[evlatests] Deep Imaging of 3C345

[George Moellenbrock]

Rick-

I think your reply deserves wider distribution.  Sorry if this is 
nit-picking...

On Fri, 12 Feb 2010, Rick Perley wrote:

> In this case, BLCAL was run and applied.  I can easily image without applying 
> it (but am now working on an X-band dataset -- higher priority).

No problem.  Hmm, I actually was expecting that it was BLCAL-less and the 
magnitude of the dynamic range quote was due mostly to measuring rms well 
away from the mess in the center.  Note that 3C345 is polarized, so the 
variable polarization-induced closure errors won't be soaked up very well 
by a constant BLCAL.  Maybe a BLCAL so-biased by source polarization 
contributes in some complex way to the zoo of artifacts you noted.

> The 'dynamic range' in the vicinity of the strong source is probably down 
> from what I quoted by a factor of a few.  I'm not sure any such quote is 
> useful -- the rms one gets is hugely dependent on where you put your box.

For a complicated source like 3C345, I agree.  On/near-source errors will 
be quite variable.  But these really are the meaningul errors, especially 
for observers looking at interesting things (not "dots").

> Probably more important is the peak artifact height -- this is about 1 mJy, 
> which is 1/10000 of the peak in the map.

That's a 35-sigma (!) artifact if your original report of dynamic 
range is meaningful and to be taken seriously....  (Nit-picking?!)

> There is a larger artifact than 
> this in fact -- but it's two cells from the core, and probably represents 
> errors associated with the real structure of 3C345 (a jet, very well known). 
> You can't see it in the image of course, as it's buried by the clean beam.

Yes, complex sources are tricky.  Interpretting the info I suggest/request 
is probably easiest for the purest of dynamic range limit tests (truly 
'dot' sources), but _whenever_ you're moved to quote dynamic range, this 
additional info (and noting dozens-sigma artifacts!) is indispensible, 
regardless of (or perhaps because of?) the effects of source complexity.

In general, I wonder if reporting some combination of realized/expected 
rms (with some indication that it varies significantly for complex fields) 
and the level of systematic artifacts above the realized rms (and as a 
fraction of peak) is more useful than the rather blunt "dynamic range" 
quote, which I find pretty misleading.  It is just not that useful without 
lots of qualification. Clearly, that you can find a spot that implies 
350k:1 is the least interesting thing about the image.  (At best, is it 
merely an indirect indication of the highest _possible_ dynamic range you 
might achieve if you can sort out the systematics correctly?  And a fairly 
poor indication since it is probably an upper limit...)

In any case, just trying to tease out a clearer picture of what you've got
there...

-George


--