Re: [alma-sw-ssr] Re: ALMA-SSR review

[Mark Holdaway <mholdawa@nrao.edu>]

On Thu, 4 May 2000, Richard Crutcher wrote:

> Comment on polarization calibration (page 10 of report: "Polarization
> calibration This can probably be derived from phase calibration
> observations."
> 
> The science goal will be to obtain maps in all 4 Stokes parameters
> routinely, at the 0.1% level, without special time-consuming
> polarization calibrations. (See report of the ALMA Science Advisory
> Committee, especially the appendix on polarization observations.)
> Polarization mapping introduces many problems for calibration. The
> telescope primary polarized beam is not a simple (near) gaussian, but
> has a much more complex structure in Q, U, and V than in I. Moreover,
> this structure is time variable with physical changes such as sag of the
> feed and reflector with elevation, focusing, etc. 

It seems we can investigate this now and say at what frequency it becomes
a problem.  Presumably the specs on the ALMA antennas are stiff enough
that it is not a big problem, say, below 200 GHz or some similar number.  
What do you think the distribution of polarization observations with
frequency looks like?  Fixing evelation-dependent beam effects would be a
nasty problem we would like to avoid if possible. For high SNR,
direction-dependent self-cal algorithms are possible, but polarization is
intrinsically low SNR; maybe it won't be a problem (ie, the SNR is low
enough so that the effects of the leakage beam changing with elevation is
masked by thermal noise in most cases) -- again, we can answer that now if
it is important.

> The polarized beam
> pattern will rotate with telescope tracking, 

Aips++ handles the beam rotation with paralactic angle quite nicely.

> and for feeds not on the
> optical axis even the polarized response at beam center will change.

Is this true?  I have seen time-dependent behavior of the beam-center
polarization responce at the VLA, but it was not simply correlated with
parallactic angle or anything.

> Moreover, single-dish data for zero and short spacings will be required.
> Single dish polarimetry is often done by subtracting total power
> measurements made with changes in the polarimeter (right - left, or
> linear at different position angles). This technique severely limits
> polarization accuracy. For ALMA, investigation of cross-correlation of
> the 2 receivers on each single dish to derive the 4 Stokes parameters is
> essential. 

They are doing this now at the 12M.  For a few months anyway.

> Mosaic mapping will require accurate knowledge of the time
> variable polarized beam pattern down to very low levels. For example,
> the proposed single-dish technique for ALMA is on the fly mapping, which
> will emphasize the effects of strong sources in the polarized primary
> beam sidelobes. 

To the extent that the polarization beam is known, confusing sources
leaking into the polarization total power can be easily removed.
But we need to specify this so that people write the software and
measure the polarization beam.

> Thus, the software requirements for polarization will be
> much more severe than can be satisfied by deriving polarization
> calibration from standard phase calibration observations.

Operationally (ie, not including the upfront work some people will have to
do to write the software and measure the beam) I think the polarization
projects will not present much overhead beyond non-polarization projects.

One guys opinion.

	-mark