[evla-sw-discuss] Atmosphere

[Bryan Butler]

On 3/28/06 18:10, Barry Clark wrote:
> I've been thinking about what to do with the atmosphere.  I can't really
> make up my mind.
> 
> Currently we just let CALC use a model climate.  This, empirically seems
> to work best for the VLBA, indicating that the upper atmosphere seems 
> to follow its seasonal average better than it follows what is happening 
> on the ground.  But for a relatively small array, most of the atmospheric
> phase path differences do lie in the lower atmosphere, and the relative
> importance of the upper atmosphere is much less.  
> 
> Pointing is handled by a separate mechanism, with the VLA climate model 
> (clearly inferior to CALC's) being used to predict refraction.

i have recommended a number of times that we should adopt the model for 
refraction described in jeff mangum's ALMA pointing memo.  i still hold that 
this is true.  is there some reason not to do this (i've asked before)?

> It is clear that the VLA does much better than the EVLA antennas in 
> pointing, and there is some suggestion that it does better with phases
> as well.
> 
> I think it is clear that we want to get the meteorological variables 
> into the pointing calculation.  This is pretty straightforward, although
> I can't make up my mind about which side of the java/C interface to do
> it on (the current calculation lives on the C side, but it might be easier
> to keep an eye on on the java side).

seems to me we ingest the variables on the java side.  for pointing this is the 
obvious thing to do, right, since the refractive pointing is handled within the 
java classes?

> I am much less clear on what we should do about the phase path calculation.
> 
> Possibilities:
> 
> 1.)  Do nothing - follow the current path until driven off of it with more
> compelling data.
> 
> 2.)  Feed the meteorology into CALC.  I'm not very clear what it might do 
> with it, but it would use it somehow or other.  Since we have only one 
> barometer, I think we have to fake barometer readings for each antenna,
> according to their elevations.  A bit messy.

does the VLA currently estimate differences based on elevation?

> 3.)  Calculate zenith wet and dry phase paths as currently done at the 
> VLA (again with corrections for the antenna elevations), and use info 
> from CALC to apply them to the current rays.  (I think I know how to do
> this, but it might take a bit of learning.)
> 
> 4.)  Turn off the CALC atmosphere calculation and install the VLA 
> atmosphere instead.

this is actually what i think is best, because i think in the end we want this 
anyway (the ability to put in our own atmosphere).

i have coded up the various liebe models in the past, which calculate phase path 
explicitly given an input atmospheric model, but they are in FORTRAN.  i think i 
could coax them into C fairly easily, but putting them into java would take alot 
more work to figure out the objects and methods...

juan pardo's ATM model has been incorporated into CASA (and the ALMA system in 
general) - we should investigate how good it is at the lower frequencies, as we 
essentially get it for free based on the ALMA work.

so, for now, we use the VLA atmosphere.  when we turn off the Modcomps (whenever 
that is - it seems to be advancing into the future), we can plug in whatever 
atmosphere we choose, if we have one available (or keep using the VLA one if we 
don't have an alternative).

> All of these are complicated to some degree or other if we use this 
> software to drive PieTown.
> 
> Comments?
> _______________________________________________
> evla-sw-discuss mailing list
> evla-sw-discuss at listmgr.cv.nrao.edu
> http://listmgr.cv.nrao.edu/mailman/listinfo/evla-sw-discuss