[daip] Re: [HelpDesk #7037] UVGLU (fwd)

[Thomas Greve]

Hi Eric, 

 ok, I'm going to stick to the UVCOP/DBCON option.

Many thanks for help!

Cheers
Thomas


On Tue, 4 Feb 2003, Eric Greisen wrote:

> 
> UVGLU is for the case where a correlator can only do some of the
> channels at a time.  It is for two "perfectly matched" data sets where
> the first contains channels 1-N of a full data set and the second
> contains channels N+1-M of that data set.
> 
> The situation you have is very much more complicated.  The UVCOP/DBCON
> route is what you will have to do here.  There are other choices which
> true spectral line observers sometimes do where the 2 IFs in one data
> set overlap in frequency.  There is a program to handle that.
> 
> I assume that what you are actually doing is not true spectral line -
> your individual channels are awfully wide.  So I do not know why you
> want the channels individually anyway.  If you were doing continuum
> imaging using the channels to avoid bandwidth smearing then normally
> you want a full data set fully sampled - which you do not have.
> 
> So long as you want the image as a cube of 9 channels then you can
> IMAGR them separately and put them back together with MCUBE before
> display.  Note that the outer channels (in this case usually channel 1
> or 7) are noisier than the others for reasons not well understood.
> 
> Eric Greisen
>