[mmaimcal] Re: [alma-sw-ssr] question about frequency/band switching speeds and rates

[Frazer Owen]

    Ed,

    I must jump in here because I have been concerned about understanding
the correct interpretation of your results. Let me try out another 
explanation
on you and see if you can shoot me down.

    I don't think the model of small cloudlets makes sense physically,
under most conditions. The VLBA phase gradients must be due to
something else. Under reasonably good conditions the power on the
atmospheric fluctuations should be mostly on very large scales. I
think we have measured this function using the tipper and it agrees
in general with what one expects theoretically.  The observed
phase fluctuations are due to the pattern being blown over the array. Fast
switching should be fast enough so that one is still seeing the same
screen. Thus the distance to the source should not matter a lot, as long
as the antennas are fast enough.

    The best explanation of your experience that I can think of, is
the zenith delay is not accurately enough predicted by the ground
measurements. Thus since one needs to correct for the difference
in delay through the atmosphere at different elevations then the
multiple calibrator approach you use will correct for the error
in the zenith model to first order. Also that would explain why
calibrators further from the source don't work as well.

    Of course, closer calibrators or multiple calibrators also
cancel out other errors to first order, like time and baseline.

    If my suggestion is right then there is a premium on knowing
the zenith delay. That can be measured astronomically and maybe
this should be a priority for ALMA. Also the ALMA site is pretty
good and a pretty constant zenith model might work well, unlike
the diverse VLBA sites.

    Mark,

    Have you though about this issue ?

---Frazer

   

Ed Fomalont wrote:

> Hi Mark,
>
>    I think you have thought of most everying - in only 10 years.
> What model do you use for the phase correlation as a function of
> calibrator-target separation?  To first order the rms phase fluctuations
> are a linear function of target-calibrator separation, but when the
> residual phase fluctuations become larger than about 50 deg, significant
> decorrelation occurs.  Also, we find with the VLBA that there are
> systematic phase gradients in the sky which persist over many minutes.
> These can be determined and removed by observing more than one
> calibrator source.  However, the VLBA baselines are much longer than 
> those
> for ALMA, so that the phase fluctuations should be dominated by the
> short-term, small-angle phase cloudlets over the array.



>
>    Also fitting the antenna-based phase fluctuations for the 
> calibrator over
> the whole array may be useful, especially for weaker calibrators when 
> fitting
> a smaller number of parameters to fit the atmosphere delay wedge (and 
> maybe
> higher order terms) rather than one value for each antenna, would
> give better SNR.  You might be able to see relatively big blobs of 
> refraction
> moving over the array, and hence apply their effect better to the target
> source.
>    If I seem to have a fixation of very weak calibrators, it is
> because they do give better phase referencing results.  Also, you 
> might have
> to position nod more frequently than 25 seconds so that the interpolated
> phase between calibrator observations really does apply to the target.
> Of course, finding weak calibrators will be time consuming.
>
>    Ed
> _______________________________________________
> mmaimcal mailing list
> mmaimcal at listmgr.cv.nrao.edu
> http://listmgr.cv.nrao.edu/mailman/listinfo/mmaimcal