I agree with Frazer that the TSYS should give good estimates of the
> weights.  However, FIXWT does a good job in determining what the
> post-fit (remove the image, look at the rms in the data, baseline by
> baseline or antenna-based) weighting really should be (caveat: my last
> use of FIXWT showed some problems in the technique, mainly too short of
> an averaging time).  In my experience, fiddling with the weighting
> properly does decrease the SNR on the image at the 10 percent level
> (assuming no bad antennas with too high weights are included), but with
> little change in the remaining artifacts.  So my guess is that you don't
> have a major weighting problem, but something more fundamentally strange
> with the data.
>
> Ed


We're doing quite odd things with WIDAR-0 compared to normal
observations -- attenuators fixed, no scaling by total power, no Tsys
scaling, etc. etc.  The weights used in imaging, unless you take special
care, will reflect the scalings required to account for all this, and may
wind up very odd indeed.  So a priori it's not obvious that one wouldn't
gain quite a bit from putting in proper weights...in fact I'm a bit
surprised Vivek's tests don't show a bigger effect.

As I mentioned to Frazer, one might do better by deriving rms from the
scatter over channels than over time, for WIDAR data.  Assuming no lines,
ignoring RFI, etc. :}

              Michael