[alma-config] simulation image library

[John Conway]

Hi Min et al,


I know you have a Cygnus A image in your .gif library already 
but the FITS version in my http area, if one settles on Cygnus A  as a
test image  might be a better one to use. Actually having  a two lobed
source isn't really important, however saving a factor of two in 
overall size can be important for running the simulations fast
(especially if one is ruunning MEM which require images which are 
twice as big).

Of course the Cyngus A image may not be voted one of the 5 to use
anyway, and it can be argued that ALMA won't be observing Cygnus A.
My philosphical position on this it is that it is more important 
to choose as wide a range of image structures as possible  (to see
first of all if there is any difference in imaging properties 
which dends on the combination of array and image structure and what
these difference are),  rather than guessing the 
exact sources ALMA would observe. In this regard the Cyngus A image 
has a compact component embedded in diffuse emisison, the M51CO is low 
dynmaic range but lots of filled beam areas, 10KDOTS is lots of
points etc etc.

This question of sampling different image structures is one reason
artifical fractal images have some appeal. I doubt that we can guess
exactly what ALMA will observe;  who in 1970 would guess the 
images that the VLA makes today?  Still 
I understand the alternative point of view; we also want to sell 
ALMA to a wide community and its much easier to sell to present mm and
future mm astronomers if we can point to an image which is familiar and of 
the type that ALMA is being sold on. As Al mentioned in Tuscon taking
optical images of galaxies is good from this point of view (.. our 
sims may also be used to sell the project..), so maybe we should 
conclude that optical image of m51 is probably a better
test image than the CO image.

I would ultimately vote for a mixture of the most complex images of
objects which ALMA will likely observe in its early years plus some
others which probe different areas of parameter space.  

I would finally argue that we can always add image to the library 
of test images later. Maybe even the final evaluation can be done 
on a bigger and more realistic set, the main thing is to get going 
on doing simulations with some common images, which then provide 
a common way of inercomparing array designs.

 John.

P.S I take your view about measuring the dynamic range of different
images. Of course sometimes dynamic range is taken for a synonym for
complextity but they are different properties. One could classify images in 
terms of both Dynamic range (ratio of max to weakest structure) and 
the complexity (number of filled beam areas). One can have images which
are high dynamic range but are not complex  (a point plus a weak halo for 
instance).



On Fri, 7 Apr 2000, Min Yun wrote:

> John,
> 
> I have looked at your images and included the 10KDOT and M51CO into
> the image library.  The 10KDOT image is a more reasonable test image
> for SKA, but it is probably less relevant for ALMA since we do not
> expect such high source density within the single primary beam.
> I will leave it in anyhow.  I worry about the lack of dynamic
> range in M51 CO image, but it prompted me to measure DR for all
> of the astronomical images for a more quantitative feel.
> The single dish images have a low DR as expected.
>