Subject: jemx mosaics

Hi Carol-anne,

I've been looking at jem-x mosaics with volker - the idea being to 
investigate if we can use j_ima_src_locator in the barn and it seems 
pretty easy to use.  However, I was trying to find a level at which to 
set the min significance to get the best results and not too many low 
sigma sources - but this has proved a bit difficult.

First off I made a mosaic of a GRS 1915 pointing ~100ksec.  I find GRS 
at ~ 1000sigma (although in the significance image the peak is only 600, 
but thats another question...) and a few other sources reported at about 
40 sigma (20 in the image) which look spurious.

Then I made a Galactic centre image, this time only 12ksec, which 
contained a few well known GC sources and again a few NEW sources - but 
this time only at ~15 sigma.

When I investigated the properties of the mosaic, both images had a nice 
Gaussian looking distribution of significances, with a positive tail 
(the sources), but crucially the standard deviation of both were not = 
1sigma, which they should be at least close to.  Moreover, the stds of 
both were different - the GC image has std = 1.6sigma, whereas GRS has 
5.5sigma.  This means that a limit of 15sigma in the GC image is really 
about 9sigma (sounds sensible), and in the GRS image 40sigma is really 
about  7sigma (also sounds sensible).

Obviously in the barn we need to just apply one limit for 
jmx_src_locator, which we can't do if different observations have 
different properties - but also it seems that the results from the 
mosaic analysis are not correct; one wouldn't necessarily expect std of 
the image distribution to be exactly 1, I would expect the sources in 
the FoV to have an effect, but not such that two obs differ by a factor 
of 3.  Is there some sort of effect due to length of exposure that isn't 
taken into account ?

Any help would be appreciated, and I can provide further details if 
necessary.

Cheers

Simon