[alma-config] BOUNCE alma-config at majordomo.cv.nrao.edu: Non-member submission from ["David Woody" <dwoody at caltech.edu>]

[Mark Holdaway]

> 
> Fortunately, nature is on our side in this situation.  Earth rotation
> naturally
> gives you ~1/BaseLine integration time for free.  So uniform coverage in
> snap shot gives you 1/UV tapering in integration time if you can wait for
> the longest baseline to change by one dish diameter.  Thus the UV coverage
> calculations and image simulations
> should cover an appropriate amount of earth rotation time to get the
> weighting ALMA data will have.
> 

I think this is a spurious argument.  While it is true that the
(u,v) points at the edge of the coverage traverse the cells more
quickly, in a ring-type array with uniform coverage, the density
of points at the edge is similar to the density of points *near*
the center (there is an overabundance of samples at the shortest
baselines and an absence at the center - we ignore these).  The
points just move into each other more quickly at the edge (or
another way to say it: the weight of one baseline in a cell
may be less, but you are more likely to have multiple baselines
contribute to that cell).

If you made an image of the weights (proportional to integration
time) and convolved with a Gaussian of a few pixels, you get a very
smooth function, not one which falls off at 1/D.  In fact, for
a circular array, there is a slight excess of the longest baselines,
which can be clearly seen in such an image.

There is a similar argument though, but at a lower magnitude.
Earth rotation will forshorten the baselines, predominantly in the E-W
dimension, so there will be some natural tapering due to earth
rotation.  My simulations 23 years ago indicated that this is not
a significant help at reducing the Lambda function sidelobes.

	-Mark